-
Notifications
You must be signed in to change notification settings - Fork 489
Pull requests: bigscience-workshop/petals
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Update peft to 0.5.0 version
#475
by artek0chumak
was merged Aug 23, 2023
Loading…
updated Aug 23, 2023
Support loading weights from Safetensors on server
#473
by borzunov
was merged Aug 22, 2023
Loading…
updated Aug 22, 2023
Change transformers version assert
#472
by justheuristic
was merged Aug 22, 2023
Loading…
updated Aug 22, 2023
Support transformers 4.32.x
#471
by justheuristic
was merged Aug 22, 2023
Loading…
updated Aug 22, 2023
Temporarily require peft<0.5.0, transformers<4.32.0
#470
by justheuristic
was merged Aug 22, 2023
Loading…
updated Aug 22, 2023
Make client compatible with transformers' GenerationMixin
#464
by borzunov
was merged Aug 20, 2023
Loading…
updated Aug 20, 2023
Update beam search algorithm with vectorized version
#109
by artek0chumak
was closed Aug 15, 2023
Loading…
updated Aug 15, 2023
Provide attention mask into the transformer layers
#206
by artek0chumak
was closed Aug 15, 2023
Loading…
updated Aug 15, 2023
Move SequenceManagerConfig -> ClientConfig, petals.dht_utils -> petals.utils.dht
#463
by borzunov
was merged Aug 14, 2023
Loading…
updated Aug 14, 2023
Add customizable input tensors
#445
by artek0chumak
was merged Aug 14, 2023
Loading…
updated Aug 14, 2023
Prioritize short inference, unmerge pools for long inference
#458
by borzunov
was merged Aug 11, 2023
Loading…
updated Aug 11, 2023
Use torch.cuda.synchronize for compute throughput
#456
by justheuristic
was merged Aug 9, 2023
Loading…
updated Aug 9, 2023
benchmarks: Aggregate speed among workers, set default dtype torch32
#454
by borzunov
was merged Aug 9, 2023
Loading…
updated Aug 9, 2023
Add LLaMA support
#323
by borzunov
was merged Jun 23, 2023
Loading…
updated Aug 8, 2023
8 of 11 tasks
Test Llama, rebalancing, throughput eval, and all CLI scripts
#452
by borzunov
was merged Aug 8, 2023
Loading…
updated Aug 8, 2023
Force using --new_swarm instead of empty --initial_peers
#451
by borzunov
was merged Aug 8, 2023
Loading…
updated Aug 8, 2023
Support loading blocks in 4-bit (QLoRA NF4 format)
#333
by borzunov
was merged Jul 3, 2023
Loading…
updated Aug 7, 2023
Prefer longer servers for fine-tuning, exclude unreachable
#448
by borzunov
was merged Aug 7, 2023
Loading…
updated Aug 7, 2023
Override float32 in config to bfloat16
#431
by borzunov
was merged Aug 7, 2023
Loading…
updated Aug 7, 2023
2 tasks
ProTip!
Type g p on any issue or pull request to go back to the pull request listing page.