README.md 29.5 KB
Newer Older
Leo Gao's avatar
Leo Gao committed
1
# Language Model Evaluation Harness
Anish Thite's avatar
Anish Thite committed
2

3
[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.10256836.svg)](https://doi.org/10.5281/zenodo.10256836)
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
4

Stella Biderman's avatar
Stella Biderman committed
5
## Announcement
lintangsutawika's avatar
lintangsutawika committed
6
**A new v0.4.0 release of lm-evaluation-harness is available** !
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
7
8
9
10
11

New updates and features include:

- Internal refactoring
- Config-based task creation and configuration
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
12
- Easier import and sharing of externally-defined task config YAMLs
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
13
14
15
16
17
18
19
20
- Support for Jinja2 prompt design, easy modification of prompts + prompt imports from Promptsource
- More advanced configuration options, including output post-processing, answer extraction, and multiple LM generations per document, configurable fewshot settings, and more
- Speedups and new modeling libraries supported, including: faster data-parallel HF model usage, vLLM support, MPS support with HuggingFace, and more
- Logging and usability changes
- New tasks including CoT BIG-Bench-Hard, Belebele, user-defined task groupings, and more

Please see our updated documentation pages in `docs/` for more details.

Anjor Kanekar's avatar
Anjor Kanekar committed
21
Development will be continuing on the `main` branch, and we encourage you to give us feedback on what features are desired and how to improve the library further, or ask questions, either in issues or PRs on GitHub, or in the [EleutherAI discord](https://discord.gg/eleutherai)!
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
22

Fabrizio Milo's avatar
Fabrizio Milo committed
23
## Overview
Anish Thite's avatar
Anish Thite committed
24

Stella Biderman's avatar
Stella Biderman committed
25
This project provides a unified framework to test generative language models on a large number of different evaluation tasks.
Leo Gao's avatar
Leo Gao committed
26

Stella Biderman's avatar
Stella Biderman committed
27
**Features:**
Stella Biderman's avatar
Stella Biderman committed
28
- Over 60 standard academic benchmarks for LLMs, with hundreds of subtasks and variants implemented.
Stella Biderman's avatar
Stella Biderman committed
29
- Support for models loaded via [transformers](https://github.com/huggingface/transformers/) (including quantization via [AutoGPTQ](https://github.com/PanQiWei/AutoGPTQ)), [GPT-NeoX](https://github.com/EleutherAI/gpt-neox), and [Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed/), with a flexible tokenization-agnostic interface.
30
- Support for fast and memory-efficient inference with [vLLM](https://github.com/vllm-project/vllm).
31
- Support for commercial APIs including [OpenAI](https://openai.com), and [TextSynth](https://textsynth.com/).
Stella Biderman's avatar
Stella Biderman committed
32
33
34
- Support for evaluation on adapters (e.g. LoRA) supported in [HuggingFace's PEFT library](https://github.com/huggingface/peft).
- Support for local models and benchmarks.
- Evaluation with publicly available prompts ensures reproducibility and comparability between papers.
Stella Biderman's avatar
Stella Biderman committed
35
- Easy support for custom prompts and evaluation metrics.
Stella Biderman's avatar
Stella Biderman committed
36

Stella Biderman's avatar
Stella Biderman committed
37
The Language Model Evaluation Harness is the backend for 🤗 Hugging Face's popular [Open LLM Leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard), has been used in [hundreds of papers](https://scholar.google.com/scholar?oi=bibs&hl=en&authuser=2&cites=15052937328817631261,4097184744846514103,1520777361382155671,17476825572045927382,18443729326628441434,14801318227356878622,7890865700763267262,12854182577605049984,15641002901115500560,5104500764547628290), and is used internally by dozens of organizations including NVIDIA, Cohere, BigScience, BigCode, Nous Research, and Mosaic ML.
38

Leo Gao's avatar
Leo Gao committed
39
40
## Install

Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
41
To install the `lm-eval` package from the github repository, run:
42

