mistralai/Devstral-Small-2505
Full fine-tune (fewer layers) vs mistralai/Mistral-Small-3.1-24B-Instruct-2503. mistralai/Devstral-Small-2505 matches part of mistralai/Mistral-Small-3.1-24B-Instruct-2503 (363 of its 585 tensors; the rest of mistralai/Mistral-Small-3.1-24B-Instruct-2503 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 10/81, MLP 120/120, Attention 160/160, lm_head 1/1
21 derivatives checked
- lmstudio-community/Devstral-Small-2505-MLX-4bit: Quantization
- unsloth/Devstral-Small-2505-GGUF: Same weights, converted
- bartowski/mistralai_Devstral-Small-2505-GGUF: Same weights, converted
- Mungert/Devstral-Small-2505-GGUF: Quantization
- mistralai/Devstral-Small-2505_gguf: Same weights, converted
- lmstudio-community/Devstral-Small-2505-GGUF: Quantization
- eaddario/Devstral-Small-2505-pruned-GGUF: Same weights, converted
- eaddario/Devstral-Small-2505-GGUF: Same weights, converted
- jkim96/Devstral-Small-2505-DASHQ-Q2-GGUF: Quantization
- mlx-community/Devstral-Small-2505-4bit: Quantization
- DevQuasar/unsloth.Devstral-Small-2505-GGUF: Same weights, converted
- unsloth/Devstral-Small-2505-unsloth-bnb-4bit: Quantization
- lmstudio-community/Devstral-Small-2505-MLX-6bit: Quantization
- mlx-community/Devstral-Small-2505-8bit: Quantization
- mlx-community/Devstral-Small-2505-3bit: Quantization
- unsloth/Devstral-Small-2505: Identical copy
- ngxson/Devstral-Small-Vision-2505-GGUF: Quantization
- mlx-community/Devstral-Small-2505-6bit: Quantization
- mlx-community/Devstral-Small-2505-bf16: Same weights, repacked
- mratsim/Devstral-Small-2505.w4a16-gptq: Quantization
- nesilabs/tosilos-24b: Full fine-tune