diff --git a/autotest/config.yaml b/autotest/config.yaml index a8563d3be..8b46ae111 100644 --- a/autotest/config.yaml +++ b/autotest/config.yaml @@ -201,9 +201,9 @@ case: base_metric: qwen3-sft/20251117105949/tracker.jsonl check_metrics: grad_norm: 1 - loss/maxvio: 0.000001 - loss/local_loss: 0.000001 - loss/reduced_balancing_loss: 0.000001 + loss/maxvio: 0.05 + loss/local_loss: 0.02 + loss/reduced_balancing_loss: 0.02 loss/reduced_llm_loss: 0.02 lr: 0 timeout: 10800 @@ -587,12 +587,12 @@ case: operator: < - metric: mismatch/mismatch_k3_kl - threshold: 0.0001 + threshold: 0.0002 method: absolute operator: <= - metric: response/response_len/mean - threshold: 0.12 + threshold: 0.25 method: relative operator: < - diff --git a/autotest/config/qwen3_5_35B_sft_vl.py b/autotest/config/qwen3_5_35B_sft_vl.py index e59d8bb70..50b72822e 100644 --- a/autotest/config/qwen3_5_35B_sft_vl.py +++ b/autotest/config/qwen3_5_35B_sft_vl.py @@ -18,7 +18,7 @@ optim_cfg = AdamWConfig(lr=6e-05) lr_cfg = LRConfig(lr_type="cosine", lr_min=1e-6) fsdp_cfg = FSDPConfig( - torch_compile=True, + torch_compile=False, cpu_offload=False, )