|
--- |
|
license: apache-2.0 |
|
language: |
|
- en |
|
pipeline_tag: text-generation |
|
library_name: transformers |
|
tags: |
|
- nlp |
|
- llm |
|
--- |
|
# K2: a Fully Reproducible Large Language Model Outperforming Llama 2 70B using 35% less compute |
|
|
|
LLM360 demystifies the training recipe used for Llama 2 - 70B with K2. |
|
|
|
About K2: |
|
* 65 billion parameter LLM |
|
* Tokens: 1.4T |
|
* Languages: English |
|
* Models Released: base, chat model |
|
* Trained in 2 stages |
|
* License: Apache 2.0 |
|
|
|
K2 was developed as a collaboration between [MBZUAI](https://mbzuai.ac.ae/institute-of-foundation-models/), [Petuum](petuum.com), and [LLM360](llm360.ai). |
|
|
|
## LLM360 Model Performance and Evaluation Collection |
|
|
|
The LLM360 Performance and Evaluation Collection is a robust evaluations set consisting of general and domain specific evaluations to assess model knowledge and function. Evaluations include standard best practice benchmarks, medical, math, and coding knowledge. More about the evaluations can be found [here](llm360.ai/evaluations). |
|
|
|
put the general table in here |
|
|
|
Detailed analysis can be found on the K2 Weights and Biases project [here](wandb.ai) |
|
|
|
|
|
## Datasets and Mix |
|
|
|
The following data mix was used to train K2 and achieve results in line with Llama 2 70B. |
|
|
|
The full data sequence can be found [here](https://huggingface.co/datasets/LLM360/K2Datasets/tree/main) |
|
|
|
| Dataset | Starting Tokens | Multiplier | Total Tokens |% of Total | |
|
| ----------- | ----------- | ----------- | ----------- | ----------- | |
|
| dm-math | 4.33B | 3x | 13B | 1% | |
|
| pubmed-abstracts | 4.77B | 3x | 14.3B | 1.1% | |
|
| uspto | 4.77B | 3x | 14.3B | 1.1% | |
|
| pubmed-central | 26B | 1x | 26B | 2% | |
|
| [redpajama.arxiv](https://huggingface.co/datasets/cerebras/SlimPajama-627B) | 27.3B | 1x | 27.3B | 2.1% | |
|
| [starcoder.spm](https://huggingface.co/datasets/bigcode/starcoderdata) | 67.6B | 0.5x | 33.8B | 2.6% | |
|
| [starcoder.fim](https://huggingface.co/datasets/bigcode/starcoderdata) | 67.6B | 0.5x | 33.8B | 2.6% | |
|
| [redpajama.stackexchange](https://huggingface.co/datasets/cerebras/SlimPajama-627B) | 61.1B | 1x | 61.1B | 4.7% | |
|
| [starcoder](https://huggingface.co/datasets/bigcode/starcoderdata) | 132.6B | 0.5x | 66.3B | 5.1% | |
|
| [pile-of-law](https://huggingface.co/datasets/pile-of-law/pile-of-law) | 76.7B | 1x | 76.7B | 5.9% | |
|
| [redpajama.book](https://huggingface.co/datasets/cerebras/SlimPajama-627B) | 80.6B | 1x | 80.6B | 6.2% | |
|
| s2orc | 107.9B | 1x | 107.9B | 8.3% | |
|
| [redpajama.wikipedia](https://huggingface.co/datasets/cerebras/SlimPajama-627B) | 22.1B | 6x | 132.6B | 10.2% | |
|
| [refinedweb](https://huggingface.co/datasets/tiiuae/falcon-refinedweb) | 612.3B | 1x | 612.3B | 47.1% | |
|
| Totals | - | - | 1.3T | 100% | |
|
|
|
|
|
# LLM360 Reasearch Suite |
|
|
|
## Stage 2 - Last 10 Checkpoints |
|
| Checkpoints | | |
|
| ----------- | ----------- | |
|
| [Checkpoint 380](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_380) | [Checkpoint 375](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_375) | |
|
| [Checkpoint 379](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_379) | [Checkpoint 374](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_374) | |
|
| [Checkpoint 378](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_378) | [Checkpoint 373](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_373) | |
|
| [Checkpoint 377](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_377) | [Checkpoint 372](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_372) | |
|
| [Checkpoint 376](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_376) | [Checkpoint 371](https://huggingface.co/LLM360/K2/tree/ministage2_ckpt_371) | |
|
|
|
## Stage 1 - Last 10 Checkpoints |
|
| Checkpoints | | |
|
| ----------- | ----------- | |
|
| [Checkpoint 360](https://huggingface.co/LLM360/K2/tree/ckpt_360) | [Checkpoint 355](https://huggingface.co/LLM360/K2/tree/ckpt_355) | |
|
| [Checkpoint 359](https://huggingface.co/LLM360/K2/tree/ckpt_359) | [Checkpoint 354](https://huggingface.co/LLM360/K2/tree/ckpt_354) | |
|
| [Checkpoint 358](https://huggingface.co/LLM360/K2/tree/ckpt_358) | [Checkpoint 353](https://huggingface.co/LLM360/K2/tree/ckpt_353) | |
|
| [Checkpoint 357](https://huggingface.co/LLM360/K2/tree/ckpt_357) | [Checkpoint 352](https://huggingface.co/LLM360/K2/tree/ckpt_352) | |
|
| [Checkpoint 356](https://huggingface.co/LLM360/K2/tree/ckpt_356) | [Checkpoint 351](https://huggingface.co/LLM360/K2/tree/ckpt_351) | |
|
|
|
[to find all branches: git branch -a] |
|
|
|
## LLM360 Pretraining Suite |
|
We provide step-by-step reproducation tutorials for tech enthusiasts, AI practitioners and academic or industry researchers who want to learn pretraining techniques here [llm360.ai/pretraining] |
|
|
|
## LLM360 Developer Suite |
|
We provide step-by-step finetuning tutorials for tech enthusiasts, AI practitioners and academic or industry researchers here [llm360.ai/pretraining]. |
|
|
|
#Put inference section here. |
|
|
|
|
|
## Model Description |
|
|
|
- **Model type:** 65 billion parameter language model with the same architecture as LLaMA. |
|
- **Language(s) (NLP):** English |
|
- **License:** Apache 2.0 |
|
- **Resources for more information:** |
|
- Training Code: TBD |
|
- Data Preparation: TBD |
|
- Metrics: TBD |
|
- Fully processed K2 pretraining dataset: TBD |
|
|
|
|
|
## About LLM360 |
|
LLM360 is an initiative for comprehensive and fully open-sourced LLMs, |
|
where all training details, model checkpoints, intermediate results, and |
|
additional analyses are made available to the community. Our goal is to advance |
|
the field by inviting the community to deepen the understanding of LLMs |
|
together. As the first step of the project LLM360, we release all intermediate |
|
model checkpoints, our fully-prepared pre-training dataset, all source code and |
|
configurations, and training details. We are |
|
committed to continually pushing the boundaries of LLMs through this open-source |
|
effort. |
|
|
|
[Visit us](https://www.llm360.ai/) |
|
|
