HuggingFaceTB/SmolVLM-Base
Full fine-tune (fewer layers) vs HuggingFaceTB/SmolLM2-1.7B-Instruct. HuggingFaceTB/SmolLM2-1.7B-Instruct matches part of HuggingFaceTB/SmolVLM-Base (217 of its 657 tensors; the rest of HuggingFaceTB/SmolVLM-Base has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Norms 49/49, MLP 72/72, Attention 96/96
3 derivatives checked
- HuggingFaceTB/SmolVLM2-2.2B-Base: Full fine-tune
- mradermacher/SmolVLM-Base-i1-GGUF: Quantization (fewer layers)
- mradermacher/SmolVLM-Base-GGUF: Same weights, converted (fewer layers)