Tag
ZeroModels is a tool that enables running any AI model on multiple backends (JAX, PyTorch, TensorFlow) using Keras 3, with weights converted from original checkpoints and no runtime dependency on transformers or torch.
A Hugging Face repository hosting MiniMax-H3 models converted for ComfyUI usage, along with a Lightx2v distill LoRA for faster inference.
This paper studies what transfers between transformer models of different sizes in the same family (Pythia), showing that representations align while weights don't, and that conversion works best via initialization rather than direct weight projection.
This paper proposes functional reconstruction for converting MHA/GQA checkpoints into MLA draft models for speculative decoding, directly optimizing attention modules to preserve token acceptance. It reports consistent improvements across 192 configurations involving Llama/Qwen models and multiple conversion methods.
Converted the Qwen 3.6 35b a3b model to ROCmfp4 format, leveraging MTP benefits for improved performance on AMD hardware.
The author provides extracted GGUF files containing only MTP tensors for Qwen3.6 models, allowing users to graft tensors with a significantly reduced download size compared to full model files.