mradermacher/Gemma3-4B-CoT-GGUF

Same weights, converted (fewer layers) vs DrRiceIO7/Gemma3-4B-CoT. mradermacher/Gemma3-4B-CoT-GGUF matches part of DrRiceIO7/Gemma3-4B-CoT (443 of its 883 tensors; the rest of DrRiceIO7/Gemma3-4B-CoT has no counterpart). On the shared part: Same weights stored in another dtype (BF16 → F16/F32).

Tensors changed: Norms 0/205, MLP 0/102, Attention 0/136

On the Hub · Downloads on Model Pulse