Skip to content

Pull requests: bigscience-workshop/petals

New pull request New
Clear current search query, filters, and sorts
16 Open 382 Closed
16 Open 382 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Spam less in server logs
#350 by borzunov was merged Jul 13, 2023 Loading… updated Jul 13, 2023
1
Estimate adapter memory overhead in choose_num_blocks()
#346 by justheuristic was merged Jul 13, 2023 Loading… updated Jul 13, 2023
3 tasks
5
Support LLaMA repos without "-hf" suffix
#349 by borzunov was merged Jul 13, 2023 Loading… updated Jul 13, 2023
Fix Docker build by avoiding Python 3.11
#348 by borzunov was merged Jul 13, 2023 Loading… updated Jul 13, 2023
Import petals.utils.peft only when needed to avoid unnecessary import of bitsandbytes
#345 by borzunov was merged Jul 12, 2023 Loading… updated Jul 12, 2023
1
Mention LLaMA in readme
#344 by borzunov was merged Jul 12, 2023 Loading… updated Jul 12, 2023
Support peft LoRA adapters
#335 by artek0chumak was merged Jul 12, 2023 Loading… updated Jul 12, 2023
15
Use bitsandbytes 0.40.0.post4 with bias hotfix
#342 by borzunov was merged Jul 12, 2023 Loading… updated Jul 12, 2023
Use bnb==0.40.0.post4 to fix bias bug, use bfloat16 by default
#341 by borzunov was closed Jul 12, 2023 Loading… updated Jul 12, 2023
4
Delete deprecated petals.cli scripts
#336 by borzunov was merged Jul 11, 2023 Loading… updated Jul 11, 2023
Use 4-bit for llama by default, use bitsandbytes 0.40.0.post3
#340 by borzunov was merged Jul 11, 2023 Loading… updated Jul 11, 2023
Implement direct server-to-server communication
#331 by borzunov was merged Jul 11, 2023 Loading… updated Jul 11, 2023
2 tasks done
15
Allow free_disk_space_for() remove arbitrary files from Petals cache
#339 by borzunov was merged Jul 5, 2023 Loading… updated Jul 5, 2023
1
Require pydantic < 2.0 (2.0 is incompatible with hivemind 1.1.8)
#337 by borzunov was merged Jul 1, 2023 Loading… updated Jul 1, 2023
Fix warmup steps and minor issues in benchmarks
#334 by borzunov was merged Jun 30, 2023 Loading… updated Jun 30, 2023
Add benchmark scripts
#319 by borzunov was merged Jun 29, 2023 Loading… updated Jun 29, 2023
4
Show license links when loading models
#332 by borzunov was merged Jun 24, 2023 Loading… updated Jun 24, 2023
Fix llama's lm_head.weight.requires_grad
#330 by borzunov was merged Jun 23, 2023 Loading… updated Jun 23, 2023
Add AutoDistributed{Model, ModelForCausalLM, ModelForSequenceClassification}
#329 by borzunov was merged Jun 23, 2023 Loading… updated Jun 23, 2023
Use number of tokens for attn_cache_size
#286 by mryab was merged Jun 17, 2023 Loading… updated Jun 17, 2023
1
Determine block dtype in a unified manner
#325 by mryab was merged Jun 16, 2023 Loading… updated Jun 16, 2023
3
Remove unused imports and attributes
#324 by mryab was merged Jun 10, 2023 Loading… updated Jun 10, 2023
1
Bump version to 1.1.5
#312 by borzunov was merged May 9, 2023 Loading… updated May 9, 2023
Abort speedtest if it runs too long
#316 by borzunov was merged May 9, 2023 Loading… updated May 9, 2023
Raise error for unexpected .generate() kwargs
#315 by borzunov was merged May 9, 2023 Loading… updated May 9, 2023
ProTip! Add no:assignee to see everything that’s not assigned.