Commits

Breadcrumbs

History for

onmain

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 27, 2023

• Add links to subfolders

  

  Nebuly

  authored
  6b79d86

  Copy full SHA for 6b79d86

• Merge pull request #305 from nebuly-ai/PierpaoloSorbellini-patch-2

  Show description for b841e22
  diegofiori

  authored
  b841e22

  Copy full SHA for b841e22

• Update apps/accelerate/chatllama/README.md

  Show description for 91441e5
  

  PierpaoloSorbellini

  and

  diegofiori

  authored
  91441e5

  Copy full SHA for 91441e5

• Update config.yaml

  PierpaoloSorbellini

  authored
  1d0b238

  Copy full SHA for 1d0b238

• Add peft_config.yaml description

  PierpaoloSorbellini

  authored
  429008e

  Copy full SHA for 429008e

• Release ChatLLaMA 0.0.4

  

  diegofiori

  committed
  156fa19

  Copy full SHA for 156fa19

• Add Peft

  

  diegofiori

  committed
  7a3b288

  Copy full SHA for 7a3b288

Commits on Mar 26, 2023

• Add tokenizer path to load_tokenizer for llama models

  PierpaoloSorbellini

  committed
  1f19e73

  Copy full SHA for 1f19e73

Commits on Mar 24, 2023

• Fix bug when diffusers is not installed (#294)

  valeriosofi

  authored
  d161fed

  Copy full SHA for d161fed

Commits on Mar 23, 2023

• Update chatllama version

  

  diegofiori

  committed
  343e0b4

  Copy full SHA for 343e0b4

• Set lr value for actor

  diegofiori

  authored
  5854da8

  Copy full SHA for 5854da8

• Set padding side to left

  diegofiori

  authored
  14382c3

  Copy full SHA for 14382c3

• Add discounted rewards coefficient to config

  PierpaoloSorbellini

  committed
  7a5c54f

  Copy full SHA for 7a5c54f

• Add fixes from the comunity that have not produced a PR

  PierpaoloSorbellini

  committed
  5ff1c67

  Copy full SHA for 5ff1c67

• Merge branch 'rlhf_accelerate' of https://github.com/nebuly-ai/nebullvm into rlhf_accelerate

  PierpaoloSorbellini

  committed
  f8fcb57

  Copy full SHA for f8fcb57

• Merge branch 'main' into rlhf_accelerate

  PierpaoloSorbellini

  committed
  e066feb

  Copy full SHA for e066feb

• Fix comments pre-release

  PierpaoloSorbellini

  committed
  dedf8cf

  Copy full SHA for dedf8cf

• fix: Fix typo of discounted_rewards.shape[0] to discounted_rewards.shape[1] in PPO loss

  mountinyy

  committed
  d083816

  Copy full SHA for d083816

Commits on Mar 21, 2023

• Update docs

  valeriosofi

  committed
  04bf989

  Copy full SHA for 04bf989

• update speedster version

  

  diegofiori

  committed
  0db45c5

  Copy full SHA for 0db45c5

• Merge branch 'main' into stable_diffusion

  valeriosofi

  authored
  c5ff3a4

  Copy full SHA for c5ff3a4

• Update docs

  valeriosofi

  committed
  85f960c

  Copy full SHA for 85f960c

• Merge pull request #269 from mfumanelli/fix_typos

  Show description for bd38730
  diegofiori

  authored
  bd38730

  Copy full SHA for bd38730

Commits on Mar 20, 2023

• Fix LLaMA with deepspeed single GPU

  PierpaoloSorbellini

  committed
  8ccfadb

  Copy full SHA for 8ccfadb

• Fix Deepspeed initialization and checkpoints single GPU

  

  Ubuntu

  committed
  5843195

  Copy full SHA for 5843195

• Add minor fix to deepspeed on reward model

  PierpaoloSorbellini

  committed
  f0f12b8

  Copy full SHA for f0f12b8

• Fix start_step value after resuming in between two epochs in actor and reward models training

  

  Ubuntu

  committed
  47c9176

  Copy full SHA for 47c9176

• Fix a bug in generate() of chatllama, add template to Standford dataset and update model loader with a method to get the path for training statistics

  PierpaoloSorbellini

  committed
  d3a09d4

  Copy full SHA for d3a09d4

Commits on Mar 18, 2023

• Remove last n checkpoint

  PierpaoloSorbellini

  committed
  10ca238

  Copy full SHA for 10ca238

Commits on Mar 17, 2023

• Add training stats logging

  PierpaoloSorbellini

  committed
  7c2cd43

  Copy full SHA for 7c2cd43

• Adjust for Anthropic dataset training, added training stats logging

  PierpaoloSorbellini

  committed
  48284ce

  Copy full SHA for 48284ce

• Merge branch 'main' into rlhf_accelerate

  PierpaoloSorbellini

  committed
  960f26d

  Copy full SHA for 960f26d

• Add indentation in json files

  PierpaoloSorbellini

  committed
  06cb02d

  Copy full SHA for 06cb02d

• Fix actor_rl save path

  PierpaoloSorbellini

  committed
  e44034b

  Copy full SHA for e44034b

• Add discounted rewards in RLHF

  PierpaoloSorbellini

  committed
  4f0fdc7

  Copy full SHA for 4f0fdc7

Pagination

Previous

Next