sknyazev/qwen3-tts-12hz-1.7b-ru-stress-gguf
LoRA-style (fewer layers) vs Qwen/Qwen3-TTS-12Hz-1.7B-Base. sknyazev/qwen3-tts-12hz-1.7b-ru-stress-gguf matches part of Qwen/Qwen3-TTS-12Hz-1.7B-Base (309 of its 480 tensors; the rest of Qwen/Qwen3-TTS-12Hz-1.7B-Base has no counterpart). On the shared part: sknyazev/qwen3-tts-12hz-1.7b-ru-stress-gguf is a GGUF (F16) of different weights. 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 28 of 28 layers; norms, biases and embeddings are untouched.
Tensors changed: Norms 0/113, MLP 84/84, Attention 112/112