Alibaba Qwen: Ķīnas AI modelis, kas izaicina Silicio ieleju

Autori : Ģirts Bai Shuai Bai Jūlijs Ču Zīle Cui Kad briesmas Mārtiņš Dengs Mīlestība Venēcija ir Jānis Hāns Fei Huang Binyuan Hui Luo Džī Maija Li Jūnijs Lin Runji Lin Dayiheng Liu Gao Liu Chengqiang Lu Atpakaļ Lu Jēkabpils ma Rui vīrieši Xingzhang Ren Jēkabpils Ren Mārtiņš Tan Mārtiņš Tan Jianhong Tu Vārds Wang Lielais Wang Wei Wang Shengguang Wu Benfeng Xu Jin Xu un Yang Hao Yang Jian Yang Shusheng Yang Yang Yao Bowen Yu Hongyi Yuan Jūlijs Jūlijs Jianwei Čang Xingxuan Zhang Yichang Zhang Zhenru Zhang Čang Čū Jingren Zhou Xiaohuan Zhou Tianhang Zhu Qwen komanda, Alibaba grupa Autori : Ģirts Bai Shuai Bai Jūlijs Ču Zīle Cui Kad briesmas Mārtiņš Dengs Mīlestība Venēcija ir Jānis Hāns Fei Huang Būvmateriāli Hui Luo Džī Maija Li Jūnijs Lin Rudzītis Lin Lielais Liu Gao Liu Chengqiang Lu izstāde Atpakaļ Lu Jēkabpils ma Rui vīrieši Xingzhang Ren Jēkabpils Ren Mārtiņš Tan Mārtiņš Tan Jianhong Tu Vārds Wang Lielais Wang Vīrietis Wang Shengguang Wu Benfeng Xu Džins Xu un Yang Hao Yang Jans Jansons Šūpuļkrēsls Jēkabs Lāčplēsis Yu Honkongas juaņa Jūlijs Jūlijs Jianwei Čang Xingxuan Zhang Jevgeņijs Žans Žanrs Žanrs Čang Čū Jingrēns Čū Xiaohuan Džū Tianhang Zhu Qwen Team, Alibaba Group Abstraktā Lieli valodu modeļi (LLM) ir revolucionizējuši mākslīgā intelekta jomu, ļaujot dabiskās valodas apstrādes uzdevumus, kas agrāk tika uzskatīti par ekskluzīviem cilvēkiem. QWEN ir visaptveroša valodas modeļu sērija, kas ietver atšķirīgus modeļus ar dažādiem parametru skaita rādītājiem. Tā ietver QWEN, pamata sagatavotos valodas modeļus un QWEN-CHAT, tērzēšanas modeļus, kas pielāgoti cilvēka pielāgošanas tehnikām. Pamata valodas modeļi konsekventi demonstrē augstāku veiktspēju daudzos turpmākajos uzdevumos, un tērzēšanas modeļi, jo īpaši tie, kas apmācīti, izmantojot Reinforcement Learning from Human Feedback (RLHF), ir ļoti konkurētspējīgi. Tērzēšanas modeļiem ir uzlabotas rīku izmantošanas un plānošanas iespējas, lai izveidotu aģentu lietojumprogrammas, parādot iespaidīgu veiktspēju pat salīdzinājumā ar lielākiem mode 1 1 Ievads Lielo valodu modeļi (LLMs) ir revolucionizējuši mākslīgā intelekta (AI) jomu, nodrošinot spēcīgu pamatu sarežģītiem apsvērumiem un problēmu risināšanas uzdevumiem. Šiem modeļiem ir spēja saspiest plašas zināšanas neironu tīklos, padarot tos neticami daudzpusīgus aģentus. Ar tērzēšanas saskarni LLM var veikt uzdevumus, kas agrāk tika uzskatīti par cilvēku ekskluzīvo domēnu, jo īpaši tos, kas ietver radošumu un pieredzi Viņi var iesaistīties dabiskās valodas sarunās ar cilvēkiem, atbildot uz jautājumiem, sniedzot informāciju un pat radot radošu saturu, piemēram, stāstus, dzejas un mūziku. (Radford et al. un citi) 2018. gadā; Lāčplēsis et al. 2018. gadā; Raivis un al. 2020 gadā; Brūns un al , 2020 gadā; atvērts , 2023. gadā Chowdhery un tā tālāk. 2022. gadā Anita un al. 2023. gadā Lāčplēsis et al. 2022. gadā Touvron un al , 2023a un b) Atvērto durvju, 2022. gadā Lāčplēsis et al. 2022. gadā Anilīns un al , 2023. gadā Google tīkls, 2023. gadā antropozīcija , 2023a un b) LLM neaprobežojas tikai ar valodas uzdevumiem, bet var darboties arī kā vispārējs aģents. sadarbojas ar ārējām sistēmām, rīkiem un modeļiem, lai sasniegtu cilvēka izvirzītos mērķus. Izpildi kodu Izmantojiet rīkus Tas paver jaunas iespējas AI lietojumprogrammām, sākot no autonomajiem transportlīdzekļiem un robotikas līdz veselības aprūpei un finansēm.Kā šie modeļi turpina attīstīties un uzlabot, mēs varam sagaidīt vēl inovatīvākus un aizraujošākus lietojumus nākamajos gados.Neatkarīgi no tā, vai tas palīdz mums atrisināt sarežģītas problēmas, radīt jaunus izklaides veidus vai pārveidot veidu, kā mēs dzīvojam un strādājam, LLM ir gatavi spēlēt centrālu lomu AI nākotnes veidošanā. (Rūdolfs un citi) 2022. gadā Bai un al. 2022. gadā Kārlis un al. 2023 a autostāvvieta , 2023. gadā Mārtiņš un al. 2023 gadā) Atvērto durvju, 2023. gadā Bai un al. 2023. gadā Mārtiņš un al. 2023 a Jūs un Al. 2023. gadā Dāvids un al. 2023. gadā Pēteris un al. 2023 b) un (Tēvzemei un Brīvībai) 2021. gadā Jēkabs un al. 2023. gadā Li un al , 2023. gadā (Pēteris un al., 2023. gadā Lāčplēša iela, Inc. 2023. gadā autostāvvieta , 2023 gadā) Neskatoties uz to iespaidīgajām iespējām, LLM bieži tiek kritizēti par to reproducējamības, vadāmības un pieejamības trūkumu pakalpojumu sniedzējiem. Šajā darbā mēs esam priecīgi iepazīstināt un izlaist mūsu LLM sērijas sākotnējo versiju, QWEN. QWEN ir nosaukums, kas izriet no ķīniešu frāzes Qianwen, kas tulkojumā nozīmē "tūkstošiem ielūgumu" un pārraida jēdzienu, kas ietver plašu vaicājumu klāstu. QWEN ir visaptveroša valodas modeļa sērija, kas ietver atšķirīgus modeļus ar dažādiem parametru skaitļiem. Modelu sērija ietver pamata priekšapmācītus valodas modeļus, tērzēšanas modeļus, kas aprakstīti ar cilvē Pamata valodas modeļi, proti, QWEN, ir pakļauti plašai apmācībai, izmantojot līdz pat 3 triljoniem dažādu tekstu un kodu žetonu, kas aptver plašu jomu klāstu. QWEN-CHAT modeļi ir rūpīgi pielāgoti kurētam datu kopumam, kas attiecas uz uzdevumu izpildi, tērzēšanu, rīku izmantošanu, aģentu, drošību utt. References novērtējums parāda, ka SFT modeļi var sasniegt augstāku veiktspēju. Turklāt mēs esam apmācījuši atlīdzības modeļus, lai imitētu cilvēka preferences un piemērotu tos RLHF tērzēšanas modeļos, kas var radīt cilvēka izvēles atbildes. Izmantojot izaicinoša testa cilvēka novērtējumu, mēs konstatējam, ka QWEN-CHAT modeļi, kas apmācīti ar RLHF, ir ļoti konkurētspējīgi, tomēr paliek aiz GPT-4 mūsu references. Turklāt mēs prezentējam specializētus modeļus, ko sauc par CODE-QWEN, kas ietver CODE-QWEN-7B un CODE-QWEN-14B, kā arī to tērzēšanas modeļus, CODE-QWEN-14B-CHAT un CODE-QWEN-7B-CHAT. Konkrēti, CODE-QWEN ir iepriekš apmācīts par plašiem koda datu kopumiem un vēl labāk pielāgots sarunām, kas saistītas ar koda ģenerēšanu, debugēšanu un interpretāciju. MBPP un HumanEvalPack pierādīt CODE-QWEN augsto prasmju līmeni koda izpratnē un ģenerēšanā. (Tēvzemei un Brīvībai) 2021. gadā – (Augustīns un al., 2021. gadā – (Mārtiņš Vītoliņš un citi, 2023 gadā) Mūsu rezultāti liecina, ka gan MATH-QWEN-7B-CHAT, gan MATH-QWEN-14B-CHAT pārspēj atvērtā koda modeļus tādā pašā izmērā ar lielām maržām un tuvojas GPT-3.5 matemātiski saistītajās atsauces datu kopās, piemēram, GSM8K un matemātika Kobe un al , 2021. gadā (Dinamo un tā tālāk) 2021. gadā Turklāt mums ir atvērtā koda QWEN-VL un QWEN-VL-CHAT, kuriem ir daudzpusīga spēja izprast vizuālas un valodas instrukcijas. Šie modeļi pārspēj pašreizējos atvērtā koda redzes valodas modeļus dažādos novērtēšanas kritērijos un atbalsta teksta atpazīšanu un vizuālo pamatni gan ķīniešu, gan angļu valodā. Turklāt šie modeļi ļauj sarunāties ar vairākiem attēliem un stāstīt stāstus. Bai et al. Tātad 2023 Now, we officially open-source the 14B-parameter and 7B-parameter base pretrained models QWEN and aligned chat models QWEN-CHAT Šis atbrīvošanas mērķis ir nodrošināt visaptverošāku un spēcīgāku LLM izstrādātājiem vai lietojumprogrammām draudzīgā mērogā. 2 Ziņojuma struktūra ir šāda: sadaļa apraksta mūsu pieeju priekšapmācībai un QWEN rezultātiem. aptver mūsu metodoloģiju saskaņošanai un ziņo par automātiskās izvērtēšanas un cilvēka izvērtēšanas rezultātiem.Turklāt šajā sadaļā ir aprakstītas detaļas par mūsu centieniem veidot tērzēšanas modeļus, kas spēj izmantot rīkus, kodu tulkotāju un aģentu. un Mēs iegremdējam specializētos kodēšanas un matemātikas modeļus un to veiktspēju. sniedz pārskatu par attiecīgo darbu, un sadaļa Noslēdz šo rakstu un norāda uz mūsu turpmāko darbu. 2 3 4 5 , 6 7 2 Apmācība Iepriekšējās apmācības posms ietver lielu datu apjomu apgūšanu, lai iegūtu visaptverošu izpratni par pasauli un tās dažādām sarežģītībām.Tas ietver ne tikai pamata valodas prasmes, bet arī uzlabotas prasmes, piemēram, aritmētisko, kodēšanu un loģisko pamatojumu.Šajā sadaļā mēs iepazīstinām ar datiem, modeļa dizainu un skalēšanu, kā arī visaptverošiem novērtējuma rezultātiem par atsauces datu kopām. 2.1 Dati Datu lielums ir izrādījies izšķirošs faktors, izstrādājot izturīgu lielo valodu modeli, kā to uzsvēra iepriekšējie pētījumi. Lai izveidotu efektīvu priekšapmācības datu kopu, ir būtiski nodrošināt, ka dati ir daudzveidīgi un aptver plašu veidu, domēnu un uzdevumu klāstu.Mūsu datu kopa ir izstrādāta, lai apmierinātu šīs prasības un ietver publiskos tīmekļa dokumentus, enciklopēdiju, grāmatas, kodus utt. Turklāt mūsu datu kopa ir daudzvalodu, un ievērojama daļa datu ir angļu un ķīniešu valodā. (Mārtiņš un al., 2022. gadā Rēzekne et al. 2023 b) Lai nodrošinātu mūsu sagatavošanas datu kvalitāti, mēs esam izstrādājuši visaptverošu datu sagatavošanas procedūru. Publiskajiem tīmekļa datiem mēs iegūstam tekstu no HTML un izmantojam valodas identifikācijas rīkus, lai noteiktu valodu. Lai palielinātu mūsu datu daudzveidību, mēs izmantojam deduplikācijas paņēmienus, tostarp precīzu atbilstību deduplikāciju pēc normalizācijas un neskaidru deduplikāciju, izmantojot MinHash un LSH algoritmus. Lai filtrētu zemas kvalitātes datus, mēs izmantojam noteikumu un mašīntulkošanas metožu kombināciju. Jo īpaši, mēs izmantojam vairākus modeļus, lai novērtētu saturu, tostarp valodas modeļus, teksta kvalitātes rezultātu modeļus un modeļus, lai identificētu potenciāli aizskarošu vai ir pierādīts, ka valodas priekšapmācības modeļi ar vairāku uzdevumu instrukcijām var uzlabot to nulles un dažu šāvienu veiktspēju.Lai vēl vairāk uzlabotu mūsu modeļa veiktspēju, mēs esam iekļāvuši augstas kvalitātes instrukciju datus mūsu priekšapmācības procesā.Lai aizsargātu mūsu atsauces vērtējuma integritāti, mēs esam pieņēmuši līdzīgu pieeju kā: un rūpīgi likvidēja visus instrukciju paraugus, kas uzrāda 13 gramu pārklāšanos ar jebkādiem datiem, kas pieejami mūsu novērtējumā izmantotajos testēšanas komplektos. (Jēkabpils un al., 2022. gadā Alūksnes un Alūksnes novads 2021. gadā Raivis un al. 2020. gadā – Brūns un al. 2020. gadā Ņemot vērā lielo skaitu lejupslīdes uzdevumu, nav iespējams atkārtot šo filtrēšanas procesu visiem uzdevumiem. Tā vietā mēs esam pārliecinājušies, ka norādījumu dati par ziņotajiem uzdevumiem ir pakļauti mūsu filtrēšanas procesam, lai nodrošinātu to precizitāti un uzticamību. 2.2 Tokenizācija Vārdu krājuma dizains būtiski ietekmē apmācības efektivitāti un lejupslīdes uzdevumu veiktspēju. Šajā pētījumā mēs izmantojam bajtu pāru kodēšanu (BPE) kā mūsu tokenizācijas metodi, sekojot GPT-3.5 un GPT-4. lai uzlabotu mūsu modeļa sniegumu daudzvalodu lejupslīdes uzdevumos, jo īpaši ķīniešu valodā, mēs papildinām vārdnīcu ar bieži lietotiem ķīniešu burtiem un vārdiem, kā arī tiem citās valodās. Mēs esam sadalījuši skaitļus vienā ciparā. gala vārdu krājuma izmērs ir aptuveni 152K. Jēzus, 2022. gadā Rēzekne et al. 2023a un b), QWEN tokenizera veiktspēja saspiešanas ziņā ir attēlota attēlā Šajā salīdzinājumā mēs esam novērtējuši QWEN salīdzinājumā ar vairākiem citiem žetoniem, tostarp XLM-R Lūgšana Baškīra Iekšējā Mūsu konstatējumi liecina, ka QWEN sasniedz augstāku saspiešanas efektivitāti nekā tā konkurenti lielākajā daļā valodu. Tas nozīmē, ka piegādes izmaksas var ievērojami samazināt, jo mazāks skaits QWEN žetonu var pārsūtīt vairāk informācijas nekā tā konkurenti. Turklāt mēs esam veikuši sākotnējus eksperimentus, lai nodrošinātu, ka QWEN vārdnīcu izmēra palielināšana nelabvēlīgi neietekmē iepriekš apmācītā modeļa lejupslīdes veiktspēju. Neskatoties uz vārdnīcu izmēra palielināšanos, mūsu eksperimenti ir parādījuši, ka QWEN saglabā savu veiktspējas līmeni lejupslīdes novērtējumā. 3. Kondicionieris et al., 2019. gadā) (Pāradresēts no Alūksnes un Alūksnes) 2023a) ir Inčukalna iela 2023a) ir Iekšlietu ministrijas komanda, 2023. gadā 2.3 Architecture QWEN is designed using a modified version of the Transformer architecture. Specifically, we have adopted the recent open-source approach of training large language models, LLaMA kas tiek plaši uzskatīts par augstāko atvērtā koda LLM. Mūsu izmaiņas arhitektūrā ietver: (Touvron et al., 2023a) ir . Based on preliminary experimental findings, we have opted for the untied embedding approach instead of tying the weights of input embedding and output projection. This decision was made in order to achieve better performance with the price of memory costs. Embedding and output projection . We have chosen RoPE (Rotary Positional Embedding) as our preferred option for incorporating positional information into our model. RoPE has been widely adopted and has demonstrated success in contemporary large language models, notably PaLM and LLaMA In particular, we have opted to use FP32 precision for the inverse frequency matrix, rather than BF16 or FP16, in order to prioritize model performance and achieve higher accuracy. Positional embedding (Su et al., 2021) (Chowdhery et al., 2022; Anil et al., 2023) (Touvron et al., 2023a;b). . For most layers, we remove biases following but we add biases in the QKV layer of attention to enhance the extrapolation ability of the model Bias Chowdhery et al. (2022), (Su, 2023b). . In modern Transformer models, pre-normalization is the most widely used approach, which has been shown to improve training stability compared to post-normalization. Recent research has suggested alternative methods for better training stability, which we plan to explore in future versions of our model. Additionally, we have replaced the traditional layer normalization technique described in with RMSNorm This change has resulted in equivalent performance while also improving efficiency. Pre-Norm & RMSNorm (Ba et al., 2016) (Jiang et al., 2023). . We have selected SwiGLU as our activation function, a combination of Swish and Gated Linear Unit Our initial experiments have shown that activation functions based on GLU generally outperform other baseline options, such as GeLU As is common practice in previous research, we have reduced the dimension of the feed-forward network (FFN) from 4 times the hidden size to 8/3 of the hidden size. Activation function (Shazeer, 2020) (Ramachandran et al., 2017) (Dauphin et al., 2017). (Hendrycks & Gimpel, 2016). 