pytorch / pytorch/executorch

[LLM runner] Correct the deprecated seq_len CLI help text

Open
#22,681 0 comments 0 reactions 2 assignees View on GitHub

@mergennachin is already working on this.

Since Sep 10, 2026.

bug module: llm
Dominant language
Python
Stars
5k
Forks
1.2k
Avg merge
2d 10h
Merged PRs (30d)
581

Description

Problem

The llama runner help for --seq_len says to use max_seq_len, but there is no corresponding CLI flag. The implementation falls back from --max_new_tokens to --seq_len, and its runtime log tells users to prefer --max_new_tokens.

This mismatch was found while validating LLM documentation in #22620. The C++ source change is intentionally excluded from that documentation-only PR.

Proposed change

Update the --seq_len help text to recommend --max_new_tokens and clearly distinguish total generation length from new-token count and the model context limit.

Test plan

Build the llama runner and inspect --help; run existing llama runner CI.

cc @larryliu0820 @cccclai @helunwencser @jackzhxng @digantdesai

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.