Skip to content

Infinity v3.0.0-alpha.9

Pre-release
Pre-release

Choose a tag to compare

View all tags
@michaelfeil michaelfeil released this 20 Oct 20:16
v3.0.0-alpha.9
e376a99
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Changes since alpha.8:

  • Enable opt-in dynamic FP8 for dense Qwen3.5 models, including qwen3_5 and qwen3_5_text. Use CUDA BF16 with --enable-fp8-dynamic; INFINITY_TORCH_FP8_SCOPE=mlp selects MLP-only quantization (#735).
  • Accept criteria: {} for OneJev yes/no (noul) questions, using the default yes/no descriptions (#734).
  • Honor modeling_overrides decision attention metadata and queue multimodal downloads instead of immediately rejecting requests when download slots are occupied (#733).

CPU and NVIDIA image builds run asynchronously; the tags become available after publication:

docker pull ghcr.io/michaelfeil/infinity_v3:nvidia-3.0.0-alpha.9
docker pull ghcr.io/michaelfeil/infinity_v3:cpu-3.0.0-alpha.9

FP8 changes numerical outputs and remains opt-in. Local H100 tests passed; PR #735 was merged with admin override while remote CI and review were still pending.

Image builds · Licensing · Full changelog

Assets 2
Loading