2.4 Training To train QWEN, we follow the standard approach of autoregressive language modeling, as described in Tas ietver modeļa apmācību, lai prognozētu nākamo žetonu, pamatojoties uz kontekstu, ko nodrošināja iepriekšējie žetoni.Mēs apmācām modeļus ar 2048 kontekstu garumiem.Lai izveidotu datu partijas, mēs shuffle un apvienojam dokumentus, un pēc tam sagriež tos līdz norādītajam kontekstu garumam.Lai uzlabotu aprēķinu efektivitāti un samazinātu atmiņas izmantošanu, mēs izmantojam Flash Attention uzmanības moduļos We adopt the standard optimizer AdamW optimizācijas priekšapmācībai. mēs iestatām hiperparametrus 1 = 0*.*9, 2 = 0*.*95, and = 10−8.Mēs izmantojam cosine mācīšanās ātruma grafiku ar noteiktu maksimālo mācīšanās ātrumu katram modeļa izmēram.Mācīšanās ātrums tiek samazināts līdz minimālajam mācīšanās ātrumam, kas ir 10% no maksimālā mācīšanās ātruma.Visi modeļi tiek apmācīti ar BFloat16 jauktu precizitāti treniņa stabilitātei. Radford et al. (2018). (Dāvis un al. 2022). (Kingma & Ba, 2014. gadā; Loshchilov & Hutter, 2017. gadā β β ϵ 2.5 Konteksta garuma pagarinājums Transformer models have a significant limitation in terms of the context length for their attention mechanism. As the context length increases, the quadratic-complexity computation leads to a drastic increase in both computation and memory costs. In this work, we have implemented simple training-free techniques that are solely applied during inference to extend the context length of the model. One of the key techniques we have used is NTK-aware interpolation (bloc97, 2023). Unlike position interpolation (PI) which scales each dimension of RoPE equally, NTK-aware interpolation adjusts the base of RoPE to prevent the loss of high-frequency information in a training-free manner. To further improve performance, we have also implemented a trivial extension called dynamic NTK-aware interpolation, which is later formally discussed in It dynamically changes the scale by chunks, avoiding severe performance degradation. These techniques allow us to effectively extend the context length of Transformer models without compromising their computational efficiency or accuracy. (Chen et al., 2023a) (Peng et al., 2023a). QWEN additionally incorporates two attention mechanisms: LogN-Scaling and window attention LogN-Scaling rescales the dot product of the query and value by a factor that depends on the ratio of the context length to the training length, ensuring that the entropy of the attention value remains stable as the context length grows. Window attention restricts the attention to a limited context window, preventing the model from attending to tokens that are too far away. (Chiang & Cholak, 2022; Viņa ir, 2023a) (Bērziņš un al. 2020 gadā) We also observed that the long-context modeling ability of our model varies across layers, with lower layers being more sensitive in context length extension compared to the higher layers. To leverage this observation, we assign different window sizes to each layer, using shorter windows for lower layers and longer windows for higher layers. 2.6 Experimental Results To evaluate the zero-shot and few-shot learning capabilities of our models, we conduct a thorough benchmark assessment using a series of datasets. We compare QWEN with the most recent open-source base models, including LLaMA LLAMA 2 MPT Falcon Baichuan2 ( ChatGLM2 InternLM Vīrietis and StableBeluga2 Our evaluation covers a total of 7 popular benchmarks, which are MMLU (5-shot) C-Eval (5-shot) GSM8K (8-shot) MATH (4-shot) Cilvēks 0 (šaušana) MBPP (0-shot) and BBH (Big Bench Hard) (3 shot) We aim to provide a comprehensive summary of the overall performance of our models across these benchmarks. (Touvron et al., 2023a), (Pāradresēts no Alūksnes un Alūksnes) 2023b), (Mosaic ML, 2023), (Almazrouei et al., 2023), Yang et al., 2023 gadā) (ChatGLM2 Team, 2023), (InternLM Team, 2023), (Inc., 2023b), (Stability AI, 2023). (Dinamo un tā tālāk) 2020), (Huang et al., 2023), (Cobbe et al., 2021), (Hendrycks et al., 2021), (Chen et al., 2021), (Austin et al., 2021), (Dzintars un al., 2022. gadā Šajā novērtējumā mēs koncentrējamies uz pamata valodas modeļiem bez saskaņošanas un vācam bāzes līniju labākos rezultātus no oficiālajiem rezultātiem un OpenCompass The results are presented in Table (OpenCompass Team, 2023). 2. Our experimental results demonstrate that the three QWEN models exhibit exceptional performance across all downstream tasks. It is worth noting that even the larger models, such as LLaMA2-70B, are outperformed by QWEN-14B in 3 tasks. QWEN-7B also performs admirably, surpassing LLaMA2-13B and achieving comparable results to Baichuan2-13B. Notably, despite having a relatively small number of parameters, QWEN-1.8B is capable of competitive performance on certain tasks and even outperforms larger models in some instances. The findings highlight the impressive capabilities of the QWEN models, particularly QWEN-14B, and suggest that smaller models, such as QWEN-1.8B, can still achieve strong performance in certain applications. To evaluate the effectiveness of context length extension, Table presents the test results on arXiv in terms of perplexity (PPL). These results demonstrate that by combining NTK-aware interpolation, LogN-Scaling, and layer-wise window assignment, we can effectively maintain the performance of our models in the context of over 8192 tokens. 3 3 3 Alignment Pretrained large language models have been found to be not aligned with human behavior, making them unsuitable for serving as AI assistants in most cases. Recent research has shown that the use of alignment techniques, such as supervised finetuning (SFT) and reinforcement learning from human feedback (RLHF), can significantly improve the ability of language models to engage in natural conversation. In this section, we will delve into the details of how QWEN models have been trained using SFT and RLHF, and evaluate their performance in the context of chat-based assistance. 3.1 Supervised Finetuning To gain an understanding of human behavior, the initial step is to carry out SFT, which finetunes a pretrained LLM on chat-style data, including both queries and responses. In the following sections, we will delve into the details of data construction and training methods. 3.1.1 DATA To enhance the capabilities of our supervised finetuning datasets, we have annotated conversations in multiple styles. While conventional datasets contain a vast amount of data prompted with questions, instructions, and answers in natural language, our approach takes it a step further by annotating human-style conversations. This practice, inspired by aims at improving the model’s helpfulness by focusing on natural language generation for diverse tasks. To ensure the model’s ability to generalize to a wide range of scenarios, we specifically excluded data formatted in prompt templates that could potentially limit its capabilities. Furthermore, we have prioritized the safety of the language model by annotating data related to safety concerns such as violence, bias, and pornography. (Wei et al., 2021. gadā a) Ouyang et al. (2022), Papildus datu kvalitātei mēs esam novērojuši, ka apmācības metode var būtiski ietekmēt modeļa gala sniegumu. which is a versatile meta language capable of describing both the metadata (such as roles) and the content of a turn. This format enables the model to effectively distinguish between various types of information, including system setup, user inputs, and assistant outputs, among others. By leveraging this approach, we can enhance the model’s ability to accurately process and analyze complex conversational data. (OpenAI, 2022), 1.3 Apmācība Consistent with pretraining, we also apply next-token prediction as the training task for SFT. We apply the loss masks for the system and user inputs. More details are demonstrated in Section A.1.1. The model’s training process utilizes the AdamW optimizer, with the following hyperparameters: 1 set to 0*. 2 set to 0. set to 10−8. The sequence length is limited to 2048, and the batch size is 128. The model undergoes a total of 4000 steps, with the learning rate gradually increased over the first 1430 steps, reaching a peak of 2 10−6. To prevent overfitting, weight decay is applied with a value of 0. . .*0. β 9, β 95 un vēl × 1, dropout ir iestatīts uz 0 1, and gradient clipping is enforced with a limit of 1 3.2 Reinforcement Learning from Human Feedback While SFT has proven to be effective, we acknowledge that its generalization and creativity capa-bilities may be limited, and it is prone to overfitting. To address this issue, we have implemented Reinforcement Learning from Human Feedback (RLHF) to further align SFT models with human preferences, following the approaches of This process involves training a reward model and using Proximal Policy Optimization (PPO) to conduct policy training. Ouyang et al. (2022); Christiano et al. (2017). (Schulman et al., 2017) 3.2.1 Reward Model To create a successful reward model, like building a large language model (LLM), it is crucial to first undergo pretraining and then finetuning. This pretraining process, also known as preference model pretraining (PMP) necessitates a vast dataset of comparison data. This dataset consists of sample pairs, each containing two distinct responses for a single query and their corresponding preferences. Similarly, finetuning is also conducted on this type of comparison data, but with a higher quality due to the presence of quality annotations. (Bai et al., 2022b), During the fine-tuning phase, we gather a variety of prompts and adjust the reward model based on human feedback for responses from the QWEN models. To ensure the diversity and complexity of user prompts are properly taken into account, we have created a classification system with around 6600 detailed tags and implemented a balanced sampling algorithm that considers both diversity and complexity when selecting prompts for annotation by the reward model To generate a wide range of responses, we have utilized QWEN models of different sizes and sampling strategies, as diverse responses can help reduce annotation difficulties and enhance the performance of the reward model. These responses are then evaluated by annotators following a standard annotation guideline, and comparison pairs are formed based on their scores. (Lu et al., 2023). In creating the reward model, we utilize the same-sized pre-trained language model QWEN to initiate the process. It is important to mention that we have incorporated a pooling layer into the original QWEN model to extract the reward for a sentence based on a specific end token. Mācīšanās ātrums šim procesam ir iestatīts uz pastāvīgu vērtību 3 10−6, and the batch size is 64. Additionally, the sequence length is set to 2048, and the training process lasts for a single epoch. × We adopted the accuracy on the test dataset as an important but not exclusive evaluation metric for the reward model. In Table we report the test pairwise accuracy of PMP and reward models on diverse human preference benchmark datasets Specifically, QWEN Helpful-base and QWEN Helpful-online are our proprietary datasets. The responses in QWEN Helpful-base are generated from QWEN without RLHF, whereas QWEN Helpful-online includes responses from QWEN with RLHF. The results show that the PMP model demonstrates high generalization capabilities on out-of-distribution data, and the reward model demonstrates significant improvement on our QWEN reward datasets. 