Leo Gao's avatar
Leo Gao committed
43
```bash
44
45
46
git clone https://github.com/EleutherAI/lm-evaluation-harness
cd lm-evaluation-harness
pip install -e .
Leo Gao's avatar
Leo Gao committed
47
```
48

Baber Abbasi's avatar
Baber Abbasi committed
49
We also provide a number of optional dependencies for extended functionality. A detailed table is available at the end of this document.
haileyschoelkopf's avatar
haileyschoelkopf committed
50

Leo Gao's avatar
Leo Gao committed
51
52
## Basic Usage

Stella Biderman's avatar
Stella Biderman committed
53
54
### Hugging Face `transformers`

Anjor Kanekar's avatar
Anjor Kanekar committed
55
To evaluate a model hosted on the [HuggingFace Hub](https://huggingface.co/models) (e.g. GPT-J-6B) on `hellaswag` you can use the following command (this assumes you are using a CUDA-compatible GPU):
jon-tow's avatar
jon-tow committed
56

Leo Gao's avatar
Leo Gao committed
57
```bash
Stella Biderman's avatar
Stella Biderman committed
58
lm_eval --model hf \
Stella Biderman's avatar
Stella Biderman committed
59
    --model_args pretrained=EleutherAI/gpt-j-6B \
Stella Biderman's avatar
Stella Biderman committed
60
    --tasks hellaswag \
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
61
62
    --device cuda:0 \
    --batch_size 8
Leo Gao's avatar
Leo Gao committed
63
64
```

Stella Biderman's avatar
Stella Biderman committed
65
Additional arguments can be provided to the model constructor using the `--model_args` flag. Most notably, this supports the common practice of using the `revisions` feature on the Hub to store partially trained checkpoints, or to specify the datatype for running a model:
Leo Gao's avatar
Leo Gao committed
66
67

```bash
Stella Biderman's avatar
Stella Biderman committed
68
lm_eval --model hf \
Stella Biderman's avatar
Stella Biderman committed
69
    --model_args pretrained=EleutherAI/pythia-160m,revision=step100000,dtype="float" \
jon-tow's avatar
jon-tow committed
70
    --tasks lambada_openai,hellaswag \
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
71
72
73
74
    --device cuda:0 \
    --batch_size 8
```

Lenni Justen's avatar
Lenni Justen committed
75
Models that are loaded via both `transformers.AutoModelForCausalLM` (autoregressive, decoder-only GPT style models) and `transformers.AutoModelForSeq2SeqLM` (such as encoder-decoder models like T5) in Huggingface are supported.
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
76

77
78
79
Batch size selection can be automated by setting the  ```--batch_size``` flag to ```auto```. This will perform automatic detection of the largest batch size that will fit on your device. On tasks where there is a large difference between the longest and shortest example, it can be helpful to periodically recompute the largest batch size, to gain a further speedup. To do this, append ```:N``` to above flag to automatically recompute the largest batch size ```N``` times. For example, to recompute the batch size 4 times, the command would be:

```bash
Stella Biderman's avatar
Stella Biderman committed
80
lm_eval --model hf \
81
82
83
84
85
86
    --model_args pretrained=EleutherAI/pythia-160m,revision=step100000,dtype="float" \
    --tasks lambada_openai,hellaswag \
    --device cuda:0 \
    --batch_size auto:4
```

87
The full list of supported arguments are provided [here](./docs/interface.md), and on the terminal by calling `lm_eval -h`. Alternatively, you can use `lm-eval` instead of `lm_eval`.
88

Stella Biderman's avatar
Stella Biderman committed
89
> [!Note]
90
91
92
> Just like you can provide a local path to `transformers.AutoModel`, you can also provide a local path to `lm_eval` via `--model_args pretrained=/path/to/model`

#### Multi-GPU Evaluation with Hugging Face `accelerate`
93

94
95
96
We support two main ways of using Hugging Face's [accelerate 🚀](https://github.com/huggingface/accelerate) library for multi-GPU evaluation.

