pestlee/Qwable-9B-Claude-Fable-5-StraTA
LoRA-style (fewer layers) vs empero-ai/Qwable-9B-Claude-Fable-5. pestlee/Qwable-9B-Claude-Fable-5-StraTA matches part of empero-ai/Qwable-9B-Claude-Fable-5 (427 of its 760 tensors; the rest of empero-ai/Qwable-9B-Claude-Fable-5 has no counterpart). On the shared part: Looks like a merged LoRA (or a fine-tune with frozen norms): only the weight matrices of down_proj, gate_proj, k_proj, o_proj, q_proj, up_proj, v_proj changed, in 32 of 32 layers; norms, biases and embeddings are untouched.
Tensors changed: lm_head 0/1, Embeddings 0/1, Norms 0/105, Attention 32/224, MLP 96/96
2 derivatives checked
- mradermacher/Qwable-9B-Claude-Fable-5-StraTA-i1-GGUF: Quantization
- mradermacher/Qwable-9B-Claude-Fable-5-StraTA-GGUF: Same weights, converted