4, (Bai et al., 2022b; Stiennon et al., 2020; Ethayarajh et al., 2022; Lightman et al., 2023. gadā 3.2.2 Reinforcement Learning Our Proximal Policy Optimization (PPO) process involves four models: the policy model, value model, reference model, and reward model. Before starting the PPO procedure, we pause the policy model’s updates and focus solely on updating the value model for 50 steps. This approach ensures that the value model can adapt to different reward models effectively. During the PPO operation, we use a strategy of sampling two responses for each query simultaneously. This strategy has proven to be more effective based on our internal benchmarking evaluations. We set the KL divergence coefficient to 0*.*04 and normalize the reward based on the running mean. The policy and value models have learning rates of 1 10−6 and 5 10−6, respectively. To enhance training stability, we utilize value loss clipping with a clip value of 0*. . *0, ir ātrāks atlīdzības pieaugums, kas galu galā noved pie konsekventi augstākas vērtēšanas atlīdzības līdzīgos apstākļos. × × 15. For inference, the policy top-p is set to 0 9. Our findings indicate that although the entropy is slightly lower than when top-p is set to 1 Additionally, we have implemented a pretrained gradient to mitigate the alignment tax. Empirical findings indicate that, with this specific reward model, the KL penalty is adequately robust to counteract the alignment tax in benchmarks that are not strictly code or math in nature, such as those that test common sense knowledge and reading comprehension. It is imperative to utilize a significantly larger volume of the pretrained data in comparison to the PPO data to ensure the effectiveness of the pretrained gradient. Additionally, our empirical study suggests that an overly large value for this coefficient can considerably impede the alignment to the reward model, eventually compromising the ultimate alignment, while an overly small value would only have a marginal effect on alignment tax reduction. Automātiska un cilvēciska saskaņoto modeļu novērtēšana To showcase the effectiveness of our aligned models, we conduct a comparison with other aligned models on well-established benchmarks, including MMLU C-Eval GSM8K HumanEval and BBH Besides the widely used few-shot setting, we test our aligned models in the zero-shot setting to demonstrate how well the models follow instructions. The prompt in a zero-shot setting consists of an instruction and a question without any previous examples in the context. The results of the baselines are collected from their official reports and OpenCompass (Hendrycks et al., 2020), (Huang et al., 2023), (Bērziņš un al., 2021), (Chen et al., 2021), (Suzgun et al., 2022). (OpenCompass Team, 2023). The results in Table demonstrate the effectiveness of our aligned models in understanding human instructions and generating appropriate responses. QWEN-14B-Chat outperforms all other models except ChatGPT (OpenAI, 2022) and LLAMA 2-CHAT-70B (Touvron et al., 2023b) in all datasets, including MMLU (Hendrycks et al., 2020), C-Eval (Huang et al., 2023), GSM8K (Cobbe et al., 2021), HumanEval (Chen et al., 2021), and BBH (Suzgun et al., 2022). 5 In particular, QWEN’s performance in HumanEval, which measures the quality of generated codes, is significantly higher than that of other open-source models. Moreover, QWEN’s performance is consistently better than that of open-source models of similar size, such as LLaMA2 ChatGLM2 InternLM and Baichuan2 This suggests that our alignment approach, which involves fine-tuning the model on a large dataset of human conversations, has been effective in improving the model’s ability to understand and generate human-like language./im (Touvron et al., 2023b), (ChatGLM2 Team, 2023), (InternLM Team, 2023), (Yang et al., 2023). Despite this, we have reservations about the ability of traditional benchmark evaluation to accurately measure the performance and potential of chat models trained with alignment techniques in today’s landscape. The results mentioned earlier provide some evidence of our competitive standing, but we believe that it is crucial to develop new evaluation methods specifically tailored to aligned models. We believe that human evaluation is crucial, which is why we have created a carefully curated dataset for this purpose. Our process involved collecting 300 instructions in Chinese that covered a wide range of topics, including knowledge, language understanding, creative writing, coding, and mathematics. To evaluate the performance of different models, we chose the SFT version of QWEN-CHAT-7B and the SFT and RLHF versions of QWEN-CHAT-14B, and added two strong baselines, GPT-3.5 and GPT-4 , for comparison. For each instruction, we asked three annotators to rank the model responses by the overall score of helpfulness, informativeness, validity, and other relevant factors. Our dataset and evaluation methodology provides a comprehensive and rigorous assessment of the capabilities of different language models in various domains. 4 Figure illustrates the win rates of the various models. For each model, we report the percentage of wins, ties, and losses against GPT-3.5, with the segments of each bar from bottom to top representing these statistics. The experimental results clearly demonstrate that the RLHF model outperforms the SFT models by significant margins, indicating that RLHF can encourage the model to generate responses that are more preferred by humans. In terms of overall performance, we find that the RLHF model significantly outperforms the SFT models, falling behind GPT-4. This indicates the effectiveness of RLHF for aligning to human preference. To provide a more comprehensive understanding of the models’ performance, we include a case study with examples from different models in Appendix Nonetheless, it remains difficult to accurately capture the gap between our models and the proprietary models. As such, a more extensive and rigorous assessment is required for the chat models. 4 A.2.2. 3.4 Tool Use, Code Interpreter, and Agent The QWEN models, which are designed to be versatile, have the remarkable ability to assist with (semi-)automating daily tasks by leveraging their skills in tool-use and planning. As such, they can serve as agents or copilots to help streamline various tasks. We explore QWEN’s proficiency in the following areas: • Utilizing unseen tools through ReAct prompting Skatīt galdu (Jānis un Alksnis) 2022) 6). • Using a Python code interpreter to enhance math reasoning, data analysis, and more (see Table and Table 7 8). • Functioning as an agent that accesses Hugging Face’s extensive collection of multimodal models while engaging with humans (see Table 9 ) To enhance QWEN’s capabilities as an agent or copilot, we employ the self-instruct strategy for SFT. Specifically, we utilize the in-context learning capability of QWEN for self-instruction. By providing a few examples, we can prompt QWEN to generate more relevant queries and generate outputs that follow a specific format, such as ReAct We then apply rules and involve human annotators to filter out any noisy samples. Afterwards, the samples are incorporated into QWEN’s training data, resulting in an updated version of QWEN that is more dependable for self-instruction. We iterate through this process multiple times until we gather an ample number of samples that possess both exceptional quality and a wide range of diversity. As a result, our final collection consists of around 2000 high-quality samples. (Wang et al., 2023c) (Yao et al., 2022). During the finetuning process, we mix these high-quality samples with all the other general-purpose SFT samples, rather than introducing an additional training stage. By doing so, we are able to retain essential general-purpose capabilities that are also pertinent for constructing agent applications. We have created and made publicly available a benchmark for evaluating QWEN’s ability to call plugins, tools, functions, or APIs using ReAct Prompting (see To ensure fair evaluation, we have excluded any plugins that were included in QWEN’s training set from the evaluation set. The benchmark assesses the model’s accuracy in selecting the correct plugin from a pool of up to five candidates, as well as the plausibility of the parameters passed into the plugin and the frequency of false positives. In this evaluation, a false positive occurs when the model incorrectly invokes a plugin in response to a query, despite not being required to do so. Using Tools via ReAct Prompting Qwen komanda, Alibaba grupa, 2023 b) The results presented in Table demonstrate that QWEN consistently achieves higher accuracy in identifying the relevance of a query to the available tools as the model size increases. However, the table also highlights that beyond a certain point, there is little improvement in performance when it comes to selecting the appropriate tool and providing relevant arguments. This suggests that the current preliminary benchmark may be relatively easy and may require further enhancement in future iterations. It is worth noting that GPT-3.5 stands out as an exception, displaying suboptimal performance on this particular benchmark. This could potentially be attributed to the fact that the benchmark primarily focuses on the Chinese language, which may not align well with GPT-3.5’s capabilities. Additionally, we observe that GPT-3.5 tends to attempt to use at least one tool, even if the query cannot be effectively addressed by the provided tools. 6 The Python code interpreter is widely regarded as a powerful tool for augmenting the capabilities of an LLM agent. It is worth investigating whether QWEN can harness the full potential of this interpreter to enhance its performance in diverse domains, such as mathematical reasoning and data analysis. To facilitate this exploration, we have developed and made publicly available a benchmark that is specifically tailored for this purpose (see Using Code Interpreter for Math Reasoning and Data Analysis Qwen komanda, Alibaba grupa, 2023a). The benchmark encompasses three primary categories of tasks: math problem-solving, data visualization, and other general-purpose tasks like file post-processing and web crawling. Within the visualization tasks, we differentiate between two levels of difficulty. The easier level can be achieved by simply writing and executing a single code snippet without the need for advanced planning skills. However, the more challenging level requires strategic planning and executing multiple code snippets in a sequential manner. This is because the subsequent code must be written based on the output of the previous code. For example, an agent may need to examine the structure of a CSV file using one code snippet before proceeding to write and execute additional code to create a plot. Regarding evaluation metrics, we consider both the executability and correctness of the generated code. To elaborate on the correctness metrics, for math problems, we measure accuracy by verifying if the ground truth numerical answer is present in both the code execution result and the final response. When it comes to data visualization, we assess accuracy by utilizing QWEN-VL a powerful multimodal language model. QWEN-VL is capable of answering text questions paired with images, and we rely on it to confirm whether the image generated by the code fulfills the user’s request. (Bai et al., 2023), The results regarding executability and correctness are presented in Table and Table respectively. It is evident that CODE LLAMA generally outperforms LLAMA 2, its generalist counterpart, which is not surprising since this benchmark specifically requires coding skills. However, it is worth noting that specialist models that are optimized for code synthesis do not necessarily outperform generalist models. This is due to the fact that this benchmark encompasses various skills beyond coding, such as abstracting math problems into equations, understanding language-specified constraints, and responding in the specified format such as ReAct. Notably, QWEN-7B-CHAT and QWEN-14B-CHAT surpass all other open-source alternatives of similar scale significantly, despite being generalist models. 7 8, Hugging Face provides a framework called the Hugging Face Agent or Transformers Agent which empowers LLM agents with a curated set of multimodal tools, including speech recognition and image synthesis. This framework allows an LLM agent to interact with humans, interpret natural language commands, and employ the provided tools as needed. Serving as a Hugging Face Agent (Hugging Face, 2023 gadā) To evaluate QWEN’s effectiveness as a Hugging Face agent, we utilized the evaluation benchmarks offered by Hugging Face. The results are presented in Table The evaluation results reveal that QWEN performs quite well in comparison to other open-source alternatives, only slightly behind the proprietary GPT-4, demonstrating QWEN’s competitive capabilities. 9. 