paint-brush
Griffin Models: Outperforming Transformers with Scalable AI Innovationby@gating

Griffin Models: Outperforming Transformers with Scalable AI Innovation

by Gating Technology
Gating Technology HackerNoon profile picture

Gating Technology

@gating

We cover HOW gating is used in technology adoption. #Meritocracy...

January 13th, 2025
Read on Terminal Reader
Read this story in a terminal
Print this story
Read this story w/o Javascript
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

This research shows Griffin models outperform Transformers in validation loss and scaling efficiency, following Chinchilla scaling laws up to 14B parameters.
featured image - Griffin Models: Outperforming Transformers with Scalable AI Innovation
1x
Read by Dr. One voice-avatar

Listen to this story

Gating Technology HackerNoon profile picture
Gating Technology

Gating Technology

@gating

We cover HOW gating is used in technology adoption. #Meritocracy #Gating #TechnologyResearch

About @gating
LEARN MORE ABOUT @GATING'S
EXPERTISE AND PLACE ON THE INTERNET.
0-item

STORY’S CREDIBILITY

Academic Research Paper

Academic Research Paper

Part of HackerNoon's growing list of open-source research papers, promoting free access to academic material.

Authors:

(1) Soham De, Google DeepMind and with Equal contributions;

(2) Samuel L. Smith, Google DeepMind and with Equal contributions;

(3) Anushan Fernando, Google DeepMind and with Equal contributions;

(4) Aleksandar Botev, Google DeepMind and with Equal contributions;

(5) George Cristian-Muraru, Google DeepMind and with Equal contributions;

(6) Albert Gu, Work done while at Google DeepMind;

(7) Ruba Haroun, Google DeepMind;

(8) Leonard Berrada, Google DeepMind;

(9) Yutian Chen, Google DeepMind;

(10) Srivatsan Srinivasan, Google DeepMind;

(11) Guillaume Desjardins, Google DeepMind;

(12) Arnaud Doucet, Google DeepMind;

(13) David Budden, Google DeepMind;

(14) Yee Whye Teh, Google DeepMind;

(15) David Budden, Google DeepMind;

(16) Razvan Pascanu, Google DeepMind;

(17) Nando De Freitas, Google DeepMind;

(18) Caglar Gulcehre, Google DeepMind.

1 Introduction

2 Model Architecture

3 Recurrent Models Scale as Efficiently as Transformers

3.1. Scaling curves

3.2. Evaluation on downstream tasks

4 Training Recurrent Models Efficiently on Device and 4.1. Model parallelism for large scale training

4.2. Efficient linear recurrences on device

4.3. Training speed on longer sequences

5. Inference Speed

5.1. A simple model of the decode step

5.2. Results

6. Long Context Modeling and 6.1. Improving next token prediction with longer contexts

6.2. Copy and retrieval capabilities

7. Related Works

8. Conclusion, Acknowledgements, and References


A. RG-LRU Recurrence Gate

B. Complex-Gated Linear Recurrent Unit (CG-LRU)

C. Model Scale Hyper-Parameters

D. Efficient Linear Recurrences on Device

E. The Local Attention Window Size of Griffin

F. Inference Speeds

G. Improving Next Token Prediction with Longer Contexts: Additional Results

H. Additional Details of the Copy and Retrieval Tasks

3.1. Scaling curves

We present our main scaling results in Figure 1(a). All three model families are trained at a range of model scales from 100M to 7B parameters, with an additional Griffin model with 14 billion parameters. We increase the number of training tokens to be roughly proportional to the number of parameters of the model, as prescribed by the Chinchilla scaling laws (Hoffmann et al., 2022). Models are trained on the MassiveText dataset (Hoffmann et al., 2022), previously used to train Gopher (Rae et al., 2021) and Chinchilla (Hoffmann et al., 2022), although we use a slightly different data subset distribution. A sequence length of 2048 tokens was used (see Section 6 for results with longer sequences.) All experiments use the AdamW optimizer (Loshchilov and Hutter, 2017). We tune the learning rate, weight decay and 𝛽2 parameters for small models, and use these runs to identify scaling rules for these hyper-parameters which predict their optimal values for the 7B and 14B models.


All three model families demonstrate a linear scaling relationship between the validation loss and training FLOPs (see Figure 1(a); note both axes are in log scale), as previously observed for Transformers by Brown et al. (2020). Notably, Griffin achieves lower validation loss than the Transformer baseline across all FLOPs budgets despite not using any global attention layers. Hawk on the other hand achieves slightly higher validation loss, but this gap appears to close as the training budget increases.


Table 1 | Character normalized accuracy. Hawk is competitive with our Transformer baseline, and exceeds the reported performance of Mamba despite being trained on half as many tokens. Griffin outperforms our Transformer baseline, and matches the performance of Llama-2 despite being trained on roughly 7 times fewer tokens. We report unnormalized accuracy with partial scoring for WinoGrande.

Table 1 | Character normalized accuracy. Hawk is competitive with our Transformer baseline, and exceeds the reported performance of Mamba despite being trained on half as many tokens. Griffin outperforms our Transformer baseline, and matches the performance of Llama-2 despite being trained on roughly 7 times fewer tokens. We report unnormalized accuracy with partial scoring for WinoGrande.


This paper is available on arxiv under CC BY 4.0 DEED license.


L O A D I N G
. . . comments & more!

About Author

Gating Technology HackerNoon profile picture
Gating Technology@gating
We cover HOW gating is used in technology adoption. #Meritocracy #Gating #TechnologyResearch

TOPICS

THIS ARTICLE WAS FEATURED IN...

Permanent on Arweave
Read on Terminal Reader
Read this story in a terminal
 Terminal
Read this story w/o Javascript
Read this story w/o Javascript
 Lite
Hackernoon
Threads
Bsky