Skip to content

Release v1.3.2

Choose a tag to compare

View all tags
@github-actions github-actions released this 07 Sep 23:15
· 37 commits to main since this release
v1.3.2
424cd5d
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

What's Changed

  • Fix(seq2seq): honor load_dtype in the decode KV cache (bf16 dtype mismatch) by @IMvision12 in #422
  • fix(generation): honor load_dtype in seq2seq KV cache; compile+cache the multimodal decode path by @IMvision12 in #423
  • resolve model_type "llama" to the LlamaModel superset + disambiguate tokenizer by vocab by @IMvision12 in #424
  • remove dead ambiguity guards (raise gemma4/qwen3_5, default depth_anything/segformer) by @IMvision12 in #425
  • HF_MODEL_TYPE siglip -> siglip2 so its own repos load by @IMvision12 in #426
  • Remove old config for on fly models by @IMvision12 in #427
  • Fix group_size, strict expert layouts, reversible dequantize by @IMvision12 in #428
  • fix failing owl vit tests by @IMvision12 in #429
  • fix-attn by @IMvision12 in #430
  • release 1.3.2 by @IMvision12 in #431

Full Changelog: v1.3.1...v1.3.2

Contributors

IMvision12
Assets 2
Loading