4 Code-Qwen: Specialized Model for Coding Training on domain-specific data has been shown to be highly effective, particularly in the case of code pretraining and finetuning. A language model that has been reinforced with training on code data can serve as a valuable tool for coding, debugging, and interpretation, among other tasks. In this work, we have developed a series of generalist models using pretraining and alignment techniques. Building on this foundation, we have created domain-specific models for coding by leveraging the base language models of QWEN, including continued pretrained model, CODE-QWEN and supervised finetuned model, CODE-QWEN-CHAT. Both models have 14 billion and 7 billion parameters versions. 4.1 Code Pretraining We believe that relying solely on code data for pretraining can result in a significant loss of the ability to function as a versatile assistant. Unlike previous approaches that focused solely on pretraining on code data we take a different approach by starting with our base models QWEN trained on a combination of text and code data, and then continuing to pretrain on the code data. We continue to pretrain the models on a total of around 90 billion tokens. During the pre-training phase, we initialize the model using the base language models QWEN. Many applications that rely on specialized models for coding may encounter lengthy contextual scenarios, such as tool usage and code interpretation, as mentioned in Section To address this issue, we train our models with context lengths of up to 8192. Similar to base model training in Section we employ Flash Attention in the attention modules, and adopt the standard optimizer AdamW setting 1 = 0*. 2 = 0. = 10−8. We set the learning rate as 6. 10−5 for CODE-QWEN-14B and 3.*0 10−5 for CODE-QWEN-7B, with 3% warm up iterations and no learning rate decays. (Li et al., 2022; 2023d), (Rozie`re et al., 2023) 3.4. 2.4, (Dao et al., 2022) (Kingma & Ba, 2014; Loshchilov & Hutter, 2017), β 9, β 95, and ϵ 0 × × 4.2 Code Supervised Fine-Tuning After conducting a series of empirical experiments, we have determined that the multi-stage SFT strategy yields the best performance compared to other methods. In the supervised fine-tuning stage, the model CODE-QWEN-CHAT initialized by the code foundation model CODE-QWEN are optimized by the AdamW optimizer ( 1 = 0*. 2 = 0.*95, = 10−8) with a learning rate of 2*. 10−6 un 1.*0 10−5 for the 14B and 7B model respectively. The learning rate increases to the peaking value with the cosine learning rate schedule (3% warm-up steps) and then remains constant. (Kingma & Ba, 2014; Loshchilov & Hutter, 2017) β 9, β ϵ 0 × × 4.3 Evaluation Our CODE-QWEN models have been compared with both proprietary and open-source language models, as shown in Tables and These tables present the results of our evaluation on the test sets of Humaneval MBPP and the multi-lingual code generation benchmark HUMANEVALPACK The comparison is based on the pass@1 performance of the models on these benchmark datasets. The results of this comparison are clearly demonstrated in Tables and 10 11. (Chen et al., 2021), (Austin et al., 2021), (Muennighoff et al., 2023. gadā 10 11. Our analysis reveals that specialized models, specifically CODE-QWEN and CODE-QWEN-CHAT, sig-nificantly outperform previous baselines with similar parameter counts, such as OCTOGEEX InstructCodeT5+ and CodeGeeX2 In fact, these models even rival the performance of larger models like Starcoder (Muen-nighoff et al., 2023 gadā) (Wang et al., 2023d), (Zheng et al., 2023. gadā (Li et al., 2023d). Salīdzinot ar dažiem ārkārtīgi lieliem slēgtā koda modeļiem, CODE-QWEN un CODE-QWEN-CHAT demonstrē skaidras priekšrocības pass@1 ziņā.Tomēr ir svarīgi atzīmēt, ka šie modeļi atpaliek no modernākajām metodēm, piemēram, GPT-4, kopumā. Ir svarīgi uzsvērt, ka iepriekš minētie novērtējumi nav pietiekami, lai pilnībā izprastu modeļu stiprās un vājās puses.Mūsu viedoklī ir nepieciešams izstrādāt stingrākus testus, lai mēs varētu precīzi novērtēt mūsu relatīvo sniegumu salīdzinājumā ar GPT-4. 5 Math-Qwen: Specialized Model for Mathematics Reasoning We have created a mathematics-specialized model series called MATH-QWEN-CHAT, which is built on top of the QWEN pretrained language models. Specifically, we have developed assistant models that are specifically designed to excel in arithmetic and mathematics and are aligned with human behavior. We are releasing two versions of this model series, MATH-QWEN-14B-CHAT and MATH-QWEN-7B-CHAT, which have 14 billion and 7 billion parameters, respectively. 5.1 Training We carry out math SFT on our augmented math instructional dataset for mathematics reasoning, and therefore we obtain the chat model, MATH-QWEN-CHAT, directly. Owing to shorter average lengths of the math SFT data, we use a sequence length of 1024 for faster training. Most user inputs in the math SFT dataset are examination questions, and it is easy for the model to predict the input format and it is meaningless for the model to predict the input condition and numbers which could be random. Thus, we mask the inputs of the system and user to avoid loss computation on them and find masking them accelerates the convergence during our preliminary experiments. For optimization, we use the AdamW optimizer with the same hyperparameters of SFT except that we use a peak learning rate of 2 10−5 and a training step of 50 000. × 5.2 Evaluation We evaluate models on the test sets of GSM8K (Grade school math) MATH (Challenging competition math problems) Math401 (Arithmetic ability) ( and Math23K (Chinese grade school math) Mēs salīdzinām MATH-QWEN-CHAT ar patentētajiem modeļiem ChatGPT un Minerva and open-sourced math-specialized model RFT ( WizardMath and GAIRMath-Abel in Table MATH-QWEN-CHAT models show better math reasoning and arithmetic abilities compared to open-sourced models and QWEN-CHAT models of similar sizes. Compared to proprietary models, MATH-QWEN-7B-CHAT outperforms Minerva-8B in MATH. MATH-QWEN-14B-CHAT is chasing Minerva-62B and GPT-3.5 in GSM8K and MATH and delivers better performance on arithmetic ability and Chinese math problems. (Cobbe et al., 2021), (Hendrycks et al., 2021), Yuan et al., 2023b), (Wang et al., 2017). (Lewkowycz et al., 2022) Yuan et al., 2023a), (Luo et al., 2023a), (Chern et al., 2023a) 12. 6 Related Work 6.1 Large Language Models The excitement of LLM began with the introduction of the Transformer architecture ( which was then applied to pretraining large-scale data by researchers such as These efforts led to significant success in transfer learning, with model sizes growing from 100 million to over 10 billion parameters Vaswani et al., 2017), Radford et al. (2018); Devlin et al. (2018); Liu et al. (2019). (Raffel et al., 2020; Šūmahers et al. 2019). In 2020, the release of GPT-3, a massive language model that is 10 times larger than T5, demonstrated the incredible potential of few-shot and zero-shot learning through prompt engineering and in-context learning, and later chain-of-thought prompting This success has led to a number of studies exploring the possibilities of further scaling these models As a result, the community has come to view these large language models as essential foundations for downstream models (Līdzīgi kā Al. 2022c). (Scao et al., 2022; Zhang et al., 2022; Du et al., 2021; Zeng et al., 2022; Lepikhin et al., 2020; Fedus et al., 2022; Du et al., 2022; Black et al., 2022; Rae et al., 2021; Hoffmann et al., 2022; Chowdhery et al., 2022; Thoppilan et al., 2022). (Bommasani et al., 2021). The birth of ChatGPT and the subsequent launch of GPT-4 marked two historic moments in the field of artificial intelligence, demonstrating that large language models (LLMs) can serve as effective AI assistants capable of communicating with humans. These events have sparked interests among researchers and developers in building language models that are aligned with human values and potentially even capable of achieving artificial general intelligence (AGI) (OpenAI, 2022) (OpenAI, 2023) (Anil et al., 2023; Anthropic, 2023a;b). One notable development in this area is the emergence of open-source LLMs, specifically LLaMA and LLAMA 2 which have been recognized as the most powerful open-source language models ever created. This has led to a surge of activity in the open-source community with a series of large language models being developed collaboratively to build upon this progress (Touvron et al., 2023a) (Touvron et al., 2023b), (Wolf et al., 2019), (Mosaic ML, 2023; Almazrouei et al., 2023; ChatGLM2 Team, 2023; Yang et al., 2023; InternLM Team, 2023). 6.2 Alignment The community was impressed by the surprising effectiveness of alignment on LLMs. Previously, LLMs without alignment often struggle with issues such as repetitive generation, hallucination, and deviation from human preferences. Since 2021, researchers have been diligently working on developing methods to enhance the performance of LLMs in downstream tasks Furthermore, researchers have been actively exploring ways to align LLMs with human instructions One major challenge in alignment research is the difficulty of collecting data. While OpenAI has utilized its platform to gather human prompts or instructions, it is not feasible for others to collect such data. (Wei et al., 2022a; Sanh et al., 2021; Longpre et al., 2023; Chung et al., 2022; Muennighoff et al., 2022). (Ouyang et al., 2022. gadā Askell et al., 2021; Bai et al., 2022b;c). However, there has been some progress in this area, such as the self-instruct approach proposed in This innovative work offers a potential solution to the data collection problem in alignment research. As a result, there has been a surge in open-source chat data, including Alpaca MOSS Dolly Evol-Instruct and others Similarly, there has been an increase in open-source chat models, such as Alpaca Vicuna Guanaco MOSS WizardLM un citi Wang et al. (2023c). (Taori et al., 2023), (Sun et al., 2023a), (Conover et al., 2023), (Xu et al., 2023b), (Sun et al., 2023b; Xu et al., 2023a;c; Chen et al., 2023c; Ding et al., 2023; Ji et al., 2023; Yang, 2023). (Taori et al., 2023), (Chiang et al., 2023), (Dettmers et al., 2023), (Sun et al., 2023a), (Xu et al., 2023b), (Xu et al., 2023c; Chen et al., 2023 C Ding et al., 2023; Wang et al., 2023b). To train an effective chat model, available solutions are mostly based on SFT and RLHF While SFT is similar to pretraining, it focuses on instruction following using the aforementioned data. However, for many developers, the limited memory capacity is a major obstacle to further research in SFT. As a result, parameter-efficient tuning methods, such as LoRA ., and Q-LoRA have gained popularity in the community. LoRA tunes only low-rank adapters, while Q-LoRA builds on LoRA and utilizes 4-bit quantized LLMs and paged attention In terms of RLHF, recent methods such as PPO have been adopted, but there are also alternative techniques aimed at addressing the complexity of optimization, such as RRHF ( DPO and PRO Despite the ongoing debate about the effectiveness of RLHF, more evidence is needed to understand how it enhances the intelligence of LLMs and what potential drawbacks it may have. (Ouyang et al., 2022). (Hu et al 2021) (Dettmers et al., 2023), (Dettmers et al., 2022; Frantar et al., 2022; Kwon et al., 2023). (Schulman et al., 2017; Touvron et al., 2023b) Yuan et al., 2023c), (Rafailov et al., 2023), (Song et al., 2023. gadā 6.3 Tool Use and Agents LLM’s planning function allows for the invocation of tools, such as APIs or agent capabilities, through in-context learning, as demonstrated by introduced ReAct, a generation format that enables the model to generate thoughts on which tool to use, accept input from API observations, and generate a response. GPT-3.5 and GPT-4, when prompted with few shots, have shown consistent and impressive performance. In addition to tool usage, LLMs can utilize external memory sources like knowledge bases or search engines to generate more accurate and informative answers. This has led to the popularity of frameworks like LangChain The research on LLMs for tool use has also sparked interest in building agents with LLM capabilities, such as agents that can call different AI models embodied lifelong learning or multimodal agents un vairāki aģenti, kas mijiedarbojas viens ar otru un pat veido mikro sabiedrību Schick et al. Tātad 2023 Yao et al. Dzīve (2022) (Hu et al., 2023; Zhong et al., 2023b) (Nakano et al., 2021; Liu et al., 2023b) (LangChain, Inc., 2023). (Shen et al., 2023; Li et al., 2023a), (Wang et al., 2023a; Driess et al., 2023), (Chen et al., 2023b; Li et al., 2023b; Xu et al., 2023d; Hong et al., 2023). 