Skip to content

Issues: EleutherAI/lm-evaluation-harness

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.

Sign up for GitHub

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

315 Open 839 Closed
315 Open 839 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Is there a good method to set the generation_kwargs for tasks so that different kv caches can be used for optimization?
#2506 opened Nov 19, 2024 by CoderChen01
2
Load dataset error
#2505 opened Nov 19, 2024 by junming-yang
french_bench_xnli dataset doesn't exist asking questions For asking for clarification / support on library usage.
#2501 opened Nov 18, 2024 by jgcb00
1
[Performance significantly drop when increase the batch_size] bug Something isn't working.
#2498 opened Nov 17, 2024 by yushengsu-thu
@baberabb
1
OOM Issues in MMLU Evaluation with lm_eval Using vllm as Backend
#2490 opened Nov 14, 2024 by wchen61
Overwrite default tasks
#2487 opened Nov 13, 2024 by jonoillar
Cannot reproduce LLaMA 3 8B on hendrycks_math validation For validation of task implementations.
#2479 opened Nov 11, 2024 by liuxiaozhu01
1
Why Different Versions Make a Big Difference in HellaSwag zero-shot validation For validation of task implementations.
#2478 opened Nov 11, 2024 by cquxl
1
MMLU-Pro Generation -> Generative
#2472 opened Nov 8, 2024 by rimashahbazyan
[QUESTION] Does lm-eval support direct evaluation from json?
#2470 opened Nov 8, 2024 by AaronZLT
Issue with Perplexity Score Using max_length > 1024 in lm-evaluation-harness
#2467 opened Nov 7, 2024 by fnusid
task load return error
#2466 opened Nov 7, 2024 by pod2c
1
Evaluate local tasks failed when using lm-eval --tasks <local-folder>
#2463 opened Nov 7, 2024 by ChenXiaoTemp
auto-detect batchsize finding too large a batchsize to fit in VRAM at the end when used with multi-gpu bug Something isn't working.
#2458 opened Nov 5, 2024 by SmerkyG
how to compute tasks' metrics by their sub-tasks
#2448 opened Oct 31, 2024 by xiaobo-Chen
1
Why is using vLLM via lm-eval-harness slower than using vLLM directly? asking questions For asking for clarification / support on library usage.
#2445 opened Oct 30, 2024 by WuXnkris
2
Wrong format of the few-shot examples in mgsm_direct tasks good first issue Good for newcomers validation For validation of task implementations.
#2444 opened Oct 30, 2024 by zxcvuser
Improve preprocessing for paws-x and xnli tasks feature request A feature that isn't implemented yet. good first issue Good for newcomers
#2442 opened Oct 30, 2024 by zxcvuser
vllm with tensor_parallel_mode is not working at all because of multiprocessing problem
#2431 opened Oct 28, 2024 by 95jinchul
1
Task winogrande does not work in 0-shot setting together with --apply_chat_template
#2430 opened Oct 27, 2024 by ArtemBiliksin
GPU with GGFU LLM
#2429 opened Oct 25, 2024 by Znbne
1
Llama3.1-8B-Instruct evaluation fails asking questions For asking for clarification / support on library usage.
#2428 opened Oct 25, 2024 by Isaaclgz
1
test speculative decode accuracy asking questions For asking for clarification / support on library usage.
#2424 opened Oct 24, 2024 by baoqianmagik
5
Question related to how to use the validation and training splits. asking questions For asking for clarification / support on library usage.
#2423 opened Oct 24, 2024 by sorobedio
3
bbh_zeroshot fails during to a custom filter issue. bug Something isn't working.
#2422 opened Oct 23, 2024 by shamanez
ProTip! Type g i on any issue or pull request to go back to the issue listing page.