NousResearch/atropos
1
0
Fork 0
mirror of https://github.com/NousResearch/atropos.git synced 2026-04-23 16:54:56 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

math zero 32k

Browse source
This commit is contained in:
Jai Suphavadeeprasit 2026-02-12 14:39:32 -05:00
parent 74d46aaa76
commit 7d96367516
1 changed files with 3 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

3
example_trainer/scripts/compare_all_modes_math_zero.sh
View file

@ -250,6 +250,7 @@ CUDA_VISIBLE_DEVICES=$SHARED_GPU python -m example_trainer.grpo \
--batch-size $BATCH_SIZE \
--training-steps $TRAINING_STEPS \
--max-model-len $MAX_MODEL_LEN \
--seq-len $MAX_MODEL_LEN \
--save-path "$LOG_DIR/checkpoints_shared" \
$WANDB_ARGS --wandb-group "shared-vllm" \
--benchmark \
@ -281,6 +282,7 @@ CUDA_VISIBLE_DEVICES=$LORA_ONLY_GPU python -m example_trainer.grpo \
--batch-size $BATCH_SIZE \
--training-steps $TRAINING_STEPS \
--max-model-len $MAX_MODEL_LEN \
--seq-len $MAX_MODEL_LEN \
--lora-r 16 \
--lora-alpha 32 \
--vllm-restart-interval 5 \
@ -304,6 +306,7 @@ CUDA_VISIBLE_DEVICES=$LORA_RESTART_GPU python -m example_trainer.grpo \
--batch-size $BATCH_SIZE \
--training-steps $TRAINING_STEPS \
--max-model-len $MAX_MODEL_LEN \
--seq-len $MAX_MODEL_LEN \
--lora-r 16 \
--lora-alpha 32 \
--vllm-restart-interval 5 \