-
Notifications
You must be signed in to change notification settings - Fork 2
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
- Status: Open.#23 In Selectus2/trainers-rb;
Icebox: 0.4.0 ergonomics — do not start until there is evidence
blockedWaiting on another issue — do not startWaiting on another issue — do not startenhancementNew feature or requestNew feature or requestStatus: Open.LoRA runs appear to checkpoint the full state dict rather than the adapters
blockedWaiting on another issue — do not startWaiting on another issue — do not startbugSomething isn't workingSomething isn't workingStatus: Open.Checkpoints cannot resume training
enhancementNew feature or requestNew feature or requestmaintainerChanges numerical semantics or public API — discuss before PRChanges numerical semantics or public API — discuss before PRStatus: Open.A saved model cannot be loaded back — no config.json is written
bugSomething isn't workingSomething isn't workingcriticalProduces incorrect results or unusable outputProduces incorrect results or unusable outputmaintainerChanges numerical semantics or public API — discuss before PRChanges numerical semantics or public API — discuss before PRStatus: Open.Add an error hierarchy; save_lora_adapters returns nil where it should raise
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.The library writes to stdout; add Trainers.logger
enhancementNew feature or requestNew feature or requestgood first issueGood for newcomersGood for newcomersStatus: Open.dataloader_drop_last and label_names are accepted and ignored
bugSomething isn't workingSomething isn't workinggood first issueGood for newcomersGood for newcomersStatus: Open.seed: is accepted, stored, and never applied
bugSomething isn't workingSomething isn't workingmaintainerChanges numerical semantics or public API — discuss before PRChanges numerical semantics or public API — discuss before PRStatus: Open.LR schedule is wrong whenever gradient_accumulation_steps > 1
bugSomething isn't workingSomething isn't workingcriticalProduces incorrect results or unusable outputProduces incorrect results or unusable outputmaintainerChanges numerical semantics or public API — discuss before PRChanges numerical semantics or public API — discuss before PRStatus: Open.Repo hygiene: SECURITY, CONTRIBUTING, README badges, description and topics
docsDocumentationDocumentationgood first issueGood for newcomersGood for newcomersStatus: Open.Add a public API surface test
help wantedExtra attention is neededExtra attention is neededtestTest coverageTest coverageStatus: Open.