[Model] EXAONE 3.0 model support #7942
Conversation
|
👋 Hi! Thank you for contributing to the vLLM project. Once the PR is approved and ready to go, please make sure to run full CI as it is required to merge (or just use auto-merge). To run full CI, you can do one of these:
🚀 |
Deepfocused
changed the title
|
/ready |
github-actions
bot
added
the
ready
There was a problem hiding this comment.
This seems to work locally for me and seems reasonable, thanks!
>>> from vllm import LLM
>>> model = LLM("LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct")
>>> model.chat([{"role":"user","content":"Hello!"}])
Processed prompts: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1/1 [00:00<00:00, 4.86it/s, est. speed input: 102.15 toks/s, output: 77.82 toks/s]
[RequestOutput(request_id=1, prompt='[|system|][|endofturn|]\n[|user|]Hello!\n[|assistant|]', prompt_token_ids=[420, 453, 47982, 453, 422, 361, 560, 420, 453, 14719, 453, 422, 33381, 362, 560, 420, 453, 1167, 8659, 453, 422], encoder_prompt=None, encoder_prompt_token_ids=None, prompt_logprobs=None, outputs=[CompletionOutput(index=0, text='Hello! How can I help you today? If you have any questions or need', token_ids=array('l', [33381, 362, 2417, 1093, 768, 2347, 904, 4963, 392, 2386, 904, 1072, 1491, 5344, 913, 1788]), cumulative_logprob=None, logprobs=None, finish_reason=length, stop_reason=None)], finished=True, metrics=RequestMetrics(arrival_time=1724860809.4073465, last_token_time=1724860809.4073465, first_scheduled_time=1724860809.408349, first_token_time=1724860809.4238176, time_in_queue=0.0010025501251220703, finished_time=1724860809.6029909, scheduler_time=0.000975252129137516, model_forward_time=None, model_execute_time=None), lora_request=None)]Also ran an eval on gsm8k
lm_eval --model vllm --model_args pretrained=LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct,max_model_len=4096,enable_chunked_prefill=True --tasks gsm8k --batch_size auto
vllm (pretrained=LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct,max_model_len=4096,enable_chunked_prefill=True), gen_kwargs: (None), limit: None, num_fewshot: None, batch_size: auto
|Tasks|Version| Filter |n-shot| Metric | |Value | |Stderr|
|-----|------:|----------------|-----:|-----------|---|-----:|---|-----:|
|gsm8k| 3|flexible-extract| 5|exact_match|↑ |0.8044|± |0.0109|
| | |strict-match | 5|exact_match|↑ |0.8021|± |0.0110|
|
The existing unopened EXAONE v1 and EXAONE v2 models had a different structure from llama, but the newly released EXAONE v3 model has the same structure as llama, but there are differences in variable names. Therefore, in order to run EXAONE v3 on vLLM, several variable names and weight names had to be changed in the llama structure. For example, There were variable changes such as these. Thank you~ |
|
Merged #7819 |
Hello, I am deepfocused from the AI engineering Team at LG CNS.
we want to integrate the recently opened model EXAONE-3.0-7.8B-Instruct model into vLLM.
Previously, we have been servicing models such as EXAONE v1, v2, etc. to our customers as vLLM through the vLLM + EXAONE build. Unlike these previous models, EXAONE 3.0 was released as an open model, so we are requesting integration.
It was written based on the llama.py model,
To run the EXAONE 3.0 model,
For offline execution,
For vllm server execution,
❗ Added exaone configuration file to vllm/transformers_utils to allow omitting trust_remote_code or --trust-remote-code at runtime.
Once Exaone is integrated into the transformer library, the above tasks will no longer be necessary ~
thank you.
BEFORE SUBMITTING, PLEASE READ THE CHECKLIST BELOW AND FILL IN THE DESCRIPTION ABOVE
PR Checklist (Click to Expand)
Thank you for your contribution to vLLM! Before submitting the pull request, please ensure the PR meets the following criteria. This helps vLLM maintain the code quality and improve the efficiency of the review process.
PR Title and Classification
Only specific types of PRs will be reviewed. The PR title is prefixed appropriately to indicate the type of change. Please use one of the following:
[Bugfix]for bug fixes.[CI/Build]for build or continuous integration improvements.[Doc]for documentation fixes and improvements.[Model]for adding a new model or improving an existing model. Model name should appear in the title.[Frontend]For changes on the vLLM frontend (e.g., OpenAI API server,LLMclass, etc.)[Kernel]for changes affecting CUDA kernels or other compute kernels.[Core]for changes in the core vLLM logic (e.g.,LLMEngine,AsyncLLMEngine,Scheduler, etc.)[Hardware][Vendor]for hardware-specific changes. Vendor name should appear in the prefix (e.g.,[Hardware][AMD]).[Misc]for PRs that do not fit the above categories. Please use this sparingly.Note: If the PR spans more than one category, please include all relevant prefixes.
Code Quality
The PR need to meet the following code quality standards:
format.shto format your code.docs/source/if the PR modifies the user-facing behaviors of vLLM. It helps vLLM user understand and utilize the new features or changes.Notes for Large Changes
Please keep the changes as concise as possible. For major architectural changes (>500 LOC excluding kernel/data/config/test), we would expect a GitHub issue (RFC) discussing the technical design and justification. Otherwise, we will tag it with
rfc-requiredand might not go through the PR.What to Expect for the Reviews
The goal of the vLLM team is to be a transparent reviewing machine. We would like to make the review process transparent and efficient and make sure no contributor feel confused or frustrated. However, the vLLM team is small, so we need to prioritize some PRs over others. Here is what you can expect from the review process:
action-requiredlabel on the PR if there are changes required. The contributor should address the comments and ping the reviewer to re-review the PR.Thank You
Finally, thank you for taking the time to read these guidelines and for your interest in contributing to vLLM. Your contributions make vLLM a great tool for everyone!