thomsonreuters/Thomson-1.0-Small
Full fine-tune vs tri-fair-lab/Snowdon1.1-Small. Fine-tune: 99% of the weights changed, across 40 of 40 layers.
Tensors changed: Embeddings 1/4, Attention 279/388, MLP 280/392, Norms 131/241, lm_head 1/1
9 derivatives checked
- bartowski/thomsonreuters_Thomson-1.0-Small-GGUF: Same weights, converted
- IsValorum/Thomson-1.0-Small-APEX-I-MiniPlus-V2.1-GGUF: Quantization
- IsValorum/Thomson-1.0-Small-APEX-I-MiniPlus-V2-GGUF: Quantization
- mradermacher/Thomson-1.0-Small-i1-GGUF: Quantization
- mradermacher/Thomson-1.0-Small-GGUF: Quantization
- hermitdave/Thomson-1.0-Small-MLX-6bit: Quantization
- McG-221/Thomson-1.0-Small-mlx-8Bit: Quantization
- hermitdave/Thomson-1.0-Small-MLX-4bit: Quantization
- ProprietaryLegal/Thomson-1.0-Small-NVFP4: Quantization
1 other repos hold byte-for-byte identical weight files.