Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update README.md #274

Closed
wants to merge 1 commit into from
Closed
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
287 changes: 12 additions & 275 deletions README.md
Original file line number Diff line number Diff line change
@@ -1,283 +1,20 @@
# OpenLM
# Linearizing Large Language Models

![](/plots/logo.png)
This repository contains the code for [Linearizing Large Language Models](https://arxiv.org/abs/2405.06640).
This is a fork of the [OpenLM](https://github.com/mlfoundations/open_lm) repository. Please refer to the original repository for set up details.

OpenLM is a minimal but performative language modeling (LM) repository, aimed to facilitate research on medium sized LMs. We have verified the performance of OpenLM up to 7B parameters and 256 GPUs.
In contrast with other repositories such as Megatron, we depend only on PyTorch, XFormers, or Triton for our core modeling code.
[[Pre-print](https://arxiv.org/abs/2405.06640)]

# Contents
- [Release Notes](#release-notes)
- [Quickstart](#quickstart)
- [Setup](#setup)
- [Process training data](#process-training-data)
- [Run training](#run-training)
- [Evaluate Model](#evaluate-model)
- [Generate Text](#generate-text)
- [Pretrained Models](#pretrained-models)
- [Team and Acknowledgements](#team-and-acknowledgements)

# Release Notes
- 09/26/23: Public release and featured on [Laion Blog](https://laion.ai/blog/open-lm/)
- 08/18/23: Updated README.md
# Quickstart
Here we'll go over a basic example where we start from a fresh install, download and preprocess some training data, and train a model.

## Setup
We require python >=3.9, and a current installation of pyTorch, as well as several other packages. The full list of requirements is contained in `requirements.txt` and can be installed in your python enviornment via
```>>> pip install -r requirements.txt```
Next, to access `open_lm` everywhere in your virtual environment, install it using pip (from within the top level github repo)
```>>> pip install --editable . ```
Some considerations:
- We like [WandB](https://wandb.ai/) and [tensorboard](https://www.tensorflow.org/tensorboard) for logging. We specify how to use these during training below.

## Process Training Data
Next you must specify a collection of tokenized data. For the purposes of this example, we will use a recent dump of english Wikipedia, available on HuggingFace. To download this locally, we've included a script located at [open_lm/datapreprocess/wiki_download.py](open_lm/datapreprocess/wiki_download.py). All you have to do is specify an output directory for where the raw data should be stored:
```
python open_lm/datapreprocess/wiki_download.py --output-dir path/to/raw_data
```

Next we process our training data by running it through a BPE tokenizer and chunk it into chunks of appropriate length. By default we use the tokenizer attached with [GPT-NeoX-20B](https://github.com/EleutherAI/gpt-neox). To do this, use the script `datapreprocess/make_2048.py`:
```
>>> python open_lm/datapreprocess/make_2048.py \
--input-files path_to_raw_data/*.jsonl
--output-dir preproc_data
--num-workers 32
--num-consumers 1
```
Where `input-files` passes all of its (possibly many) arguments through the python `glob` module, allowing for wildcards. Optionally, data can be stored in S3 by setting the environment variables: `S3_BASE`, and passing the flag `--upload-to-s3` to the script. This saves sharded data to the given bucket with prefix of `S3_BASE`. E.g.
```
>>> export S3_BASE=preproc_data-v1/
>>> python open_lm/datapreprocess/make2048.py --upload-to-s3 ... # same arguments as before
```

## Run Training
Tokenized data can now be passed to the main training script, `open_lm/main.py`. Distributed computatation is handled via `torchrun`, and hyperparameters are specified by a variety of keyword arguments. We highlight several of the most important ones here:
- `train-data`: location of the sharded tokenized training data. If locally generated and stored, this will point to a directory containing files like `preproc_data/2048-v1/0/XXXXXXX.tar`. Data are processed using the [webdataset](https://github.com/webdataset/webdataset) package where wildcards are supported like `preproc_data/2048-v1/0/{0000000..0000099}.tar` to select the first 100 .tar files.
- `model`: Which model to use. See the table below to see valid options and parameter sizes for each.
- `train-num-samples`: how many samples to use from the specified training dataset
- `name`: name of this particular training run for logging purposes
- `report-to`: if present, can be `wandb`, `tensorboard`, or `all` to stash logging information on WandB or Tensorboard.


