BSC-LT/salamandra-2b-instruct
Partial fine-tune vs BSC-LT/salamandra-2b. Partial change: 77% of the weights changed (attention, lm_head, mlp, norms), in 24 of 24 layers.
Tensors changed: lm_head 1/1, Embeddings 0/1, Norms 39/49, MLP 72/72, Attention 96/96
10 derivatives checked
- robbiemu/salamandra-2b-instruct: Full fine-tune
- mradermacher/salamandra-2b-instruct-GGUF: Same weights, converted
- mradermacher/salamandra-2b-instruct-i1-GGUF: Quantization
- hdnh2006/BSC-LT-salamandra-2b-instruct-gguf: Same weights, converted
- tensorblock/salamandra-2b-instruct-GGUF: Quantization
- BSC-LT/salamandraTA-2B-academic: Partial fine-tune
- franciscobdl/salamandra-estigia: LoRA-style
- franciscobdl/salamandra-estigiaV7: LoRA-style
- gplsi/Aitana-2B-S-LF: Partial fine-tune
- franciscobdl/salamandra-estigiaV2: LoRA-style