Commit History
-
Update README.md
Verified signature4214f88 -
Update README.md
Verified signature962808c -
Update README.md
Verified signatureA custom LLM that is based on Microsoft's Phi-3-mini-4k-instruct, model is instructed to act as an AI Law assistant that takes court recordings and transcripts and summarizes and cross references with the Indian penal code and past precedence formed by the supreme court. This model is fine-tuned by QLoRa on a by a 6 GB GTX 1660 ti gpu card. The hyper parameters include: 1) max_seq_length: 512 2) optimizer: paged_adamw_8bit 3) rank (r): 16 4) steps: 500 Training behavior was logged using weight and biases and can be viewed [here](https://api.wandb.ai/links/aryankankheria331-myself/aogjrdy0)de2e475 -
initial commit
Verified signature61d2c03