Model choices are contained in the following table, where, for instance `11m` indicates an 11 million parameter model and `1b` indicates a 1 billion parameter model.
<center>

| Model Name |
|---------------|
| `open_lm_11m` |
| `open_lm_25m` |
| `open_lm_87m` |
| `open_lm_160m`|
| `open_lm_411m`|
| `open_lm_830m`|
| `open_lm_1b` |
| `open_lm_3b` |
| `open_lm_7b` |

</center>

An example training run can be called as follows:
```
>>> export CUDA_VISIBLE_DEVICES=0,1,2,3
>>> torchrun --nproc-per-node 4 -m open_lm.main \
--model open_lm_3b \
--train-data /preproc_data/shard-{0000000..0000099}.tar \
--train-num-samples 1000000000 \
--workers 8 \
--dataset-resampled \
--precision amp_bfloat16 \
--batch-size 8 \
--grad-checkpointing \
--log-every-n-steps 100 \
--grad-clip-norm 1 \
--data-key txt \
--lr 3e-4 \
--fsdp --fsdp-amp \
--warmup 2000 \
--wd 0.1 \
--beta2 0.95 \
--epochs 100 \
--report-to wandb \
--wandb-project-name open_lm_example \
--name open_lm_ex_$RANDOM \
--resume latest \
--logs path/to/logging/dir/
```
Checkpoints and final model weights will be saved to the specified logs directory.

During training, the above command will pick shards to train on via sampling with replacement. Training can also be done by picking shards via sampling without replacement. To do this, the input dataset(s) must first be preprocessed using the following command:
```
python -m open_lm.utils.make_wds_manifest --data-dir /preproc_data/
```
This will create a file called ```manifest.jsonl``` under ```/preproc_data```. Training can then be done by sampling wihout replacement via the following example commands:
```
>>> export CUDA_VISIBLE_DEVICES=0,1,2,3
>>> torchrun --nproc-per-node 4 -m open_lm.main \
--model open_lm_3b \
--dataset-manifest /preproc_data/manifest.jsonl \
--train-num-samples 1000000000 \
--workers 8 \
--precision amp_bfloat16 \
--batch-size 8 \
--grad-checkpointing \
--log-every-n-steps 100 \
--grad-clip-norm 1 \
--data-key txt \
--lr 3e-4 \
--fsdp --fsdp-amp \
--warmup 2000 \
--wd 0.1 \
--beta2 0.95 \
--epochs 100 \
--report-to wandb \
--wandb-project-name open_lm_example \
--name open_lm_ex_$RANDOM \
--resume latest \
--logs path/to/logging/dir/
```

### Dataset manifest

The manifest created with `open_lm/utils/make_wds_manifest.py` is a `jsonl` file describing the dataset. Each line in this file corresponds to a shard of the dataset and is a `json` object containing two fields:

- `"shard"`: the name of a shard in the dataset.
- `"num_sequences"`: the number of sequences contained in the shards. Each sequence contains a set length of tokens.

This manifest file provides auxiliary information about the dataset, and is assumed to be found within the same directory as the shards.

## Evaluate Model
Once trained, we can evaluate the model. This requires [LLM Foundry](https://github.com/mosaicml/llm-foundry), which can be installed via `pip install llm-foundry`. Next some configurations are required to pass to the evaluator: a skeleton of these parameters is located at [eval/in_memory_hf_eval.yaml](eval/in_memory_hf_eval.yaml). Then just run the following script, making sure to point it at the checkpoint of your trained model (and it's correspending config .json file):
```
cd eval

python eval_openlm_ckpt.py \
--eval-yaml in_memory_hf_eval.yaml \
--model open_lm_1b \
--checkpoint /path/to/openlm_checkpoint.pt
--positional_embedding_type head_rotary

```
Note that `--positional-embedding-type head_rotary` is only necessary if using the pretrained `open_lm_1b` model hosted below. See discussion in the next section about this.

## Generate Text
One can also use a trained model to generate text. This is accessible via the script located at [scripts/generate.py](scripts/generate.py). The parameters are similar to those used in evaluation:
```
cd scripts

python generate.py \
--model open_lm_1b \
--checkpoint /path/to/openlm_checkpoint.pt \
--positional-embedding-type head_rotary \
--input-text "Please give me a recipe for chocolate chip cookies"
```

Again, note that `--positional-embedding-type head_rotary` is only necessary for the pretrained `open_lm_1b` model hosted below.

