[go: nahoru, domu]

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

Fix index error when profiling token generation with -ct 1 #1898

Open
wants to merge 1 commit into
base: main
Choose a base branch
from

Conversation

lvhan028
Copy link
Collaborator
@lvhan028 lvhan028 commented Jul 2, 2024
Traceback (most recent call last):
  File "profile_generation.py", line 482, in <module>
    main()
  File "profile_generation.py", line 435, in main
    output = _process_map(profile_target, (args.model_path, ))
  File "profile_generation.py", line 377, in _process_map
    raise ret
IndexError: index 0 is out of bounds for axis 0 with size 0

@zhulinJulia24
Copy link
Collaborator

(lmdeploy) [zhulin1@SH-IDC1-10-140-0-187 lmdeployLvhan]$ python benchmark/profile_generation.py /nvme/qa_test_models/meta-llama/Meta-Llama-3-8B-Instruct --tp 1 -c 8 256 -ct 256 2048 -pt 1
Traceback (most recent call last):
  File "/home/zhulin1/lmdeployLvhan/benchmark/profile_generation.py", line 492, in <module>
    main()
  File "/home/zhulin1/lmdeployLvhan/benchmark/profile_generation.py", line 396, in main
    assert len(args.prompt_tokens) == len(args.completion_tokens), \
AssertionError: mismatched size between `prompt-tokens` and `completion-tokenes`, 1 vs 2

]

throughput = np.round(token_latency_stats.size / elapsed_time, 2)
if output_seqlen > 1:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we need to support this? Prompts with output_len<4 were filtered.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

3 participants