FINAL-Bench/Darwin-TTS-1.7B-Cross
Full fine-tune (fewer layers) vs Qwen/Qwen3-1.7B. Qwen/Qwen3-1.7B matches part of FINAL-Bench/Darwin-TTS-1.7B-Cross (309 of its 976 tensors; the rest of FINAL-Bench/Darwin-TTS-1.7B-Cross has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Norms 113/113, MLP 84/84, Attention 112/112
3 derivatives checked
- AMAImedia/NOESIS-Qwen3-TTS-1.7B-Talker-Combo-LLM2-ASR2-BF16: LoRA-style (fewer layers)
- zeropointnine/Darwin-TTS-1.7B-Cross-Qwen3Tokenizer: Identical copy
- cstr/darwin-tts-1.7b-cross-GGUF: Not comparable yet
1 other repos hold byte-for-byte identical weight files.