6.4 LLM for Coding Previous research has demonstrated that LLMs possess remarkable capabilities in code understanding and generation, particularly those with massive numbers of parameters Moreover, several LLMs have been pre-trained, continued pre-trained, or fine-tuned on coding-related data, which has resulted in significantly improved performance compared to general-purpose LLMs. These models include Codex AlphaCode SantaCoder Starcoder-Base InCoder CodeT5 CodeGeeX and CODE LLAMA In addition to these models, recent studies have focused on developing specialized alignment techniques for coding, such as Code Llama-Instruct and StarCoder These models can assist developers in various code-related tasks, including code generation code completion code translation bug fixing code refinement and code question answering In a word, LLMs have the potential to revolutionize the field of coding by providing developers with powerful tools for code comprehension, generation, and related tasks. (Chowdhery et al., 2022; Anil et al., 2023; Rae et al., 2021; Hoffmann et al., 2022). Chen et al. (2021), (Li et al., 2022), (Allal et al., 2023), (Li et al., 2023d), (Fried et al., 2022), (Wang et al., 2021), (Zheng et al., 2023), (Rozie`re et al., 2023). (Rozie`re et al., 2023) (Li et al., 2023d). (Chen et al., 2021; Austin et al., 2021), (Zhang et al., 2023a), (Szafraniec et al., 2023), (Muennighoff et al., 2023), (Liu et al., 2023c), (Liu & Wan, 2021). 6.5 LLM matemātikā LLMs with a certain model scale have been found to possess the ability to perform mathematical reasoning In order to encourage LLMs to achieve better performance on math-related tasks, researchers have employed techniques such as chain-of-thought prompting and scratchpad which have shown promising results. Additionally, self-consistency and least-to-most prompting have further improved the performance of these models on these tasks. However, prompt engineering is a time-consuming process that requires a lot of trial and error, and it is still difficult for LLMs to consistently perform well or achieve satisfactory results in solving mathematical problems. Moreover, simply scaling the data and model size is not an efficient way to improve a model’s mathematical reasoning abilities. Instead, pretraining on math-related corpora has been shown to consistently enhance these capabilities Additionally, fine-tuning on math-related instruction-following datasets has also been effective and more cost-effective than math-specific pretraining. Despite their limitations in terms of accuracy, LLMs still have significant potential to assist users with practical mathematical problems. There is ample scope for further development in this area. (Wei et al., 2022b; Suzgun et al., 2022). (Wei et al., 2022c) (Nye et al., 2021), (Wang et al., 2022) (Zhou et al., 2022) (Dinamo un tā tālāk) 2021; Lewkowycz et al., 2022; Taylor et al., 2022; Lāčplēsis et al. 2023). (Si et al., 2023; Yuan et al., 2023 a Luo et al., 2023a; Yue et al., 2023; Chern et al., 2023a; Yu et al., 2023 gadā) 7 Conclusion In this report, we present the QWEN series of large language models, which showcase the latest advancements in natural language processing. With 14B, 7B, and 1.8B parameters, these models have been pre-trained on massive amounts of data, including trillions of tokens, and fine-tuned using cutting-edge techniques such as SFT and RLHF. Additionally, the QWEN series includes specialized models for coding and mathematics, such as CODE-QWEN, CODE-QWEN-CHAT, and MATH-QWEN-CHAT, which have been trained on domain-specific data to excel in their respective fields. Our results demonstrate that the QWEN series is competitive with existing open-source models and even matches the performance of some proprietary models on comprehensive benchmarks and human evaluation. We believe that the open access of QWEN will foster collaboration and innovation within the community, enabling researchers and developers to build upon our work and push the boundaries of what is possible with language models. By providing these models to the public, we hope to inspire new research and applications that will further advance the field and contribute to our understanding of the variables and techniques introduced in realistic settings. In a nutshell, the QWEN series represents a major milestone in our development of large language models, and we are excited to see how it will be used to drive progress and innovation in the years to come. References Loubna Ben Allal, Raymond Li, Denis Kocetkov, Chenghao Mou, Christopher Akiki, Carlos Munoz Ferrandis, Niklas Muennighoff, Mayank Mishra, Alex Gu, Manan Dey, et al. SantaCoder: Don’t reach for the stars! , 2023. arXiv preprint arXiv:2301.03988 Ebtesam Almazrouei, Hamza Alobeidli, Abdulaziz Alshamsi, Alessandro Cappelli, Ruxandra Co-jocaru, Merouane Debbah, Etienne Goffinet, Daniel Heslow, Julien Launay, Quentin Malartic, Badreddine Noune, Baptiste Pannier, and Guilherme Penedo. Falcon-40B: An open large language model with state-of-the-art performance, 2023. Rohan Anil, Andrew M Dai, Orhan Firat, Melvin Johnson, Dmitry Lepikhin, Alexandre Passos, Siamak Shakeri, Emanuel Taropa, Paige Bailey, Zhifeng Chen, et al. PaLM 2 technical report. , 2023. arXiv preprint arXiv:2305.10403 Anthropic. Introducing Claude, 2023a. URL .nthropic. Claude 2. Technical report, Anthropic, 2023b. URL . . https://www.anthropic.com/index/ introducing-claude https://www-files anthropic.com/production/images/Model-Card-Claude-2.pdf Vamsi Aribandi, Yi Tay, Tal Schuster, Jinfeng Rao, Huaixiu Steven Zheng, Sanket Vaibhav Mehta, Honglei Zhuang, Vinh Q Tran, Dara Bahri, Jianmo Ni, et al. ExT5: Towards extreme multi-task scaling for transfer learning. , 2021. arXiv preprint arXiv:2111.10952 Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Ben Mann, Nova DasSarma, et al. A general language assistant as a laboratory for alignment. , 2021. arXiv preprint arXiv:2112.00861 Jacob Austin, Augustus Odena, Maxwell Nye, Maarten Bosma, Henryk Michalewski, David Dohan, Ellen Jiang, Carrie Cai, Michael Terry, Quoc Le, et al. Program synthesis with large language models. , 2021. arXiv preprint arXiv:2108.07732 AutoGPT. AutoGPT: The heart of the open-source agent ecosystem, 2023. URL . https:// github.com/Significant-Gravitas/Auto-GPT Lei Jimmy Ba, Jamie Ryan Kiros, and Geoffrey E. Hinton. Layer normalization. , abs/1607.06450, 2016. URL . CoRR http://arxiv.org/abs/1607.06450 Jinze Bai, Rui Men, Hao Yang, Xuancheng Ren, Kai Dang, Yichang Zhang, Xiaohuan Zhou, Peng Wang, Sinan Tan, An Yang andf Zeyu Cui, Yu Han, Shuai Bai, Wenbin Ge, Jianxin Ma, Junyang Lin, Jingren Zhou, and Chang Zhou. OFASys: A multi-modal multi-task learning system for building generalist models. , abs/2212.04408, 2022a. doi: 10.48550/arXiv.2212.04408. URL . CoRR https://doi.org/10.48550/arXiv.2212.04408 Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, and Jingren Zhou. Qwen-VL: A versatile vision-language model for understanding, localization, text reading, and beyond. , abs/2308.12966, 2023. doi: 10.48550/arXiv.2308.12966. URL . CoRR https://doi.org/10.48550/arXiv.2308.12966 Yuntao Bai, Andy Jones, Kamal Ndousse, Amanda Askell, Anna Chen, Nova DasSarma, Dawn Drain, Stanislav Fort, Deep Ganguli, Tom Henighan, et al. Training a helpful and harmless assistant with reinforcement learning from human feedback. , 2022b. arXiv preprint arXiv:2204.05862 Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, et al. Constitutional AI: Harmlessness from AI feedback. , 2022c. arXiv preprint arXiv:2212.08073 Iz Beltagi, Matthew E. Peters un Arman Cohan. Longformer: ilgdokumentu transformators. , 2020. arXiv preprint arXiv:2004.05150 Yonatan Bisk, Rowan Zellers, Ronan Le Bras, Jianfeng Gao, and Yejin Choi. PIQA: reasoning about physical commonsense in natural language. In , pp. 7432–7439. AAAI Press, 2020. doi: The Thirty-Fourth AAAI Conference on Artificial Intelligence, AAAI 2020, The Thirty-Second Innovative Applications of Artificial Intelligence Con-ference, IAAI 2020, The Tenth AAAI Symposium on Educational Advances in Artificial Intelligence, EAAI 2020, New York, NY, USA, February 7-12, 2020 10.1609/aaai.v34i05.6239. URL . https://doi.org/10.1609/aaai.v34i05.6239 Sid Black, Stella Biderman, Eric Hallahan, Quentin Anthony, Leo Gao, Laurence Golding, Horace He, Connor Leahy, Kyle McDonell, Jason Phang, et al. GPT-NeoX-20B: An open-source autoregressive language model. , 2022. arXiv preprint arXiv:2204.06745 bloc97. NTK-aware scaled RoPE allows LLaMA models to have extended (8k+) con-text size without any fine-tuning and minimal perplexity degradation., 2023. URL . https://www.reddit.com/r/LocalLLaMA/comments/14lz7j5/ntkaware_ scaled_rope_allows_llama_models_to_have/ Rishi Bommasani, Drew A Hudson, Ehsan Adeli, Russ Altman, Simran Arora, Sydney von Arx, Michael S Bernstein, Jeannette Bohg, Antoine Bosselut, Emma Brunskill un citi. , 2021. arXiv preprint arXiv:2108.07258 Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell un citi. , 33:1877–1901, 2020. Advances in neural information processing systems ChatGLM2 Team. ChatGLM2-6B: An open bilingual chat LLM, 2023. URL . . https://github com/THUDM/ChatGLM2-6B Mark Chen, Jerry Tworek, Heewoo Jun, Qiming Yuan, Henrique Ponde´ de Oliveira Pinto, Jared Kaplan, Harrison Edwards, Yuri Burda, Nicholas Joseph, Greg Brockman, Alex Ray, Raul Puri, Gretchen Krueger, Michael Petrov, Heidy Khlaaf, Girish Sastry, Pamela Mishkin, Brooke Chan, Scott Gray, Nick Ryder, Mikhail Pavlov, Alethea Power, Lukasz Kaiser, Mohammad Bavarian, Clemens Winter, Philippe Tillet, Felipe Petroski Such, Dave Cummings, Matthias Plappert, Fotios Chantzis, Elizabeth Barnes, Ariel Herbert-Voss, William Hebgen Guss, Alex Nichol, Alex Paino, Nikolas Tezak, Jie Tang, Igor Babuschkin, Suchir Balaji, Shantanu Jain, William Saunders, Christopher Hesse, Andrew N. Carr, Jan Leike, Joshua Achiam, Vedant Misra, Evan Morikawa, Alec Radford, Matthew Knight, Miles Brundage, Mira Murati, Katie Mayer, Peter Welinder, Bob McGrew, Dario Amodei, Sam McCandlish, Ilya Sutskever, and Wojciech Zaremba. Evaluating large language models trained on code. , abs/2107.03374, 2021. URL . . CoRR https://arxiv org/abs/2107.03374 Shouyuan Chen, Sherman Wong, Liangjian Chen, and Yuandong Tian. Extending context window of large language models via positional interpolation. , 2023a. arXiv preprint arXiv:2306.15595 Weize Chen, Yusheng Su, Jingwei Zuo, Cheng Yang, Chenfei Yuan, Chen Qian, Chi-Min Chan, Yujia Qin, Yaxi Lu, Ruobing Xie, et al. Agentverse: Facilitating multi-agent collaboration and exploring emergent behaviors in agents. , 2023b. arXiv preprint arXiv:2308.10848 Zhihong Chen, Feng Jiang, Junying Chen, Tiannan Wang, Fei Yu, Guiming Chen, Hongbo Zhang, Juhao Liang, Chen Zhang, Zhiyi Zhang, et al. Phoenix: Democratizing ChatGPT across languages. , 2023c. arXiv preprint arXiv:2304.10453 Ethan Chern, Haoyang Zou, Xuefeng Li, Jiewen Hu, Kehua Feng, Junlong Li, and Pengfei Liu. Generative ai for math: Abel. , 2023a. https://github.com/GAIR-NLP/abel I Chern, Steffi Chern, Shiqi Chen, Weizhe Yuan, Kehua Feng, Chunting Zhou, Junxian He, Graham Neubig, Pengfei Liu, et al. Factool: Factuality detection in generative ai–a tool augmented framework for multi-task and multi-domain scenarios. , 2023b. arXiv preprint arXiv:2307.13528 David Chiang and Peter Cholak. Overcoming a theoretical limitation of self-attention. In , pp. 7654–7664, 2022. Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) Wei-Lin Chiang, Zhuohan Li, Zi Lin, Ying Sheng, Zhanghao Wu, Hao Zhang, Lianmin Zheng, Siyuan Zhuang, Yonghao Zhuang, Joseph E. Gonzalez, Ion Stoica, and Eric P. Xing. Vicuna: An open-source chatbot impressing GPT-4 with 90%* ChatGPT quality, March 2023. URL . https://lmsys.org/blog/2023-03-30-vicuna/ Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, et al. PaLM: Scaling language modeling with pathways. , 2022. arXiv preprint arXiv:2204.02311 Paul F. Christiano, Jan Leike, Tom B. Brown, Miljan Martic, Shane Legg, and Dario Amodei. Deep reinforcement learning from human preferences. In Isabelle Guyon, Ulrike von Luxburg, Samy Bengio, Hanna M. Wallach, Rob Fergus, S. V. N. Vishwanathan, and Roman Garnett (eds.), , 4299–4307, 2017. gada URL . Neironu informācijas apstrādes sistēmu attīstība 30: gada konference par Neu-ral informācijas apstrādes sistēmām 2017, 2017. gada 4. un 9. decembris, Long Beach, CA, ASV https://proceedings.neurips.cc/paper/2017/hash/ d5e2c0adad503c91f91df240d0cd4e49-Abstract.html Hyung Won Chung, Le Hou, Shayne Longpre, Barret Zoph, Yi Tay, William Fedus, Eric Li, Xuezhi Wang, Mostafa Dehghani, Siddhartha Brahma, et al. Scaling instruction-finetuned language models. , 2022. arXiv preprint arXiv:2210.11416 Christopher Clark, Kenton Lee, Ming-Wei Chang, Tom Kwiatkowski, Michael Collins, and Kristina Toutanova. Boolq: Exploring the surprising difficulty of natural yes/no questions. In Jill Burstein, Christy Doran, and Thamar Solorio (eds.), , pp. 2924–2936. Association for Computational Linguistics, 2019. doi: 10.18653/v1/n19-1300. URL . Proceedings of the 2019 Conference of the North Amer-ican Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2019, Minneapolis, MN, USA, June 2-7, 2019, Volume 1 (Long and Short Papers) https://doi.org/10.18653/v1/n19-1300 Peter Clark, Isaac Cowhey, Oren Etzioni, Tushar Khot, Ashish Sabharwal, Carissa Schoenick un Oyvind Tafjord. , abs/1803.05457, 2018. URL . CoRR http://arxiv.org/abs/1803.05457 Karl Cobbe, Vineet Kosaraju, Mohammad Bavarian, Mark Chen, Heewoo Jun, Lukasz Kaiser, Matthias Plappert, Jerry Tworek, Jacob Hilton, Reiichiro Nakano, et al. Training verifiers to solve math word problems. , 2021. arXiv preprint arXiv:2110.14168 Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Fran-cisco Guzma´n, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. Unsupervised cross-lingual representation learning at scale. , 2019. arXiv pirmizrāde arXiv:1911.02116 Mike Conover, Matt Hayes, Ankit Mathur, Jianwei Xie, Jun Wan, Sam Shah, Ali Ghodsi, Patrick Wendell, Matei Zaharia, and Reynold Xin. Free Dolly: Introducing the world’s first truly open instruction-tuned LLM, 2023. URL . https://www.databricks.com/blog/2023/04/ 12/dolly-first-open-commercially-viable-instruction-tuned-llm Wenliang Dai, Junnan Li, Dongxu Li, Anthony Meng, Huat Tiong, Junqi Zhao, Weisheng Wang, Boyang Li, Pascale Fung un Steven Hoi. 2023. gadā arXiv preprint arXiv:2305.06500 Tri Dao, Daniel Y. Fu, Stefano Ermon, Atri Rudra un Christopher Re ́. FlashAt-tention: ātra un atmiņas efektīva precīza uzmanība ar io-apziņu. , 2022. URL . NeurIPS http://papers.nips.cc/paper_files/paper/2022/hash/ 67d57c32e20fd0a7a302cb81d36e40d5-Abstract-Konference.html Yann N Dauphin, Angela Fan, Michael Auli, and David Grangier. Language modeling with gated convolutional networks. In , pp. 933–941. PMLR, 2017. International conference on machine learning Tim Dettmers, Mike Lewis, Younes Belkada, and Luke Zettlemoyer. LLM.int8(): 8-bit matrix multiplication for transformers at scale. , 2022. arXiv preprint arXiv:2208.07339 Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. QLoRA: Efficient finetuning of quantized LLMs. 2023. gadā arXiv preprint arXiv:2305.14314 Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre-training of deep bidirectional transformers for language understanding. , 2018. arXiv preprint arXiv:1810.04805 Ning Ding, Yulin Chen, Bokai Xu, Yujia Qin, Zhi Zheng, Shengding Hu, Zhiyuan Liu, Maosong Sun, and Bowen Zhou. Enhancing chat language models by scaling high-quality instructional conversations. , 2023. arXiv preprint arXiv:2305.14233 Danny Driess, Fei Xia, Mehdi SM Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, et al. Palm-e: An embodied multimodal language model. 2023. gadā arXiv preprint arXiv:2303.03378 Nan Du, Yanping Huang, Andrew M Dai, Simon Tong, Dmitry Lepikhin, Yuanzhong Xu, Maxim Krikun, Yanqi Zhou, Adams Wei Yu, Orhan Firat, et al. GLaM: Efficient scaling of language models with mixture-of-experts. In , pp. 5547–5569. PMLR, 2022. International Conference on Machine Learning Zhengxiao Du, Yujie Qian, Xiao Liu, Ming Ding, Jiezhong Qiu, Zhilin Yang, and Jie Tang. GLM: General language model pretraining with autoregressive blank infilling. , 2021. arXiv preprint arXiv:2103.10360 Kawin Ethayarajh, Yejin Choi, and Swabha Swayamdipta. Understanding dataset difficulty with -usable information. In Kamalika Chaudhuri, Stefanie Jegelka, Le Song, Csaba Szepesvari, Gang Niu, and Sivan Sabato (eds.), , volume 162 of , 5988–6008. PMLR, 17–23 Jūlijs 2022. V Proceedings of the 39th International Conference on Machine Learning Mašīntulkošanas pētījumu rezultāti William Fedus, Barret Zoph, and Noam Shazeer. Switch transformers: Scaling to trillion parameter models with simple and efficient sparsity. , 23(1): 5232–5270, 2022. Mašīntulkošanas pētījumu žurnāls Elias Frantar, Saleh Ashkboos, Torsten Hoefler un Dan Alistarh. GPTQ: Precīza pēcapmācības kvantificēšana ģeneratīviem iepriekš apmācītajiem transformatoriem. , 2022. arXiv pirmizrāde arXiv:2210.17323 Daniel Fried, Armen Aghajanyan, Jessy Lin, Sida I. Wang, Eric Wallace, Freda Shi, Ruiqi Zhong, Wen tau Yih, Luke Zettlemoyer un Mike Lewis. , abs/2204.05999, 2022. ArXiv Google. svarīgs nākamais solis mūsu AI ceļojumā, 2023. URL . https://blog.google/ technology/ai/bard-google-ai-search-updates/ Dan Hendrycks un Kevin Gimpel: nonlinearitātes un stochastic regularizers ar Gaussian kļūdas lineārās vienības. , abs/1606.08415, 2016. gada URL Tātad . Kārlis http://arxiv.org/abs/1606 08415 Dan Hendrycks, Collin Burns, Steven Basart, Andy Zou, Mantas Mazeika, Dawn Song, and Jacob Steinhardt. Measuring massive multitask language understanding. , 2020. arXiv pirmizrāde arXiv:2009.03300 Dan Hendrycks, Collin Burns, Saurav Kadavath, Akul Arora, Steven Basart, Eric Tang, Dawn Song, and Jacob Steinhardt. Measuring mathematical problem solving with the math dataset. , 2021. arXiv preprint arXiv:2103.03874 Jordan Hoffmann, Sebastian Borgeaud, Arthur Mensch, Elena Buchatskaya, Trevor Cai, Eliza Rutherford, Diego de Las Casas, Lisa Anne Hendricks, Johannes Welbl, Aidan Clark, et al. Training compute-optimal large language models. , 2022. arXiv preprint arXiv:2203.15556 Sirui Hong, Xiawu Zheng, Jonathan Chen, Yuheng Cheng, Ceyao Zhang, Zili Wang, Steven Ka Shing Yau, Zijuan Lin, Liyang Zhou, Chenyu Ran, et al. Metagpt: Meta programmēšana vairāku aģentu sadarbības sistēmai. , 2023. arXiv preprint arXiv:2308.00352 Chenxu Hu, Jie Fu, Chenzhuang Du, Simian Luo, Junbo Zhao, and Hang Zhao. Chatdb: Augmenting llms with databases as their symbolic memory. , 2023. arXiv preprint arXiv:2306.03901 Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. , 2021. arXiv preprint arXiv:2106.09685 Hai Hu, Kyle Richardson, Liang Xu, Lu Li, Sandra Ku¨bler, and Lawrence S. Moss. OCNLI: original chinese natural language inference. In Trevor Cohn, Yulan He, and Yang Liu (eds.), , apjoms EMNLP 2020 no , pp. 3512–3526. Association for Computational Linguistics, 2020. doi: 10.18653/v1/2020.findings-emnlp.314. URL . Findings of the Association for Computational Linguistics: EMNLP 2020, Online Event, 16-20 November 2020 ACL konstatējumi https://doi.org/10.18653/v1/2020.findings-emnlp.314 Yuzhen Huang, Yuzhuo Bai, Zhihao Zhu, Junlei Zhang, Jinghan Zhang, Tangjun Su, Junteng Liu, Chuancheng Lv, Yikai Zhang, Jiayi Lei, et al. C-Eval: A multi-level multi-discipline chinese evaluation suite for foundation models. , 2023. arXiv preprint arXiv:2305.08322 Transformatoru aģenti, 2023 URL . https://huggingface.co/docs/ Pārveidotāji / Transformers Baichuan Inc. Baichuan-7B: A large-scale 7B pretraining language model developed by BaiChuan-Inc, 2023a. URL . https://github.com/baichuan-inc/Baichuan-7B XVERSE Technology Inc. XVERSE-13B: A multilingual large language model devel-oped by XVERSE Technology Inc., 2023b. URL . https://github.com/xverse-ai/ XVERSE-13B InternLM Team. InternLM: A multilingual language model with progressively enhanced capabilities, 2023. URL . https://github.com/InternLM/InternLM Shantanu Jain. tiktoken: A fast BPE tokeniser for use with OpenAI’s models, 2022. URL . https://github.com/openai/tiktoken/ Yunjie Ji, Yong Deng, Yan Gong, Yiping Peng, Qiang Niu, Lei Zhang, Baochang Ma, and Xiangang Li. Exploring the impact of instruction data scaling on large language models: An empirical study on real-world use cases. , 2023. arXiv preprint arXiv:2303.14742 Zixuan Jiang, Jiaqi Gu, Hanqing Zhu, and David Z. Pan. Pre-RMSNorm and Pre-CRMSNorm transformers: Equivalent and efficient pre-LN transformers. , abs/2305.14858, 2023. doi: 10.48550/arXiv.2305.14858. URL . Kārlis https://doi.org/10.48550/arXiv.2305.14858 Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. 2014. gadā arXiv preprint arXiv:1412.6980 Tom Kwiatkowski, Jennimaria Palomaki, Olivia Redfield, Michael Collins, Ankur P. Parikh, Chris Alberti, Danielle Epstein, Illia Polosukhin, Jacob Devlin, Kenton Lee, Kristina Toutanova, Llion Jones, Matthew Kelcey, Ming-Wei Chang, Andrew M. Dai, Jakob Uszkoreit, Quoc Le, and Slav Petrov. Natural questions: a benchmark for question answering research. , 7:452–466, 2019. doi: 10.1162/tacl*\* a*\* 00276. URL . . Trans. Assoc. Comput. Linguistics https://doi.org/10 1162/tacl_a_00276 Woosuk Kwon, Zhuohan Li, Siyuan Zhuang, Ying Sheng, Lianmin Zheng, Cody Hao Yu, Joseph E. Gonzalez, Hao Zhang, and Ion Stoica. Efficient memory management for large language model serving with PagedAttention. In 2023. gadā Proceedings of the ACM SIGOPS 29th Symposium on Operating Systems Principles LangChain, Inc. LangChain: Building applications with LLMs through composability, 2023. URL . https://python.langchain.com/ Dmitry Lepikhin, HyoukJoong Lee, Yuanzhong Xu, Dehao Chen, Orhan Firat, Yanping Huang, Maxim Krikun, Noam Shazeer, and Zhifeng Chen. GShard: Scaling giant models with conditional computation and automatic sharding. 2020. gadā arXiv preprint arXiv:2006.16668 Aitor Lewkowycz, Anders Andreassen, David Dohan, Ethan Dyer, Henryk Michalewski, Vinay Ramasesh, Ambrose Slone, Cem Anil, Imanol Schlag, Theo Gutman-Solo, Yuhuai Wu, Behnam Neyshabur, Guy Gur-Ari, and Vedant Misra. Solving quantitative reasoning problems with language models, 2022. Chenliang Li, Hehong Chen, Ming Yan, Weizhou Shen, Haiyang Xu, Zhikai Wu, Zhicheng Zhang, Wenmeng Zhou, Yingda Chen, Chen Cheng, et al. ModelScope-Agent: Building your customizable agent system with open-source large language models. 2023. gadā a. arXiv preprint arXiv:2309.00986 Guohao Li, Hasan Abed Al Kader Hammoud, Hani Itani, Dmitrii Khizbullin, and Bernard Ghanem. Camel: Communicative agents for “mind” exploration of large scale language model society. , 2023b. arXiv preprint arXiv:2303.17760 Haonan Li, Yixuan Zhang, Fajri Koto, Yifei Yang, Hai Zhao, Yeyun Gong, Nan Duan, and Timothy Baldwin. CMMLU: Measuring massive multitask language understanding in Chinese. , 2023c. arXiv preprint arXiv:2306.09212 Raymond Li, Loubna Sean Allal, Yangtian Zi, Niklas Muennighoff, Denis Kocetkov, Chenghao Mou, Marc Marone, Christopher Akiki, Jia Li, Jenny Chim, Qian Liu, Evgenii Zheltonozhskii, Terry Yue Zhuo, Thomas Wang, Lipkin, Olivier Dehaene, Mishig Davaadorj, Joel Lamy-Poirier, Rudra Murthy V, Jason Stiller, Siva Sank Patel, Dmitry Abulkhanov Anderson, Nicolas Gontier, Nicholas Meade, Armel Zebaze, Ming-Ho Yee, Logesh Kumar Umapathi, Jian Zhuo, Jian Zhu Contiley, Benjamin, Muhtamokulov, Zuo Red, Rudra Murthy V, Jason Stiller , abs/2305.06161, 2023d. doi: 10.48550/arXiv.2305.06161. URL . Kārlis https://doi.org/10.48550/arXiv.2305.06161 Yujia Li, David H. Choi, Junyoung Chung, Nate Kushman, Julian Schrittwieser, Re’mi Leblond, Tom Eccles, James Keeling, Felix Gimeno, Agustin Dal Lago, Thomas Hubert, Peter Choy, Cyprien de Masson d’Autume, Igor Babuschkin, Xinyun Chen, Po-Sen Huang, Johannes Welbl, Sven Gowal, Alexey Cherepanov, James Molloy, Daniel J. Mankowitz, Esme Sutherland Robson, Pushmeet Kohli, Nando de Freitas, Koray Kavayukcuoglu un Oriol Vinyals. Konkurences līmeņa koda ģenerācija ar AlphaCode. , abs/2203.07814, 2022. CoRR Hunter Lightman, Vineet Kosaraju, Yura Burda, Harri Edwards, Bowen Baker, Teddy Lee, Jan Leike, John Schulman, Ilya Sutskever, and Karl Cobbe. Let’s verify step by step. , 2023. arXiv pirmizrāde arXiv:2305.20050 Chenxiao Liu and Xiaojun Wan. CodeQA: A question answering dataset for source code com-prehension. In Marie-Francine Moens, Xuanjing Huang, Lucia Specia, and Scott Wen-tau Yih (eds.), , pp. 2618–2632. Associa-tion for Computational Linguistics, 2021. doi: 10.18653/v1/2021.findings-emnlp.223. URL . Aprēķinu lingvistikas asociācijas konstatējumi: EMNLP 2021, Virtuālais notikums / Punta Cana, Dominikānas Republika, 2021. gada 16. un 20. novembris https://doi.org/10.18653/v1/2021.findings-emnlp.223 Haotian Liu, Chunyuan Li, Qingyang Wu un Yong Jae Lee. , 2023a. arXiv preprint arXiv:2304.08485 Xiao Liu, Hanyu Lai, Hao Yu, Yifan Xu, Aohan Zeng, Zhengxiao Du, Peng Zhang, Yuxiao Dong, and Jie Tang. WebGLM: Towards an efficient web-enhanced question answering system with human preferences. , 2023b. arXiv preprint arXiv:2306.07906 Yinhan Liu, Myle Ott, Naman Goyal, Jingfei Du, Mandar Joshi, Danqi Chen, Omer Levy, Mike Lewis, Luke Zettlemoyer, and Veselin Stoyanov. RoBERTa: A robustly optimized BERT pretraining approach. , 2019. arXiv preprint arXiv:1907.11692 Yue Liu, Thanh Le-Cong, Ratnadira Widyasari, Chakkrit Tantithamthavorn, Li Li, Xuan-Bach Dinh Le, and David Lo. Refining ChatGPT-generated code: Characterizing and mitigating code quality issues. , abs/2307.12596, 2023c. doi: 10.48550/arXiv.2307.12596. URL . Kārlis https://doi.org/10.48550/arXiv.2307.12596 Shayne Longpre, Le Hou, Tu Vu, Albert Webson, Hyung Won Chung, Yi Tay, Denny Zhou, Quoc V Le, Barret Zoph, Jason Wei, et al. The Flan collection: Designing data and methods for effective instruction tuning. , 2023. arXiv preprint arXiv:2301.13688 Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. , 2017. arXiv preprint