tencent/ContextPilot-E4B
Full fine-tune vs google/gemma-4-E4B-it. Fine-tune: 94% of the weights changed, across 42 of 42 layers.
Tensors changed: MLP 126/606, Norms 129/484, other 85/261, Attention 132/719, Embeddings 2/6
2 derivatives checked
- mradermacher/ContextPilot-E4B-i1-GGUF: Quantization (fewer layers)
- mradermacher/ContextPilot-E4B-GGUF: Same weights, converted (fewer layers)
2 other repos hold byte-for-byte identical weight files.