Edit model card

Uploaded model

  • Developed by: sidvash
  • License: apache-2.0
  • Finetuned from model : unsloth/llama-3-8b-Instruct-bnb-4bit

This llama model was trained 2x faster with Unsloth and Huggingface's TRL library.

Task

GIven a document, and an event type from FrameNet, extract all instances of that event type from the document.

Data

v0.1

This model used FAMuS train set (759 examples):

  • Each instance had a gold extracted event + silver generated extracted events of the same type as extracted from Gemini-1.5-pro model with a 10-shot ICL gold annotated examples
  • All instances are positive data (i.e. there is at least one instance of the event type present in the data)

More details: TBD

Downloads last month
11
Safetensors
Model size
8.03B params
Tensor type
BF16
·
Inference API
Input a message to start chatting with sidvash/famus_exh_task2_unsloth_llama-3-8b-Instruct-bnb-4bit-merged_16bit.
Model is too large to load in Inference API (serverless). To try the model, launch it on Inference Endpoints (dedicated) instead.

Finetuned from