From 2035e75040abfb6ec3125191ee544f3a2bc0823a Mon Sep 17 00:00:00 2001 From: Yinjie Wang Date: Tue, 21 Apr 2026 00:15:33 -0500 Subject: [PATCH] Update README.md --- toolcall-rl/README.md | 7 +++++++ 1 file changed, 7 insertions(+) diff --git a/toolcall-rl/README.md b/toolcall-rl/README.md index b5a2bbf..a41bb03 100644 --- a/toolcall-rl/README.md +++ b/toolcall-rl/README.md @@ -82,6 +82,13 @@ PYTHONPATH=/path/to/Megatron-LM python tools/convert_hf_to_torch_dist.py \ --rotary-base 5000000 \ --save /path/to/qwen3-4b-sft_torch_dist +# or for qwen3.5: +source scripts/models/qwen3-4B.sh +PYTHONPATH=/path/to/Megatron-LM python tools/convert_hf_to_torch_dist.py \ + ${MODEL_ARGS[@]} \ + --hf-checkpoint /path/to/qwen3.5-4b \ + --save /path/to/qwen3.5-4b-torch_dist + # Download RL data and run huggingface-cli download --repo-type dataset BytedTsinghua-SIA/DAPO-Math-17k --local-dir /path/to/dapo-math-17k huggingface-cli download --repo-type dataset zhuzilin/aime-2024 --local-dir /path/to/aime-2024