To perform *data-parallel evaluation* (where each GPU loads a **separate full copy** of the model), we leverage the `accelerate` launcher as follows:
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
97
98

```
Stella Biderman's avatar
Stella Biderman committed
99
accelerate launch -m lm_eval --model hf \
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
100
    --tasks lambada_openai,arc_easy \
101
    --batch_size 16
Leo Gao's avatar
Leo Gao committed
102
```
103
104
105
(or via `accelerate launch --no-python lm_eval`).

For cases where your model can fit on a single GPU, this allows you to evaluate on K GPUs K times faster than on one.
Leo Gao's avatar
Leo Gao committed
106

107
**WARNING**: This setup does not work with FSDP model sharding, so in `accelerate config` FSDP must be disabled, or the NO_SHARD FSDP option must be used.
Stella Biderman's avatar
Stella Biderman committed
108

109
The second way of using `accelerate` for multi-GPU evaluation is when your model is *too large to fit on a single GPU.*
110

111
112
113
114
115
116
117
118
119
120
121
122
In this setting, run the library *outside of the `accelerate` launcher*, but passing `parallelize=True` to `--model_args` as follows:

```
lm_eval --model hf \
    --tasks lambada_openai,arc_easy \
    --model_args parallelize=True \
    --batch_size 16
```

This means that your model's weights will be split across all available GPUs.

For more advanced users or even larger models, we allow for the following arguments when `parallelize=True` as well:
123
124
125
126
127
- `device_map_option`: How to split model weights across available GPUs. defaults to "auto".
- `max_memory_per_gpu`: the max GPU memory to use per GPU in loading the model.
- `max_cpu_memory`: the max amount of CPU memory to use when offloading the model weights to RAM.
- `offload_folder`: a folder where model weights will be offloaded to disk if needed.

128
These two options (`accelerate launch` and `parallelize=True`) are mutually exclusive.
Zach Nussbaum's avatar
Zach Nussbaum committed
129

130
131
**Note: we do not currently support multi-node evaluations natively, and advise using either an externally hosted server to run inference requests against, or creating a custom integration with your distributed framework [as is done for the GPT-NeoX library](https://github.com/EleutherAI/gpt-neox/blob/main/eval_tasks/eval_adapter.py).**

132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
### NVIDIA `nemo` models

[NVIDIA NeMo Framework](https://github.com/NVIDIA/NeMo) is a generative AI framework built for researchers and pytorch developers working on language models.

To evaluate a `nemo` model, start by installing NeMo following [the documentation](https://github.com/NVIDIA/NeMo?tab=readme-ov-file#installation). We highly recommended to use the NVIDIA PyTorch or NeMo container, especially if having issues installing Apex or any other dependencies (see [latest released containers](https://github.com/NVIDIA/NeMo/releases)). Please also install the lm evaluation harness library following the instructions in [the Install section](https://github.com/EleutherAI/lm-evaluation-harness/tree/main?tab=readme-ov-file#install).

NeMo models can be obtained through [NVIDIA NGC Catalog](https://catalog.ngc.nvidia.com/models) or in [NVIDIA's Hugging Face page](https://huggingface.co/nvidia). In [NVIDIA NeMo Framework](https://github.com/NVIDIA/NeMo/tree/main/scripts/nlp_language_modeling) there are conversion scripts to convert the `hf` checkpoints of popular models like llama, falcon, mixtral or mpt to `nemo`.

Run a `nemo` model on one GPU:
```bash
lm_eval --model nemo_lm \
    --model_args path=<path_to_nemo_model> \
    --tasks hellaswag \
    --batch_size 32
```

It is recommended to unpack the `nemo` model to avoid the unpacking inside the docker container - it may overflow disk space. For that you can run:

```
mkdir MY_MODEL
tar -xvf MY_MODEL.nemo -c MY_MODEL
```

#### Multi-GPU evaluation with NVIDIA `nemo` models

By default, only one GPU is used. But we do support either data replication or tensor/pipeline parallelism during evaluation, on one node.

