6 Pull requests merged by 4 people

• Fix Mistral conversion script

  #2306 merged Jun 23, 2025

• Update preset versions for Mixtral,Qwen-MoE and Mistral models

  #2307 merged Jun 23, 2025

• Fix Gemma3InterleaveEmbeddings JAX inference error by ensuring indices are int32

  #2305 merged Jun 19, 2025

• Modify TransformerEncoder masking documentation

  #2297 merged Jun 18, 2025

• Bump the python group with 2 updates

  #2282 merged Jun 18, 2025

• Incorrect condition on sliding_window_size

  #2289 merged Jun 18, 2025

1 Pull request opened by 1 person

• [OpenVINO backend] supporting inference for gemma with ov backend

  #2310 opened Jun 22, 2025

2 Issues closed by 2 people

• Multimodal Gemma 3 models fail to run inference using JAX backend

  #2304 closed Jun 19, 2025

• TransformerEncoder silently ignores mask_zero=True despite docs claim

  #2220 closed Jun 18, 2025

11 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Safetensors conversion

  #2290 commented on Jun 23, 2025 • 13 new comments

• [WIP] PARSeq Model

  #2089 commented on Jun 23, 2025 • 5 new comments

• update python to 3.10 and Keras to 3.6

  #2292 commented on Jun 18, 2025 • 1 new comment

• Add StableLM-3B 4E1T to Keras Hub

  #2151 commented on Jun 17, 2025 • 0 new comments

• WIP [DeepSeek R1] Add DeepSeekV3 Base + Weight Conversion

  #2171 commented on Jun 22, 2025 • 0 new comments

• Add return_attention_scores support to CachedMultiHeadAttention

  #2213 commented on Jun 24, 2025 • 0 new comments

• [DeepSeek R1] Qwen2.5 Distillations

  #2236 commented on Jun 22, 2025 • 0 new comments

• Add Esm

  #2244 commented on Jun 17, 2025 • 0 new comments

• Loosen class requirements in from_preset

  #2276 commented on Jun 18, 2025 • 0 new comments

• Add HGNetV2 to KerasHub

  #2293 commented on Jun 23, 2025 • 0 new comments

• Add Swin Transformer Backbone

  #2298 commented on Jun 18, 2025 • 0 new comments