-
Notifications
You must be signed in to change notification settings - Fork 485
Issues
is:issue state:open
is:issue state:open
Issue creation is restricted in this repository
Search results
Division by Zero in GPTRewardModel with Empty Batches
bugSomething isn't workingSomething isn't workingStatus: Open.#609 In CarperAI/trlx;ValueError: Invalid pattern: '**' can only be an entire path component
bugSomething isn't workingSomething isn't workingStatus: Open.#607 In CarperAI/trlx;ImportError: cannot import name 'prepare_model_for_int8_training' from 'peft'
bugSomething isn't workingSomething isn't workingStatus: Open.#606 In CarperAI/trlx;Extracting total-loss, PPO-loss, rewards per step, returns per step in RLHF-PPO implementation
documentationImprovements or additions to documentationImprovements or additions to documentationStatus: Open.#605 In CarperAI/trlx;- Status: Open.#604 In CarperAI/trlx;
Does the framework support PPO training for Qwen2?
feature requestNew feature or requestNew feature or requestStatus: Open.#603 In CarperAI/trlx;- Status: Open.#602 In CarperAI/trlx;
OOM error with PEFT LoRA on Llama2-7B
bugSomething isn't workingSomething isn't workingStatus: Open.#601 In CarperAI/trlx;Load the checkpoint fails
bugSomething isn't workingSomething isn't workingStatus: Open.#600 In CarperAI/trlx;cannot import name 'flatten_dataclass' from 'trlx.data.ilql_types'
bugSomething isn't workingSomething isn't workingStatus: Open.#599 In CarperAI/trlx;maybe bug in prepare & load's order
bugSomething isn't workingSomething isn't workingStatus: Open.#598 In CarperAI/trlx;Error when running Ray Tune to launch hyperparameter sweep
bugSomething isn't workingSomething isn't workingStatus: Open.#597 In CarperAI/trlx;