-
Notifications
You must be signed in to change notification settings - Fork 5
Issues
is:issue state:open
is:issue state:open
Issue creation is restricted in this repository
Search results
Consider query-only test time training
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.Consider modern RoPE variants
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.Generalization of SmartInitialLastRecentlyInsertedKVCache: Cater for left-padding
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.Dashboards for training and evaluation loss
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.Test comparing attention weights fails for FlashInfer if
q_len=1bugSomething isn't workingSomething isn't workingStatus: Open.Implement LoRA improvements
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.Create auto-tuning script to find best parameter combinations
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.Roll back small number of KV cache updates
enhancementNew feature or requestNew feature or requestStatus: Open.Backward CPU offloading: Asynchronous transfer
enhancementNew feature or requestNew feature or requestStatus: Open.Fix bug with autograd hooks and old training replay cache
bugSomething isn't workingSomething isn't workingStatus: Open.Configuration of fine-tuning scripts by YAML file
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.KV cache seems not reset before applied to new sequence
bugSomething isn't workingSomething isn't workingStatus: Open.