Search results

Running ppo with accelerate fails.

⚡accelerateRelated to accelerateRelated to accelerate

🏋 PPORelated to PPORelated to PPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2792 In huggingface/trl;

· ruiqi-zhong opened on Feb 7, 2025

GRPOTrainer in the current main branch doesn't work (v0.14.0 works)

⏳ needs more infoAdditional information or clarification is required to proceedAdditional information or clarification is required to proceed

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2791 In huggingface/trl;

· zhengqigao opened on Feb 7, 2025

vLLM doesn't estimate the model size properly

⚡ PEFTRelated to PEFTRelated to PEFT

⚡accelerateRelated to accelerateRelated to accelerate

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2788 In huggingface/trl;

· Superskyyy opened on Feb 6, 2025

Load/Savings Checkpoint Fails using DeepSpeed - GRPO

🐛 bugSomething isn't workingSomething isn't working

🚀 deepspeedRelated to deepspeedRelated to deepspeed

Status: Open.

#2787 In huggingface/trl;

· zaddy6 opened on Feb 6, 2025

Add vLLM support to LogCompletionsCallback

⚡accelerateRelated to accelerateRelated to accelerate

✨ enhancementNew feature or requestNew feature or request

🏋 DPORelated to DPORelated to DPO

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#2786 In huggingface/trl;

· tchang1997 opened on Feb 6, 2025

deepspeed zero stage 3 does not affect the speed and VRAM consumption at all

⚡accelerateRelated to accelerateRelated to accelerate

🏋 PPORelated to PPORelated to PPO

🚀 deepspeedRelated to deepspeedRelated to deepspeed

Status: Open.

#2784 In huggingface/trl;

· JohnConnor123 opened on Feb 6, 2025

NashMD trainer sampling policy wrong

⚡ PEFTRelated to PEFTRelated to PEFT

⚡accelerateRelated to accelerateRelated to accelerate

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2781 In huggingface/trl;

· zhourunlong opened on Feb 6, 2025

lora don't work! OOM

⚡ PEFTRelated to PEFTRelated to PEFT

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2780 In huggingface/trl;

· zhangguoxin1 opened on Feb 6, 2025

ORPOTrainer crashes due to pickling failure if dataloader_num_workers > 0

🏋 ORPORelated to ORPORelated to ORPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2779 In huggingface/trl;

· kiratp opened on Feb 6, 2025

Allow vllm sub-batching to avoid CUDA out of memory

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#2775 In huggingface/trl;

· cfpark00 opened on Feb 5, 2025

GRPO tests failing in multi-device setting on main

🏋 GRPORelated to GRPORelated to GRPO

🐛 bugSomething isn't workingSomething isn't working

Status: Open.

#2774 In huggingface/trl;

· tyler-romero opened on Feb 5, 2025

How to log more metrics with wandb when using GRPO trainer and accelerate

⚡accelerateRelated to accelerateRelated to accelerate

✨ enhancementNew feature or requestNew feature or request

🏋 GRPORelated to GRPORelated to GRPO

Status: Open.

#2768 In huggingface/trl;

· andrewsiah opened on Feb 5, 2025