# Pretrained Models

## [OpenLM 1B](https://huggingface.co/mlfoundations/open_lm_1B)
OpenLM 1B is a ~1Billion parameter model trained on a 1.6T token dataset which consists of a mix of RedPajama, Pile, S2ORC, The Pile of Law, Deepmind Math, and RealNews (the full mixture of training data is described in [more detail here](https://docs.google.com/spreadsheets/d/1YW-_1vGsSPmVtEt2oeeJOecH6dYX2SuEuhOwZyGwy4k/edit?usp=sharing)).
The model checkpoint can be downloaded from [HuggingFace here](https://huggingface.co/mlfoundations/open_lm_1B/tree/main).
The script used to train this model (for config-copying purposes) is [located here](https://github.com/mlfoundations/open_lm/blob/main/scripts/train_example.sh).
Once this checkpoint has been downloaded, you can evaluate it by following the directions in the [Evaluate Model](#evaluate-model) section above and passing `--positional-embedding-type head_rotary` or setting `"positional_embedding_type": "head_rotary"` in the model config (see note below).

Note: We trained this model with rotary embeddings applied to the _head_
dimension, which is the default in xformers as of 09/01/2023. Since these models
were trained, we have updated openlm to correctly apply the rotary embeddings to
the sequence dimension (see
[this issue](https://github.com/mlfoundations/open_lm/issues/4) and [this
issue](https://github.com/facebookresearch/xformers/issues/841) for details).
To evaluate these models, ensure you use the `"positional_embedding_type": "head_rotary"` in the model config.

| **OpenLM-1B** | **250B Tokens** | **500B tokens** | **750B tokens** | **1T Tokens** | **1.25T Tokens** | **1.5T Tokens** | **1.6T Tokens** |
|----------------|-----------------|-----------------|-----------------|---------------|------------------|-----------------|-----------------|
| | | | | | | | |
| arc_challenge | 0.27 | 0.28 | 0.29 | 0.28 | 0.29 | 0.31 | 0.31 |
| arc_easy | 0.49 | 0.50 | 0.51 | 0.53 | 0.54 | 0.56 | 0.56 |
| boolq | 0.60 | 0.61 | 0.62 | 0.62 | 0.65 | 0.64 | 0.65 |
| copa | 0.71 | 0.70 | 0.70 | 0.78 | 0.71 | 0.73 | 0.70 |
| hellaswag | 0.50 | 0.54 | 0.54 | 0.57 | 0.59 | 0.61 | 0.61 |
| lambada_openai | 0.56 | 0.57 | 0.61 | 0.61 | 0.65 | 0.65 | 0.66 |
| piqa | 0.70 | 0.70 | 0.71 | 0.72 | 0.73 | 0.74 | 0.74 |
| triviaqa | | | | | | | |
| winogrande | 0.55 | 0.57 | 0.58 | 0.59 | 0.61 | 0.60 | 0.60 |
| MMLU | 0.24 | 0.24 | 0.24 | 0.23 | 0.26 | 0.24 | 0.25 |
| Jeopardy | 0.01 | 0.02 | 0.01 | 0.01 | 0.04 | 0.09 | 0.10 |
| Winograd | 0.75 | 0.77 | 0.77 | 0.79 | 0.81 | 0.80 | 0.79 |
| | | | | | | | |
| **Average** | **0.49** | **0.50** | **0.51** | **0.52** | **0.53** | **0.54** | **0.54** |


| **1B Baselines** | **OPT-1.3B** | **Pythia-1B** | **Neox-1.3B** | **OPT-IML-1.3B** |
|------------------|-------------:|--------------:|--------------:|-----------------:|
| arc_challenge | 0.27 | 0.26 | 0.26 | 0.30 |
| arc_easy | 0.49 | 0.51 | 0.47 | 0.58 |
| boolq | 0.58 | 0.61 | 0.62 | 0.72 |
| copa | 0.75 | 0.68 | 0.72 | 0.73 |
| hellaswag | 0.54 | 0.49 | 0.48 | 0.54 |
| lambada_openai | 0.59 | 0.58 | 0.57 | 0.57 |
| piqa | 0.72 | 0.70 | 0.72 | 0.73 |
| triviaqa | | | | |
| winogrande | 0.59 | 0.53 | 0.55 | 0.59 |
| MMLU | 0.25 | 0.26 | 0.26 | 0.30 |
| Jeopardy | 0.01 | 0.00 | 0.00 | 0.12 |
| Winograd | 0.74 | 0.71 | 0.75 | 0.73 |
| **Average** | **0.50** | **0.48** | **0.49** | **0.54** |


## [OpenLM 7B](https://huggingface.co/mlfoundations/open_lm_7B_1.25T)
OpenLM 7B is not yet done training, but we've released a checkpoint at 1.25T tokens. Information is the same as for OpenLM-1B above, including the information pertaining to rotary embeddings.


