impactframes/QWQCoder
Full fine-tune (fewer layers) vs Qwen/QwQ-32B-Preview. impactframes/QWQCoder matches part of Qwen/QwQ-32B-Preview (387 of its 771 tensors; the rest of Qwen/QwQ-32B-Preview has no counterpart). On the shared part: Fine-tune: 97% of the weights changed, across 32 of 32 layers.
Tensors changed: Embeddings 1/1, Norms 65/65, MLP 93/96, Attention 217/224, lm_head 1/1
2 derivatives checked
- mradermacher/QWQCoder-i1-GGUF: Quantization
- mradermacher/QWQCoder-GGUF: Quantization
1 other repos hold byte-for-byte identical weight files.