mistralai/Voxtral-Small-24B-2507
Full fine-tune (fewer layers) vs mistralai/Mistral-Small-24B-Base-2501. mistralai/Mistral-Small-24B-Base-2501 matches part of mistralai/Voxtral-Small-24B-2507 (363 of its 852 tensors; the rest of mistralai/Voxtral-Small-24B-2507 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 40 of 40 layers.
Tensors changed: Embeddings 1/1, Norms 81/81, MLP 120/120, Attention 160/160, lm_head 1/1
11 derivatives checked
- handy-computer/Voxtral-Small-24B-2507-gguf: Not comparable yet
- bartowski/mistralai_Voxtral-Small-24B-2507-GGUF: Same weights, converted (fewer layers)
- mradermacher/Voxtral-Small-24B-2507-i1-GGUF: Quantization (fewer layers)
- VincentGOURBIN/voxtral-small-4bit-mixed: Quantization
- Abu-Dju/mistralai_Voxtral-Small-24B-2507-GGUF: Same weights, converted (fewer layers)
- mradermacher/Voxtral-Small-24B-2507-GGUF: Quantization (fewer layers)
- RedHatAI/Voxtral-Small-24B-2507-FP8-dynamic: Quantization
- kukidevalml/mistralai_Voxtral-Small-24B-2507-GGUF: Same weights, converted (fewer layers)
- MarkusKaemmerer/Voxtral-Small-24B-2507-4bit-dense-encoder: Quantization
- MarkusKaemmerer/Voxtral-Small-24B-2507-8bit-dense-encoder: Quantization
- stduhpf/Voxtral-Small-24B-2507-GGUF: Quantization (fewer layers)