| **OpenLM-7B** | **275B Tokens** | **500B tokens** | **675B tokens** | **775B tokens** | **1T Tokens** | **1.25T Tokens** | **1.5T Tokens** | **1.6T Tokens** | **LLAMA-7B** | **MPT-7B** |
|-----------------|-----------------|-----------------|-----------------|-----------------|---------------|------------------|-----------------|-----------------|--------------|------------|
| arc_challenge | 0.35 | 0.35 | 0.36 | 0.37 | 0.39 | 0.39 | | | 0.41 | 0.39 |
| arc_easy | 0.60 | 0.61 | 0.62 | 0.62 | 0.63 | 0.66 | | | 0.65 | 0.67 |
| boolq | 0.67 | 0.66 | 0.69 | 0.69 | 0.70 | 0.70 | | | 0.77 | 0.75 |
| copa | 0.75 | 0.79 | 0.75 | 0.80 | 0.80 | 0.78 | | | 0.78 | 0.81 |
| hellaswag | 0.64 | 0.67 | 0.68 | 0.68 | 0.69 | 0.70 | | | 0.75 | 0.76 |
| lambada_openai | 0.67 | 0.68 | 0.69 | 0.70 | 0.70 | 0.70 | | | 0.74 | 0.70 |
| piqa | 0.75 | 0.76 | 0.76 | 0.76 | 0.77 | 0.77 | | | 0.79 | 0.80 |
| triviaqa | | | | | | | | | | |
| winogrande | 0.62 | 0.65 | 0.65 | 0.65 | 0.67 | 0.67 | | | 0.68 | 0.68 |
| MMLU-0 shot | 0.25 | 0.25 | 0.27 | 0.27 | 0.28 | 0.30 | | | 0.30 | 0.30 |
| Jeopardy | 0.15 | 0.18 | 0.23 | 0.22 | 0.16 | 0.21 | | | 0.33 | 0.31 |
| Winograd | 0.82 | 0.81 | 0.84 | 0.84 | 0.85 | 0.86 | | | 0.81 | 0.88 |
| | | | | | | | | | | |
| **Average** | **0.57** | **0.58** | **0.60** | **0.60** | **0.60** | **0.61** | | | **0.64** | **0.64** |
| **MMLU-5 shot** | | | | | | **0.34** | | | **0.34** | |

# Unit tests

For unit tests we use `pytest`. Either

```
pip install pytest
```
or create the `open_lm_tests` conda environment by running
```
conda env create --file environment-tests.yml
```

Tests live in the `tests/` folder.

To run tests make sure you are on a machine with a GPU and run:
```
pytest tests/
## Citations
```bibtex
@inproceedings{Mercat2024LinearizingLL,
title={Linearizing Large Language Models},
author={Jean-Pierre Mercat and Igor Vasiljevic and Sedrick Scott Keh and Kushal Arora and Achal Dave and Adrien Gaidon and Thomas Kollar},
year={2024},
url={https://arxiv.org/abs/2405.06640}
}
```

# Team and acknowledgements

Team (so-far, * = equal contrib): Suchin Gururangan*, Mitchell Wortsman*, Samir Yitzhak Gadre*, Achal Dave*, Maciej Kilian, Weijia Shi, Jean Mercat, Georgios Smyrnis, Gabriel Ilharco, Matt Jordan, Reinhard Heckel, Alex Dimakis, Ali Farhadi, Vaishaal Shankar*, Ludwig Schmidt.

Code is based heavily on [open-clip](https://github.com/mlfoundations/open_clip) developed by a team including Ross Wightman, Romain Beaumont, Cade Gordon, Mehdi Cherti, Jenia Jitsev, and [open-flamingo](https://github.com/mlfoundations/open_flamingo), developed by a team including Anas Awadalla and Irena Gao. Additional inspiration is from [lit-llama](https://github.com/Lightning-AI/lit-llama).
We are greatful to stability.ai for resource support.
OpenLM is developed by researchers from various affiliations including the [RAIVN Lab](https://raivn.cs.washington.edu/) at the University of Washington, [UWNLP](https://nlp.washington.edu/), [Toyota Research Institute](https://www.tri.global/), [Columbia University](https://www.columbia.edu/), and more.

Citation
--------

If you use this model in your work, please use the following BibTeX citation:
```bibtex
@misc{open_lm,
author = {Gururangan, Suchin and Wortsman, Mitchell and Gadre, Samir Yitzhak and Dave, Achal and Kilian, Maciej and Shi, Weijia and Mercat, Jean and Smyrnis, Georgios and Ilharco, Gabriel and Jordan, Matt and Heckel, Reinhard and Dimakis, Alex and Farhadi, Ali and Shankar, Vaishaal and Schmidt, Ludwig},
Expand Down
Loading