arXiv:1711.05101 Keming Lu, Hongyi Yuan, Zheng Yuan, Runji Lin, Junyang Lin, Chuanqi Tan, Chang Zhou, and Jingren Zhou. #InsTag: Instruction tagging for analyzing supervised fine-tuning of large language models. , abs/2308.07074, 2023. doi: 10.48550/arXiv.2308.07074. URL . . CoRR https://doi org/10.48550/arXiv.2308.07074 Haipeng Luo, Qingfeng Sun, Can Xu, Pu Zhao, Jianguang Lou, Chongyang Tao, Xiubo Geng, Qingwei Lin, Shifeng Chen, and Dongmei Zhang. WizardMath: Empowering mathematical reasoning for large language models via reinforced evol-instruct. , 2023a. arXiv preprint arXiv:2308.09583 Ziyang Luo, Can Xu, Pu Zhao, Qingfeng Sun, Xiubo Geng, Wenxiang Hu, Chongyang Tao, Jing Ma, Qingwei Lin un Daxin Jiang. , 2023b. arXiv pirmizrāde arXiv:2306.08568 Mosaic ML. MPT-30B: Raising the bar for open-source foundation models, 2023. URL . https://www.mosaicml.com/blog/mpt-30b Niklas Muennighoff, Thomas Wang, Lintang Sutawika, Adam Roberts, Stella Biderman, Teven Le Scao, M Saiful Bari, Sheng Shen, Zheng-Xin Yong, Hailey Schoelkopf, et al. Crosslingual generalization through multitask finetuning. , 2022. arXiv preprint arXiv:2211.01786 Niklas Muennighoff, Qian Liu, Armel Zebaze, Qinkai Zheng, Binyuan Hui, Terry Yue Zhuo, Swayam Singh, Xiangru Tang, Leandro von Werra, and Shayne Longpre. OctoPack: Instruction tuning code large language models. , abs/2308.07124, 2023. CoRR Reiichiro Nakano, Jacob Hilton, Suchir Balaji, Jeff Wu, Long Ouyang, Christina Kim, Christopher Hesse, Shantanu Jain, Vineet Kosaraju, William Saunders, et al. WebGPT: Browser-assisted question-answering with human feedback. , 2021. arXiv preprint arXiv:2112.09332 Maxwell Nye, Anders Andreassen, Guy Gur-Ari, Henryk Michalewski, Jacob Austin, David Bieber, David Dohan, Aitor Lewkowycz, Maarten Bosma, David Luan, Charles Sutton, and Augustus Odena. Show your work: Scratchpads for intermediate computation with language models. , abs/2112.00114, 2021. ArXiv OpenAI. Introducing ChatGPT, 2022. URL . https://openai.com/blog/chatgpt OpenAI. ChatML, 2022. URL . https://github.com/openai/openaipython/blob/e389823ba013a24b4c32ce38fa0bd87e6bccae94/chatml.md OpenAI. GPT4 technical report. , 2023. arXiv preprint arXiv:2303.08774 OpenCompass Team. OpenCompass: A universal evaluation platform for foundation models, 2023. URL . https://opencompass.org.cn/leaderboard-llm Long Ouyang, Jeffrey Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schulman, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul F. Christiano, Jan Leike, and Ryan Lowe. Training language models to follow instructions with human feedback. In , 2022. URL . NeurIPS http://papers.nips.cc/paper_files/paper/2022/hash/ b1efde53be364a73914f58805a001731-Abstract-Conference.html Denis Paperno, Germa´n Kruszewski, Angeliki Lazaridou, Quan Ngoc Pham, Raffaella Bernardi, Sandro Pezzelle, Marco Baroni, Gemma Boleda, and Raquel Ferna´ndez. The LAMBADA dataset: Word prediction requiring a broad discourse context. In . The Association for Computer Linguistics, 2016. doi: 10.18653/v1/ p16-1144. URL . Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, ACL 2016, August 7-12, 2016, Berlin, Germany, Volume 1: Long Papers https://doi.org/10.18653/v1/p16-1144 Bowen Peng, Jeffrey Quesnelle, Honglu Fan, and Enrico Shippole. YaRN: Efficient context window extension of large language models. , 2023a. arXiv preprint arXiv:2309.00071 Zhiliang Peng, Wenhui Wang, Li Dong, Yaru Hao, Shaohan Huang, Shuming Ma, and Furu Wei. Kosmos-2: Grounding multimodal large language models to the world. , 2023b. arXiv preprint arXiv:2306.14824 Qwen Team, Alibaba Group. Novērtējuma benchmark par kodu intepreter, 2023a. URL https://github.com/QwenLM/Qwen-Agent/tree/main/benchmark. Qwen Team, Alibaba Group. Evaluation benchmark for tool usage through ReAct prompting, 2023b. URL . https://github.com/QwenLM/Qwen-7B/tree/main/eval Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever, et al. Improving language understanding by generative pre-training. Technical report, OpenAI, 2018. Jack W Rae, Sebastian Borgeaud, Trevor Cai, Katie Millican, Jordan Hoffmann, Francis Song, John Aslanides, Sarah Henderson, Roman Ring, Susannah Young, et al. Scaling language models: Methods, analysis & insights from training gopher. , 2021. arXiv preprint arXiv:2112.11446 Rafael Rafailov, Archit Sharma, Eric Mitchell, Stefano Ermon, Christopher D Manning, and Chelsea Finn. Direct preference optimization: Your language model is secretly a reward model. , 2023. arXiv preprint arXiv:2305.18290 Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J Liu. Exploring the limits of transfer learning with a unified text-to-text transformer. , 21(1):5485–5551, 2020. The Journal of Machine Learning Research Prajit Ramachandran, Barret Zoph un Quoc V Le. Aktivizācijas funkciju meklēšana. , 2017. arXiv pirmizrāde arXiv:1710.05941 Scott E. Reed, Konrad Zolna, Emilio Parisotto, Sergio Go´mez Colmenarejo, Alexander Novikov, Gabriel Barth-Maron, Mai Gimenez, Yury Sulsky, Jackie Kay, Jost Tobias Springenberg, Tom Eccles, Jake Bruce, Ali Razavi, Ashley Edwards, Nicolas Heess, Yutian Chen, Raia Hadsell, Oriol Vinyals, Mahyar Bordbar, and Nando de Freitas. A generalist agent. , 2022, 2022. URL . Trans. Mach. Learn. Res. https://openreview.net/forum?id=1ikK0kHjvj Baptiste Rozie`re, Jonas Gehring, Fabian Gloeckle, Sten Sootla, Itai Gat, Xiaoqing Ellen Tan, Yossi Adi, Jingyu Liu, Tal Remez, Je´re´my Rapin, et al. Code Llama: Open foundation models for code. , 2023. arXiv preprint arXiv:2308.12950 Victor Sanh, Albert Webson, Colin Raffel, Stephen H Bach, Lintang Sutawika, Zaid Alyafeai, Antoine Chaffin, Arnaud Stiegler, Teven Le Scao, Arun Raja un citi. , 2021. arXiv preprint arXiv:2110.08207 Maarten Sap, Hannah Rashkin, Derek Chen, Ronan Le Bras, and Yejin Choi. SocialIQA: Com-monsense reasoning about social interactions. , abs/1904.09728, 2019. URL CoRR http: . //arxiv.org/abs/1904.09728 Teven Le Scao, Angela Fan, Christopher Akiki, Ellie Pavlick, Suzana Ilic’, Daniel Hesslow, Roman Castagne’, Alexandra Sasha Luccioni, Franc’ois Yvon, Matthias Galle’ un citi. , 2022. arXiv preprint arXiv:2211.05100 Timo Schick, Jane Dwivedi-Yu, Roberto Dess`ı, Roberta Raileanu, Maria Lomeli, Luke Zettlemoyer, Nicola Cancedda, and Thomas Scialom. Toolformer: Language models can teach themselves to use tools. , 2023. arXiv pirmizrāde arXiv:2302.04761 John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. Proximal policy optimization algorithms. , 2017. arXiv preprint arXiv:1707.06347 Noam Shazeer. GLU varianti uzlabo transformatoru. 2020. gadā arXiv preprint arXiv:2002.05202 Yongliang Shen, Kaitao Song, Xu Tan, Dongsheng Li, Weiming Lu, and Yueting Zhuang. Hug-gingGPT: Solving AI tasks with ChatGPT and its friends in HuggingFace. , 2023. arXiv preprint arXiv:2303.17580 Mohammad Shoeybi, Mostofa Patwary, Raul Puri, Patrick LeGresley, Jared Casper, and Bryan Catan-zaro. Megatron-LM: Training multi-billion parameter language models using model parallelism. , 2019. arXiv preprint arXiv:1909.08053 Qingyi Si, Tong Wang, Naibin Gu, Rui Liu, and Zheng Lin. Alpaca-CoT: An instruction-tuning platform with unified interface of instruction collection, parameter-efficient methods, and large language models, 2023. URL . https://github.com/PhoebusSi/alpaca-CoT Feifan Song, Bowen Yu, Minghao Li, Haiyang Yu, Fei Huang, Yongbin Li, and Houfeng Wang. Preference ranking optimization for human alignment. , 2023. arXiv preprint arXiv:2306.17492 Stabilitāte AI. StableBeluga2, 2023. URL . https://huggingface.co/stabilityai/ StableBeluga2 Nisan Stiennon, Long Ouyang, Jeffrey Wu, Daniel Ziegler, Ryan Lowe, Chelsea Voss, Alec Radford, Dario Amodei un Paul F Christiano. Mācīties apkopot ar cilvēka atsauksmēm. , 33:3008 līdz 3021, 2020. Advances in Neural Information Processing Systems Jianlin Su. Transformatora uzlabošana: garuma ekstrapolācijas spēja un pozīcijas izturība, 2023a. . https://spaces.ac.cn/archives/9444 Jianlin Su. The magical effect of the Bias term: RoPE + Bias = better length extrapolation, 2023b. URL . https://spaces.ac.cn/archives/9577 Jianlin Su, Yu Lu, Shengfeng Pan, Ahmed Murtadha, Bo Wen, and Yunfeng Liu. Roformer: Enhanced transformer with rotary position embedding. 2021. gadā arXiv preprint arXiv:2104.09864 Tianxiang Sun, Xiaotian Zhang, Zhengfu He, Peng Li, Qinyuan Cheng, Hang Yan, Xiangyang Liu, Yunfan Shao, Qiong Tang, Xingjian Zhao, Ke Chen, Yining Zheng, Zhejian Zhou, Ruixiao Li, Jun Zhan, Yunhua Zhou, Linyang Li, Xiaogui Yang, Lingling Wu, Zhangyue Yin, Xuanjing Huang un Xipeng Qiu. Zhiqing Sun, Yikang Shen, Qinhong Zhou, Hongxin Zhang, Zhenfang Chen, David Cox, Yiming Yang, and Chuang Gan. Principle-driven self-alignment of language models from scratch with minimal human supervision. , 2023b. arXiv preprint arXiv:2305.03047 Mirac Suzgun, Nathan Scales, Nathanael Scha¨rli, Sebastian Gehrmann, Yi Tay, Hyung Won Chung, Aakanksha Chowdhery, Quoc V Le, Ed H Chi, Denny Zhou, et al. Challenging big-bench tasks and whether chain-of-thought can solve them. , 2022. arXiv pirmizrāde arXiv:2210.09261 Marc Szafraniec, Baptiste Rozie`re, Hugh Leather, Patrick Labatut, Franc-ois Charton un Gabriel Synnaeve. Koda tulkošana ar kompilatoru pārstāvniecībām. . OpenReview.net, 2023. URL . 11. starptautiskā konference par mācīšanās pārstāvniecību, ICLR 2023, Kigali, Ruanda, 2023. gada 1. un 5. maijā https://openreview.net/pdf?id=XomEU3eNeSQ Alon Talmor, Jonathan Herzig, Nicholas Lourie, and Jonathan Berant. CommonsenseQA: A question answering challenge targeting commonsense knowledge. In Jill Burstein, Christy Doran, and Thamar Solorio (eds.), , pp. 4149–4158. Association for Computational Linguistics, 2019. doi: 10.18653/v1/n19-1421. URL . Ziemeļamerikas Asociācijas aprēķinu lingvistikas nodaļas 2019. gada konferences protokoli: Cilvēku valodas tehnoloģijas, NAACL-HLT 2019, Minneapolis, MN, ASV, 2019. gada 2.-7. jūnijs, 1. apjoms (Ilgi un īsi dokumenti) https://doi.org/10.18653/v1/n19-1421 Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B. Hashimoto. Stanford Alpaca: An instruction-following LLaMA model, 2023. URL . https://github.com/tatsu-lab/stanford_alpaca Ross Taylor, Marcin Kardas, Guillem Cucurull, Thomas Scialom, Anthony Hartshorn, Elvis Saravia, Andrew Poulton, Viktor Kerkez, and Robert Stojnic. Galactica: A large language model for science, 2022. Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, YaGuang Li, Hongrae Lee, Huaixiu Steven Zheng, Amin Ghafouri, Marcelo Menegali, Yanping Huang, Maxim Krikun, Dmitry Lepikhin, James Qin, Dehao Chen, Yuanzhong Xu, Zhifeng Chen, Adam Roberts, Maarten Bosma, Yanqi Zhou, Chung-Ching Chang, Igor Krivokon, Will Rusch, Marc Pickett, Kathleen S. Meier-Hellstern, Meredith Ringel Morris, Tulsee Doshi, Renelito Delos Santos, Toju Duke, Johnny Soraker, Ben Zevenbergen, Vinodkumar Prabhakaran, Mark Diaz, Ben Hutchinson, Kristen Ol-son, Alejandra Molina, Erin Hoffman-John, Josh Lee, Lora Aroyo, Ravi Rajakumar, Alena Butryna, Matthew Lamm, Viktoriya Kuzmina, Joe Fenton, Aaron Cohen, Rachel Bernstein, Ray Kurzweil, Blaise Agu¨era y Arcas, Claire Cui, Marian Croak, Ed H. Chi, and Quoc Le. LaMDA: Language models for dialog applications. , abs/2201.08239, 2022. URL . CoRR https://arxiv.org/abs/2201.08239 Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothe´e Lacroix, Baptiste Rozie`re, Naman Goyal, Eric Hambro, Faisal Azhar, et al. LLaMA: Open and efficient foundation language models. 2023. gadā a. arXiv preprint arXiv:2302.13971 Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, Yasmine Babaei, Nikolay Bashlykov, Soumya Batra, Prajjwal Bhargava, Shruti Bhosale, Dan Bikel, Lukas Blecher, Cristian Canton-Ferrer, Moya Chen, Guillem Kerkez, Madian Khabsa, Isabel Kloumann, Jude Fernandes, Jeremy Fu, Wenyin Fu, Brian Fuller, Cynthia Gao, Vedanuj Goswami, Naman Goyal, Anthony Hartshorn, Saghar Hosseini, Rui Hou, Hakan Inan, Marcin Kardas, Viktor Kerkez, Madian Khabsa, Isabel Kloumann, Artemadur Korenev, Punit Singh Koura, Marie-Anneaux Lach, Thibaut Lavril, Robert , abs/2307.09288, 2023b. doi: 10.48550/arXiv.2307.09288. URL . Kārlis https://doi.org/ 10.48550/arXiv.2307.09288 Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. , 30, 2017. Neironu informācijas apstrādes sistēmu attīstība Guanzhi Wang, Yuqi Xie, Yunfan Jiang, Ajay Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi Fan un Anima Anandkumar. 2023. gadā a. arXiv preprint arXiv:2305.16291 Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Huai hsin Chi, and Denny Zhou. Self-consistency improves chain of thought reasoning in language models. , abs/2203.11171, 2022. Arhīvs Yan Wang, Xiaojiang Liu un Shuming Shi. dziļa neironu risinātājs matemātikas vārdu problēmām. , 2017. URL . semanticscholar.org/CorpusID:910689. Conference on Empirical Methods in Natural Language Processing https://api Yizhong Wang, Hamish Ivison, Pradeep Dasigi, Jack Hessel, Tushar Khot, Khyathi Raghavi Chandu, David Wadden, Kelsey MacMillan, Noah A Smith, Iz Beltagy, et al. How far can camels go? Exploring the state of instruction tuning on open resources. 