1) To enable data replication, set the `model_args` of `devices` to the number of data replicas to run. For example, the command to run 8 data replicas over 8 GPUs is:
```bash
torchrun --nproc-per-node=8 --no-python lm_eval \
    --model nemo_lm \
    --model_args path=<path_to_nemo_model>,devices=8 \
    --tasks hellaswag \
    --batch_size 32
```

2) To enable tensor and/or pipeline parallelism, set the `model_args` of `tensor_model_parallel_size` and/or `pipeline_model_parallel_size`. In addition, you also have to set up `devices` to be equal to the product of `tensor_model_parallel_size` and/or `pipeline_model_parallel_size`. For example, the command to use one node of 4 GPUs with tensor parallelism of 2 and pipeline parallelism of 2 is:
```bash
torchrun --nproc-per-node=4 --no-python lm_eval \
    --model nemo_lm \
    --model_args path=<path_to_nemo_model>,devices=4,tensor_model_parallel_size=2,pipeline_model_parallel_size=2 \
    --tasks hellaswag \
    --batch_size 32
```
Note that it is recommended to substitute the `python` command by `torchrun --nproc-per-node=<number of devices> --no-python` to facilitate loading the model into the GPUs. This is especially important for large checkpoints loaded into multiple GPUs.

Not supported yet: multi-node evaluation and combinations of data replication with tensor or pipeline parallelism.
179

180
### Tensor + Data Parallel and Optimized Inference with `vLLM`
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
181

182
We also support vLLM for faster inference on [supported model types](https://docs.vllm.ai/en/latest/models/supported_models.html), especially faster when splitting a model across multiple GPUs. For single-GPU or multi-GPU — tensor parallel, data parallel, or a combination of both — inference, for example:
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
183
184

```bash
Stella Biderman's avatar
Stella Biderman committed
185
lm_eval --model vllm \
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
186
    --model_args pretrained={model_name},tensor_parallel_size={GPUs_per_model},dtype=auto,gpu_memory_utilization=0.8,data_parallel_size={model_replicas} \
187
    --tasks lambada_openai \
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
188
189
    --batch_size auto
```
190
To use vllm, do `pip install lm_eval[vllm]`. For a full list of supported vLLM configurations, please reference our [vLLM integration](https://github.com/EleutherAI/lm-evaluation-harness/blob/e74ec966556253fbe3d8ecba9de675c77c075bce/lm_eval/models/vllm_causallms.py) and the vLLM documentation.
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
191

192
vLLM occasionally differs in output from Huggingface. We treat Huggingface as the reference implementation, and provide a [script](./scripts/model_comparator.py) for checking the validity of vllm results against HF.
193

194
195
196
197
198
199
> [!Tip]
> For fastest performance, we recommend using `--batch_size auto` for vLLM whenever possible, to leverage its continuous batching functionality!

> [!Tip]
> Passing `max_model_len=4096` or some other reasonable default to vLLM through model args may cause speedups or prevent out-of-memory errors when trying to use auto batch size, such as for Mistral-7B-v0.1 which defaults to a maximum length of 32k.

200
### Model APIs and Inference Servers
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
201
202

Our library also supports the evaluation of models served via several commercial APIs, and we hope to implement support for the most commonly used performant local/self-hosted inference servers.
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
203

204
205
206
To call a hosted model, use:

```bash
207
export OPENAI_API_KEY=YOUR_KEY_HERE
208
lm_eval --model openai-completions \
Anjor Kanekar's avatar
Anjor Kanekar committed
209
    --model_args model=davinci \
210
211
212
    --tasks lambada_openai,hellaswag
```

213
We also support using your own local inference server with servers that mirror the OpenAI Completions and ChatCompletions APIs.
214
215
216
217

```bash
lm_eval --model local-chat-completions --tasks gsm8k --model_args model=facebook/opt-125m,base_url=http://{yourip}:8000/v1
```
218
219
Note that for externally hosted models, configs such as `--device` and `--batch_size` should not be used and do not function. Just like you can use `--model_args` to pass arbitrary arguments to the model constructor for local models, you can use it to pass arbitrary arguments to the model API for hosted models. See the documentation of the hosting service for information on what arguments they support.

