nomic-ai/CodeRankEmbed

Full fine-tune vs Snowflake/snowflake-arctic-embed-m-long. Fine-tune: 100% of the weights changed, across 12 of 12 layers.

Tensors changed: other 2/2, Embeddings 2/2, Attention 24/24, MLP 36/36, Norms 48/48

3 derivatives checked

• handwoven8588/CodeRankEmbed-flash-attn: Same weights, converted
• Salesforce/SweRankEmbed-Small: Full fine-tune
• brandtcormorant/CodeRankEmbed-Q8_0-GGUF: Quantization

On the Hub · Downloads on Model Pulse