2023. gadā B arXiv preprint arXiv:2306.04751 Yizhong Wang, Yeganeh Kordi, Swaroop Mishra, Alisa Liu, Noah A. Smith, Daniel Khashabi, and Hannaneh Hajishirzi. Self-Instruct: Aligning language models with self-generated instructions. In Anna Rogers, Jordan L. Boyd-Graber, and Naoaki Okazaki (eds.), , 13484–13508. lpp. Asociācija skaitļošanas lingvistikai, 2023c. doi: 10.18653/v1/2023.acl-long.754. URL . Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), ACL 2023, Toronto, Canada, July 9-14, 2023 https://doi.org/10.18653/v1/ 2023.acl-long.754 Yue Wang, Weishi Wang, Shafiq Joty, and Steven CH Hoi. CodeT5: Identifier-aware unified pre-trained encoder-decoder models for code understanding and generation. , 2021. arXiv preprint arXiv:2109.00859 Yue Wang, Hung Le, Akhilesh Deepak Gotmare, Nghi D. Q. Bui, Junnan Li, and Steven C. H. Hoi. CodeT5+: Open code large language models for code understanding and generation. , abs/2305.07922, 2023d. doi: 10.48550/arXiv.2305.07922. URL . . Kārlis https://doi.org/10 48550/arXiv.2305.07922 Jason Wei, Maarten Bosma, Vincent Y. Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M. Dai un Quoc V. Le. Finetuned valodu modeļi ir nulles šāvienu mācītāji. OpenReview.net, 2022a. URL . Desmitā starptautiskā konference par mācīšanās pārstāvniecību, ICLR 2022, Virtuālais notikums, 2022. gada 25.-29. aprīlis https://openreview.net/forum?id= gEZrGCozdqR Jason Wei, Yi Tay, Rishi Bommasani, Colin Raffel, Barret Zoph, Sebastian Borgeaud, Dani Yogatama, Maarten Bosma, Denny Zhou, Donald Metzler, Ed Huai hsin Chi, Tatsunori Hashimoto, Oriol Vinyals, Percy Liang, Jeff Dean un William Fedus. , 2022, 2022b. URL CorpusID:249674500. Trans. Mach. Learn. Res. https://api.semanticscholar.org/ Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed Chi, Quoc V Le, Denny Zhou un citi. , 35:24824–24837, 2022c. Advances in Neural Information Processing Systems Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Re'mi Louf, Morgan Funtowicz un citi. , 2019. arXiv preprint arXiv:1910.03771 Benfeng Xu, An Yang, Junyang Lin, Quan Wang, Chang Zhou, Yongdong Zhang un Zhendong Mao. 2023. gadā a. arXiv preprint arXiv:2305.14688 Can Xu, Qingfeng Sun, Kai Zheng, Xiubo Geng, Pu Zhao, Jiazhan Feng, Chongyang Tao un Daxin Jiang. , 2023b. arXiv preprint arXiv:2304.12244 Canwen Xu, Daya Guo, Nan Duan un Julian McAuley. Baize: atvērtā koda tērzēšanas modelis ar parametru efektīvu pielāgošanu paša tērzēšanas datiem. 2023 c. arXiv preprint arXiv:2304.01196 Yuzhuang Xu, Shuo Wang, Peng Li, Fuwen Luo, Xiaolong Wang, Weidong Liu, and Yang Liu. Exploring large language models for communication games: An empirical study on werewolf. , 2023d. arXiv preprint arXiv:2309.04658 Aiyuan Yang, Bin Xiao, Bingning Wang, Borong Zhao, Chao Yin, Chenxu Lv, Da Pan, Dian Wang, Dong Yan, Fan Yang, Fei Deng, Feng Wang, Feng Liu, Guangwei Ai, Guosheng Dong, Haizhou Zhao, Hang Xu, Haoze Sun, Hongda Zhang, Hui Liu, Jiaming Ji, Jian Xie, Juntao Dai, Kun Fang, Lei Su, Liang Song, Lifeng Liu, Liyun Ru, Luyao Ma, Mang Wang, Mickel Liu Ming, An Lin, Nuolan Nie, Peidong Guo, Ruiyang Sun, Tao Zhang, Tianpeng Liu, Tianyu Liu, Wei Chip, Weeng Chen, Xiangrong Zeng, Xiaochuan Wang, . . https://cdn.baichuan-ai.com/paper/Baichuan2-technical-report pdf Jianxin Yang. ugunsdzēsēji 2023. gadā https://github.com/yangjianxin1/Firefly Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan un Yuan Cao. , 2022. arXiv preprint arXiv:2210.03629 Qinghao Ye, Haiyang Xu, Guohai Xu, Jiabo Ye, Ming Yan, Yiyang Zhou, Junyang Wang, Anwen Hu, Pengcheng Shi, Yaya Shi, et al. mPLUG-Owl: Modularization empowers large language models with multimodality. , 2023. arXiv preprint arXiv:2304.14178 Longhui Yu, Weisen Jiang, Han Shi, Jincheng Yu, Zhengying Liu, Yu Zhang, James T. Kwok, Zhenguo Li, Adrian Weller, and Weiyang Liu. Metamath: Bootstrap your own mathematical questions for large language models, 2023. Zheng Yuan, Hongyi Yuan, Chengpeng Li, Guanting Dong, Keming Lu, Chuanqi Tan, Chang Zhou un Jingren Zhou. Zheng Yuan, Hongyi Yuan, Chuanqi Tan, Wei Wang, and Songfang Huang. How well do large language models perform in arithmetic tasks? , 2023b. arXiv preprint arXiv:2304.02015 Zheng Yuan, Hongyi Yuan, Chuanqi Tan, Wei Wang, Songfang Huang, and Fei Huang. RRHF: Rank responses to align language models with human feedback without tears, 2023c. Xiang Yue, Xingwei Qu, Ge Zhang, Yao Fu, Wenhao Huang, Huan Sun, Yu Su, and Wenhu Chen. MAmmoTH: Building math generalist models through hybrid instruction tuning. , 2023. arXiv preprint arXiv:2309.05653 Rowan Zellers, Ari Holtzman, Yonatan Bisk, Ali Farhadi, and Yejin Choi. HellaSwag: Can a machine really finish your sentence? In Anna Korhonen, David R. Traum, and Llu´ıs Ma`rquez (eds.), , 4791–4800. asociācija skaitļošanas lingvistikai, 2019. doi: 10.18653/v1/p19-1472. URL . Asociācijas skaitļošanas lingvistikas 57. konferences protokoli, ACL 2019, Florence, Itālija, 2019. gada 28. jūlijs, 2. augusts, 1: garie dokumenti https://doi.org/10.18653/v1/p19-1472 Aohan Zeng, Xiao Liu, Zhengxiao Du, Zihan Wang, Hanyu Lai, Ming Ding, Zhuoyi Yang, Yifan Xu, Wendi Zheng, Xiao Xia, et al. GLM-130B: An open bilingual pre-trained model. , 2022. arXiv preprint arXiv:2210.02414 Fengji Zhang, Bei Chen, Yue Zhang, Jin Liu, Daoguang Zan, Yi Mao, Jian-Guang Lou un Weizhu Chen. RepoCoder: Repozitorijas līmeņa koda pabeigšana, izmantojot iteratīvu meklēšanu un ģenerēšanu. , abs/2303.12570, 2023a. doi: 10.48550/arXiv.2303.12570. URL . CoRR https://doi.org/ 10.48550 / arXiv.2303.12570 Susan Zhang, Stephen Roller, Naman Goyal, Mikel Artetxe, Moya Chen, Shuohui Chen, Christopher Dewan, Mona Diab, Xian Li, Xi Victoria Lin, et al. OPT: Atvērti iepriekš apmācīti transformatoru valodas modeļi. , 2022. arXiv pirmizrāde arXiv:2205.01068 Xiaotian Zhang, Chunyang Li, Yi Zong, Zhengyu Ying, Liang He, and Xipeng Qiu. Evaluating the performance of large language models on GAOKAO benchmark. , abs/2305.12474, 2023b. doi: 10.48550/arXiv.2305.12474. URL . . CoRR https://doi.org/10.48550/arXiv 2305.12474 Qinkai Zheng, Xiao Xia, Xu Zou, Yuxiao Dong, Shan Wang, Yufei Xue, Zihan Wang, Lei Shen, Andi Wang, Yang Li, Teng Su, Zhilin Yang un Jie Tang. , abs/2303.17568, 2023. doi: 10.48550/arXiv.2303.17568. URL . CoRR https://doi.org/10.48550/arXiv.2303.17568 Wanjun Zhong, Ruixiang Cui, Yiduo Guo, Yaobo Liang, Shuai Lu, Yanlin Wang, Amin Saied, Weizhu Chen, and Nan Duan. AGIEval: A human-centric benchmark for evaluating foundation models. , abs/2304.06364, 2023a. doi: 10.48550/arXiv.2304.06364. URL . CoRR https://doi.org/ 10.48550/arXiv.2304.06364 Wanjun Zhong, Lianghong Guo, Qiqi Gao, and Yanlin Wang. MemoryBank: Enhancing large language models with long-term memory. 2023. gadā B arXiv preprint arXiv:2305.10250 Denny Zhou, Nathanael Scharli, Le Hou, Jason Wei, Nathan Scales, Xuezhi Wang, Dale Schuurmans, Olivier Bousquet, Quoc Le, and Ed Huai hsin Chi. Least-to-most prompting enables complex reasoning in large language models. , abs/2205.10625, 2022. Arhīvs A pielikums A.1 More Training Details A.1.1 Datu formāts Qwen-Chat Different from conventional pretraining based on autoregressive next-token prediction, despite using a similar training task, there should be a specially design data format for SFT and RLHF to build a conversational AI assistant model. Common formats include “human-assistant” and ChatML formats. As to our knowledge, one of the earliest examples of the human-assistant format comes from Anthropic which adds a special phrase “\n\nhuman: ” in front of the user input and “\n\nassistant: ” in front of the assistant response. It is easy for the base language model to transfer to the pattern of conversational AI. However, as the specific phrases are common words, it might be hard for the model to disambiguate from these words in other contexts. (Bai et al., 2022. gadā b) Tā vietā mēs pievērsāmies OpenAI ierosinātajam ChatML formātam. This format allows the use of special tokens, i.e., “ ” and “ ”, kas neparādās priekšapmācībā un tādējādi atrisina iepriekš minēto problēmu. 5 A.2 Evaluation A.2.1 Automātiskā novērtēšana To provide a whole picture of the performance of our model series QWEN, here in this section we illustrate the detailed performance of our models as well as the baselines in the comprehensive benchmark evaluation proposed by We report the results in multiple tables based on the officially provided categories, including examination, language, knowledge, understanding, and reasoning. In terms of the performance of the baseline models, we report the higher results between the reported ones and those on the leaderboard. OpenCompass komandas Tātad 2023 Šeit mēs novērtējam modeļus uz virkni datu kopu, kas ir svarīgi eksāmenam.Datu kopas ietver: Examination MMLU (Hendrycks et al., 2020) Masveida vairāku uzdevumu valodas izpratne ir paredzēta valodas izpratnes spēju mērīšanai. C-Eval (Huang et al., 2023) C-Eval ir ķīniešu novērtēšanas datu kopums, kas aptver 52 dažādas disciplīnas. CMMLU is designed for assessing language understanding capabilities in Chinese. We report 5-shot results. CMMLU (Li et al., 2023c) AGIEval (Zhong et al., 2023a) Tas ir kritērijs, kas sastāv no cilvēka centrā eksāmeniem, ieskaitot koledžas uzņemšanas eksāmenus, tiesību skolas uzņemšanas testus, matemātikas sacensības un advokāta kvalifikācijas testus. Gaokao-Bench (Zhang et al., 2023b) Šis ir kritērijs ar Gaokao (ķīniešu koledžas uzņemšanas eksāmenu) jautājumiem. ARC is a dataset consisting of grade-school level, multiple-choice science questions. It includes an easy set and a challenge set, which are referred by ARC-e and ARC-c. We report zero-shot results. ARC (Clark et al., 2018) Runājot par MMLU, mēs ziņojam par detalizētiem rezultātiem tabulā Attiecībā uz C-Eval mēs ziņojam rezultātus tabulā Par pārējām datu kopām mēs ziņojam rezultātus tabulā Ņemiet vērā, ka AGIEval ietver ķīniešu un angļu valodas daļas, bet LLAMA 2 ziņoja tikai par rezultātiem angļu valodā, tāpēc mēs izmantojam OpenCompass rezultātus. 13. 14. 15. Additionally, while CMMLU, AGIEval, and Gaokao-Bench are related to Chinese, and MPT, Falcon, and the LLaMA series were not optimized for Chinese, these models achieved low performance on the datasets. Šeit mēs novērtējam modeļus uz virkni datu kopu, kas attiecas uz zināšanām un dabiskās valodas izpratni. Knowledge and Understanding BoolQ (Clark et al., 2019) Tas ir QA datu kopums, kur jautājumi ir par Wikipedia sadaļām, un modelim vajadzētu atbildēt jā vai nē uz sniegto iespējamo atbildi. CommonsenseQA (Talmor et al., 2019) Šis ir datu kopums, kas atbild uz vairāku izvēles jautājumu, kas novērtē izpratni par saprāta zināšanām. NaturalQuestions (Kwiatkowski et al., 2019) Tas ir QA datu kopums, kurā jautājumi ir no lietotājiem un atbildes pārbauda eksperti. LAMBADA (Paperno et al., 2016) Tas ir datu kopums, lai novērtētu valodas izpratni, izmantojot vārdu prognozi. Tas sastāv no daļām, kas saistītas ar cilvēka subjektiem. Mēs ziņojam par nulles šāvienu rezultātiem. Mēs ziņojam par rezultātiem tabulā 16. Mēs ziņojam par novērtējuma rezultātiem uz datu kopām, kas attiecas uz pamatojumu, koncentrējoties uz dabiskās valodas pamatojumu. Citiem, piemēram, matemātikā un kodēšanā, kā mēs esam ilustrējuši detalizētus rezultātus, šeit mēs nerunājam par šiem rezultātiem atkārtoti. Reasoning HellaSwag (Zellers et al., 2019) Tas ir dabiskās valodas izpratnes (NLI) datu kopums, kur jautājumi ir viegli cilvēkiem, bet cīnās par iepriekšējiem valodas modeļiem. PIQA (Bisk et al., 2020) Šis ir NLI datu kopums, kas novērtē fiziskās zināšanas. SIQA (Sap et al., 2019) Šis ir NLI datu kopums, kas novērtē sociālo saprāta izlūkošanu. OCNLI (Hu et al., 2020) Šis ir NLI datu kopums, kas koncentrējas uz ķīniešu valodu. We report the results in Table 17. A.2.2 Cilvēka novērtējums In this section, we demonstrate the cases of human analysis. In our self-constructed evaluation dataset, the instructions are either manually written data or manual revised from public datasets, such as CLiB C-Evāls faktu LeetCode 3) un tā tālāk 6 (Huang et al., 2023 gadā) (Krievijas un citām valstīm) 2023 b) un 7 Attiecībā uz katru gadījumu mēs demonstrējam atbildes un Elo vērtējumus of all models for comparison. Specifically, as the data in our human evaluation are in Chinese, we also provide their translations in English. 8 A.3 Koda tulkotāja analīze Šeit mēs sniedzam salīdzinājuma gadījumu starp CODE LLAMA un QWEN-CHAT. Šis gadījums demonstrē QWEN-CHAT priekšrocības tabulu datu apstrādē un sarežģītu uzdevumu veikšanā. Šis dokuments ir pieejams arhīvā saskaņā ar CC by 4.0 Deed (attribution 4.0 International) licenci. Šis dokuments ir Attiecības ir licencētas saskaņā ar CC by 4.0 Deed (Attribution 4.0 International) licenci. available on arxiv