Oriserve/Whisper-Hindi2Hinglish-Apex
Partial fine-tune vs openai/whisper-large-v3-turbo. Partial change: 24% of the weights changed (attention, embeddings, mlp, norms, other), in 6 of 32 layers.
Tensors changed: Embeddings 2/3, Norms 28/156, Attention 42/280, MLP 24/144, other 4/4