trillionlabs/Trida2.0-4B
Full fine-tune (fewer layers) vs Qwen/Qwen3.5-4B. trillionlabs/Trida2.0-4B matches part of Qwen/Qwen3.5-4B (425 of its 738 tensors; the rest of Qwen/Qwen3.5-4B has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 32 of 32 layers.
Tensors changed: MLP 96/96, Attention 179/224, Norms 57/105
2 derivatives checked
- mradermacher/Trida2.0-4B-i1-GGUF: Quantization
- mradermacher/Trida2.0-4B-GGUF: Same weights, converted