220
221
| API or Inference Server                                                                                                   | Implemented?                    | `--model <xxx>` name                                                | Models supported:                                                                             | Request Types:                                             |
|---------------------------------------------------------------------------------------------------------------------------|---------------------------------|---------------------------------------------------------------------|-----------------------------------------------------------------------------------------------|------------------------------------------------------------|
222
| OpenAI Completions                                                                                                        | :heavy_check_mark:              | `openai-completions`, `local-completions` | All OpenAI Completions API models                                            | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
223
224
225
226
| OpenAI ChatCompletions                                                                                                    | :heavy_check_mark:        | `openai-chat-completions`, `local-chat-completions`                                                               | [All ChatCompletions API models](https://platform.openai.com/docs/guides/gpt)                 | `generate_until` (no logprobs)                             |
| Anthropic                                                                                                                 | :heavy_check_mark:              | `anthropic`                                                         | [Supported Anthropic Engines](https://docs.anthropic.com/claude/reference/selecting-a-model)  | `generate_until` (no logprobs)                             |
| Textsynth                                                                                                                 | :heavy_check_mark:                   | `textsynth`                                                         | [All supported engines](https://textsynth.com/documentation.html#engines)                     | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
| Cohere                                                                                                                    | [:hourglass: - blocked on Cohere API bug](https://github.com/EleutherAI/lm-evaluation-harness/pull/395) | N/A                                                                 | [All `cohere.generate()` engines](https://docs.cohere.com/docs/models)                        | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
227
| [Llama.cpp](https://github.com/ggerganov/llama.cpp) (via [llama-cpp-python](https://github.com/abetlen/llama-cpp-python)) | :heavy_check_mark:              | `gguf`, `ggml`                                                      | [All models supported by llama.cpp](https://github.com/ggerganov/llama.cpp)                   | `generate_until`, `loglikelihood`, (perplexity evaluation not yet implemented) |
228
| vLLM                                                                                                                      | :heavy_check_mark:       | `vllm`                                                              | [Most HF Causal Language Models](https://docs.vllm.ai/en/latest/models/supported_models.html) | `generate_until`, `loglikelihood`, `loglikelihood_rolling` |
229
| Mamba                       | :heavy_check_mark:       | `mamba_ssm`                                                                      | [Mamba architecture Language Models via the `mamba_ssm` package](https://huggingface.co/state-spaces) | `generate_until`, `loglikelihood`, `loglikelihood_rolling`                             |
230
| Huggingface Optimum (Causal LMs)    | ✔️         | `openvino`                                 |     Any decoder-only AutoModelForCausalLM converted with Huggingface Optimum into OpenVINO™ Intermediate Representation (IR) format                           |  `generate_until`, `loglikelihood`, `loglikelihood_rolling`                         | ...                                                      |
231
| Neuron via AWS Inf2 (Causal LMs)    | ✔️         | `neuronx`                                 |     Any decoder-only AutoModelForCausalLM supported to run on [huggingface-ami image for inferentia2](https://aws.amazon.com/marketplace/pp/prodview-gr3e6yiscria2)                         |  `generate_until`, `loglikelihood`, `loglikelihood_rolling`                         | ...                                                      |
232
| Your local inference server!                                                                                              | :heavy_check_mark:                             | `local-completions` or `local-chat-completions` (using `openai-chat-completions` model type)    | Any server address that accepts GET requests using HF models and mirror's OpenAI's Completions or ChatCompletions interface                                  | `generate_until`                                           |                                | ...                |
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
233

234
Models which do not supply logits or logprobs can be used with tasks of type `generate_until` only, while local models, or APIs that supply logprobs/logits of their prompts, can be run on all task types: `generate_until`, `loglikelihood`, `loglikelihood_rolling`, and `multiple_choice`.
235
236

For more information on the different task `output_types` and model request types, see [our documentation](https://github.com/EleutherAI/lm-evaluation-harness/blob/main/docs/model_guide.md#interface).
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
237

Stella Biderman's avatar
Stella Biderman committed
238
239
### Other Frameworks

lintangsutawika's avatar
lintangsutawika committed
240
A number of other libraries contain scripts for calling the eval harness through their library. These include [GPT-NeoX](https://github.com/EleutherAI/gpt-neox/blob/main/eval_tasks/eval_adapter.py), [Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed/blob/main/examples/MoE/readme_evalharness.md), and [mesh-transformer-jax](https://github.com/kingoflolz/mesh-transformer-jax/blob/master/eval_harness.py).
Jason Phang's avatar
Jason Phang committed
241

242
243
To create your own custom integration you can follow instructions from [this tutorial](https://github.com/EleutherAI/lm-evaluation-harness/blob/main/docs/interface.md#external-library-usage).

Stella Biderman's avatar
Stella Biderman committed
244
### Additional Features
Baber Abbasi's avatar
Baber Abbasi committed
245
246
> [!Note]
> For tasks unsuitable for direct evaluation — either due risks associated with executing untrusted code or complexities in the evaluation process — the `--predict_only` flag is available to obtain decoded generations for post-hoc evaluation.
Stella Biderman's avatar
Stella Biderman committed
247

baberabb's avatar
baberabb committed
248
If you have a Metal compatible Mac, you can run the eval harness using the MPS back-end by replacing `--device cuda:0` with `--device mps` (requires PyTorch version 2.1 or higher).
Stella Biderman's avatar
Stella Biderman committed
249

250
251
252
253
> [!Note]
> You can inspect what the LM inputs look like by running the following command:
> ```bash
> python write_out.py \
254
>     --tasks <task1,task2,...> \
255
256
257
258
259
>     --num_fewshot 5 \
>     --num_examples 10 \
>     --output_base_path /path/to/output/folder
> ```
> This will write out one text file for each task.
jon-tow's avatar
jon-tow committed
260

Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
261
262
263
To verify the data integrity of the tasks you're performing in addition to running the tasks themselves, you can use the `--check_integrity` flag:

```bash
Stella Biderman's avatar
Stella Biderman committed
264
lm_eval --model openai \
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
265
266
267
268
269
    --model_args engine=davinci \
    --tasks lambada_openai,hellaswag \
    --check_integrity
```

Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
270
## Advanced Usage Tips
Stella Biderman's avatar
Stella Biderman committed
271
272
273

For models loaded with the HuggingFace  `transformers` library, any arguments provided via `--model_args` get passed to the relevant constructor directly. This means that anything you can do with `AutoModel` can be done with our library. For example, you can pass a local path via `pretrained=` or use models finetuned with [PEFT](https://github.com/huggingface/peft) by taking the call you would run to evaluate the base model and add `,peft=PATH` to the `model_args` argument:
```bash
Stella Biderman's avatar
Stella Biderman committed
274
lm_eval --model hf \
275
    --model_args pretrained=EleutherAI/gpt-j-6b,parallelize=True,load_in_4bit=True,peft=nomic-ai/gpt4all-j-lora \
Stella Biderman's avatar
Stella Biderman committed
276
277
278
    --tasks openbookqa,arc_easy,winogrande,hellaswag,arc_challenge,piqa,boolq \
    --device cuda:0
```
279

280
[GPTQ](https://github.com/PanQiWei/AutoGPTQ) quantized models can be loaded by specifying their file names in `,autogptq=NAME` (or `,autogptq=True` for default names) in the `model_args` argument:
281
282

```bash
Stella Biderman's avatar
Stella Biderman committed
283
lm_eval --model hf \
284
    --model_args pretrained=model-name-or-path,autogptq=model.safetensors,gptq_use_triton=True \
Stella Biderman's avatar
Stella Biderman committed
285
    --tasks hellaswag
286
287
```

Stella Biderman's avatar
Stella Biderman committed
288
289
We support wildcards in task names, for example you can run all of the machine-translated lambada tasks via `--task lambada_openai_mt_*`.

290
291
To save evaluation results provide an `--output_path`. We also support logging model responses with the `--log_samples` flag for post-hoc analysis.

Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
292
Additionally, one can provide a directory with `--use_cache` to cache the results of prior runs. This allows you to avoid repeated execution of the same (model, task) pairs for re-scoring.
293

294
For a full list of supported arguments, check out the [interface](https://github.com/EleutherAI/lm-evaluation-harness/blob/main/docs/interface.md) guide in our documentation!
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
295

296
297
## Visualizing Results

298
299
300
301
You can seamlessly visualize and analyze the results of your evaluation harness runs using both Weights & Biases (W&B) and Zeno.

### Zeno

302
303
You can use [Zeno](https://zenoml.com) to visualize the results of your eval harness runs.

Anjor Kanekar's avatar
Anjor Kanekar committed
304
First, head to [hub.zenoml.com](https://hub.zenoml.com) to create an account and get an API key [on your account page](https://hub.zenoml.com/account).
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
Add this key as an environment variable:

```bash
export ZENO_API_KEY=[your api key]
```

You'll also need to install the `lm_eval[zeno]` package extra.

To visualize the results, run the eval harness with the `log_samples` and `output_path` flags.
We expect `output_path` to contain multiple folders that represent individual model names.
You can thus run your evaluation on any number of tasks and models and upload all of the results as projects on Zeno.

```bash
lm_eval \
    --model hf \
    --model_args pretrained=EleutherAI/gpt-j-6B \
    --tasks hellaswag \
    --device cuda:0 \
    --batch_size 8 \
    --log_samples \
    --output_path output/gpt-j-6B
```

Then, you can upload the resulting data using the `zeno_visualize` script:

```bash
python scripts/zeno_visualize.py \
    --data_path output \
    --project_name "Eleuther Project"
```

This will use all subfolders in `data_path` as different models and upload all tasks within these model folders to Zeno.
If you run the eval harness on multiple tasks, the `project_name` will be used as a prefix and one project will be created per task.

339
340
You can find an example of this workflow in [examples/visualize-zeno.ipynb](examples/visualize-zeno.ipynb).

341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
### Weights and Biases

With the [Weights and Biases](https://wandb.ai/site) integration, you can now spend more time extracting deeper insights into your evaluation results. The integration is designed to streamline the process of logging and visualizing experiment results using the Weights & Biases (W&B) platform.

The integration provide functionalities

- to automatically log the evaluation results,
- log the samples as W&B Tables for easy visualization,
- log the `results.json` file as an artifact for version control,
- log the `<task_name>_eval_samples.json` file if the samples are logged,
- generate a comprehensive report for analysis and visualization with all the important metric,
- log task and cli specific configs,
- and more out of the box like the command used to run the evaluation, GPU/CPU counts, timestamp, etc.

First you'll need to install the lm_eval[wandb] package extra. Do `pip install lm_eval[wandb]`.

Authenticate your machine with an your unique W&B token. Visit https://wandb.ai/authorize to get one. Do `wandb login` in your command line terminal.

Run eval harness as usual with a `wandb_args` flag. Use this flag to provide arguments for initializing a wandb run ([wandb.init](https://docs.wandb.ai/ref/python/init)) as comma separated string arguments.

```bash
lm_eval \
    --model hf \
    --model_args pretrained=microsoft/phi-2,trust_remote_code=True \
    --tasks hellaswag,mmlu_abstract_algebra \
    --device cuda:0 \
    --batch_size 8 \
    --output_path output/phi-2 \
    --limit 10 \
    --wandb_args project=lm-eval-harness-integration \
    --log_samples
```

374
In the stdout, you will find the link to the W&B run page as well as link to the generated report. You can find an example of this workflow in [examples/visualize-wandb.ipynb](examples/visualize-wandb.ipynb), and an example of how to integrate it beyond the CLI.
375

Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
376
377
## How to Contribute or Learn More?

378
For more information on the library and how everything fits together, check out all of our [documentation pages](https://github.com/EleutherAI/lm-evaluation-harness/tree/main/docs)! We plan to post a larger roadmap of desired + planned library improvements soon, with more information on how contributors can help.
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
379

Stella Biderman's avatar
Stella Biderman committed
380
381
382
383
### Implementing new tasks

To implement a new task in the eval harness, see [this guide](./docs/new_task_guide.md).

384
In general, we follow this priority list for addressing concerns about prompting and other eval details:
Stella Biderman's avatar
Stella Biderman committed
385
386
387
388
1. If there is widespread agreement among people who train LLMs, use the agreed upon procedure.
2. If there is a clear and unambiguous official implementation, use that procedure.
3. If there is widespread agreement among people who evaluate LLMs, use the agreed upon procedure.
4. If there are multiple common implementations but not universal or widespread agreement, use our preferred option among the common implementations. As before, prioritize choosing from among the implementations found in LLM training papers.
Stella Biderman's avatar
Stella Biderman committed
389

Stella Biderman's avatar
Stella Biderman committed
390
These are guidelines and not rules, and can be overruled in special circumstances.
Stella Biderman's avatar
Stella Biderman committed
391

baberabb's avatar
baberabb committed
392
We try to prioritize agreement with the procedures used by other groups to decrease the harm when people inevitably compare runs across different papers despite our discouragement of the practice. Historically, we also prioritized the implementation from [Language Models are Few Shot Learners](https://arxiv.org/abs/2005.14165) as our original goal was specifically to compare results with that paper.
Hailey Schoelkopf's avatar
Hailey Schoelkopf committed
393

394
395
### Support

baberabb's avatar
baberabb committed
396
The best way to get support is to open an issue on this repo or join the [EleutherAI Discord server](https://discord.gg/eleutherai). The `#lm-thunderdome` channel is dedicated to developing this project and the `#release-discussion` channel is for receiving support for our releases. If you've used the library and have had a positive (or negative) experience, we'd love to hear from you!
397

Baber Abbasi's avatar
Baber Abbasi committed
398
399
400
401
402
403
404
405
## Optional Extras
Extras dependencies can be installed via `pip install -e ".[NAME]"`

| Name          | Use                                   |
|---------------|---------------------------------------|
| anthropic     | For using Anthropic's models          |
| dev           | For linting PRs and contributions     |
| gptq          | For loading models with GPTQ          |
Michael Feil's avatar
Michael Feil committed
406
| hf_transfer   | For speeding up HF Hub file downloads |
Baber Abbasi's avatar
Baber Abbasi committed
407
| ifeval        | For running the IFEval task           |
408
| neuronx       | For running on AWS inf2 instances     |
Baber Abbasi's avatar
Baber Abbasi committed
409
410
411
412
413
414
415
416
417
418
419
420
421
| mamba         | For loading Mamba SSM models          |
| math          | For running math task answer checking |
| multilingual  | For multilingual tokenizers           |
| openai        | For using OpenAI's models             |
| optimum       | For running Intel OpenVINO models     |
| promptsource  | For using PromptSource prompts        |
| sentencepiece | For using the sentencepiece tokenizer |
| testing       | For running library test suite        |
| vllm          | For loading models with vLLM          |
| zeno          | For visualizing results with Zeno     |
|---------------|---------------------------------------|
| all           | Loads all extras (not recommended)    |

Leo Gao's avatar
Leo Gao committed
422
423
424
## Cite as

```
Stella Biderman's avatar
Stella Biderman committed
425
426
427
428
429
430
431
432
433
@misc{eval-harness,
  author       = {Gao, Leo and Tow, Jonathan and Abbasi, Baber and Biderman, Stella and Black, Sid and DiPofi, Anthony and Foster, Charles and Golding, Laurence and Hsu, Jeffrey and Le Noac'h, Alain and Li, Haonan and McDonell, Kyle and Muennighoff, Niklas and Ociepa, Chris and Phang, Jason and Reynolds, Laria and Schoelkopf, Hailey and Skowron, Aviya and Sutawika, Lintang and Tang, Eric and Thite, Anish and Wang, Ben and Wang, Kevin and Zou, Andy},
  title        = {A framework for few-shot language model evaluation},
  month        = 12,
  year         = 2023,
  publisher    = {Zenodo},
  version      = {v0.4.0},
  doi          = {10.5281/zenodo.10256836},
  url          = {https://zenodo.org/records/10256836}
Leo Gao's avatar
Leo Gao committed
434
435
}
```