NVIDIA-NeMo / NVIDIA-NeMo/RL

nightly test fail after mbridge bump in #3906

Open
#3,946 0 comments 2 reactions 1 assignee Claimed by @terrykong View on GitHub
Automation bug
Dominant language
Python
Stars
2k
Forks
561
Avg merge
4d 5h
Merged PRs (30d)
145

Description

should be fixed in #3906, not validated yet:
1. `grpo-nemotron3-super-120BA12B-16n8g-megatron.sh`
2. `grpo-nanov3-30BA3B-2n8g-megatron-pack-cp.sh`
3. `grpo-nano-v2-12b-1n8g-megatron.sh`

should be fixed in #3914:
1. `vlm_grpo-nemotron-omni-30ba3b-mmpr-4n8g-megatron-tp8ep16.v1.sh`
2. `vlm_grpo-nemotron-omni-30ba3b-clevr-1n8g-megatron-tp8ep8.v1.sh`

should be fixed in https://github.com/NVIDIA/Megatron-LM/pull/6993:
1. `grpo-qwen2.5-math-1.5b-instruct-1n8g-megatron_generation-noncolocated-single-controller-sync.sh`

unknown:
1. `grpo-moonlight-16ba3b-4n8g-megatron-tq_simple.sh`
2. `grpo-moonlight-16ba3b-4n8g-megatron.sh`
3. `vlm_grpo-qwen3-omni-30ba3b-audiomcq-4n8g-megatron.v1.sh`
4. `vlm_grpo-qwen2.5-omni-7b-intent-1n8g-megatron.v1.sh`

pre-existed:
1. `grpo-dapomath17k-dsv3-megatron.sh`
2. `grpo-moonlight-16ba3b-4n8g-megatron-fp8-e2e.sh`

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.