Skip to content

Latest commit

 

History

History
99 lines (71 loc) · 2.45 KB

CHANGELOG.md

File metadata and controls

99 lines (71 loc) · 2.45 KB

Changelog

All notable changes to this project will be documented in this file.
The format is based on Keep a Changelog,
and this project adheres to Semantic Versioning.

Unreleased

  • ...

[0.1.4] - 2025-02-27

Added

  • EG 07.23 — example usage of train_model_dpo_simple (#339)
  • train_model_dpo_simple (#338)
  • evaluate_dpo_loss_loader (#337)
  • EG 07.22 (#335)
  • compute_dpo_loss_loader (#334)
  • EG 07.21 - example usage of compute_dpo_loss_batch (#333)
  • dpo_loss_batch (#332)
  • compute_logprobs (#331)
  • compute_dpo_loss (#326)
  • preference data loader (#319)
  • [Fix] rejected/chosen masking & update EG 07.19 (#318)
  • EG 07.19 Example usage of PreferenceDataCollator (#314)
  • PreferenceDatasetCollator (#309)
  • PreferenceDataset (#308)
  • EncodedPreferenceExample and InstructionExample trait (#303)
  • generate_preference_dataset + EG 07.18 (#299)
  • Bonus DPO - Use Ollama to generate chosen/rejection response for an instruction entry + EG 07.17 (#294)

Changed

Exercise 7.4 + use of GPT trait in listings::ch05 instead of GPTModel (#289)

[0.1.3] - 2025-01-23

Added

  • docs Appendix E (#287)
  • EG E.07 (#286)
  • EG E.06 (#285)
  • Listing E.7 re-export of train_classifier_simple (#284)
  • EG E.05 (#283)
  • EG E.04 (#282)
  • EG E.03 (LoRA model loading) (#281)
  • GPTModelWithLoRA (#279)
  • TransformerBlockWithLoRA (#278)
  • FeedForwardWithLoRA (#277)
  • MultiHeadAttentionWithLoRA (#270)
  • Listing E.6 LinearWithLoRA (#269)
  • Listing E.5 (LoRALayer) (#268)
  • EG E.02 and set listing E.4 as re-export (#267)
  • Listing E.4 (#266)
  • Example E.01 (#265)
  • Listing E.3 (#264)
  • Listing E.2 (#263)
  • Listing E.1 (#262)

Changed

  • parametrize batch size (#285)
  • GPT trait to "consolidate" GPTModel and GPTModelWithLoRA (#279)
  • Rip out Sequential and SequentialT in favor of explicit sequential-like structs (#276)

[0.1.2] - 2025-01-13

Added

  • Exercise 7.3 (#252)

Fixed

  • [docs] make listings more visible for ch07 (#260)

[0.1.1] - 2025-01-11

Added

  • Exercise 7.2 (#248)

Changed

  • Make listings::ch07::InstructionDataBatcher more generic (#248)
  • Add associated type to listings::ch07::CustomCollator (#248)

Fixed

  • Incorrect cast of keep indices to u8 in calc_loss_loader (#250)
  • Missing ignore_index in calc_loss_loader fn params (#250)

[0.1.0] - 2025-01-09

Added

  • Listings ch02 up to ch07
  • Examples ch02 up to ch07
  • Exercise ch02 up to ch06 and Exercise 7.1