microsoft/chatbench-distilgpt2

Full fine-tune vs distilbert/distilgpt2. Fine-tune: 100% of the weights changed, across 6 of 6 layers.

Tensors changed: Attention 24/24, Norms 26/26, MLP 24/24, Embeddings 2/2

3 derivatives checked

• ysn-rfd/chatbench-distilgpt2-GGUF: Same weights, converted
• ysn-rfd/OpenELM-3B-Instruct-GGUF: Same weights, converted
• DevQuasar/microsoft.chatbench-distilgpt2-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse