Datasetpublic

Nutanix/transformers_zero_shot_llama70b_llama8b_results

Results for 0-shot prompting on transformers code using queries provided by HuggingFace team. Description The RAG responses are generated using the dataset: https://huggingface.co/datasets/Nutanix/transformers_processed_data_llama70b. Llama-3-128k-70b-Instruct-awq was used for feature extraction of transformers code(v4.42.3) to create the dataset, and Llama-3-8b-Instruct was used for generation of both RAG and Raw outputs. Data Fields id - The id for the… See the full description on the dataset page: https://huggingface.co/datasets/Nutanix/transformers_zero_shot_llama70b_llama8b_results.

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes12downloads

Dataset Card

Results for 0-shot prompting on transformers code using queries provided by HuggingFace team.

Description

The RAG responses are generated using the dataset: https://huggingface.co/datasets/Nutanix/transformersprocesseddata_llama70b. Llama-3-128k-70b-Instruct-awq was used for feature extraction of transformers code(v4.42.3) to create the dataset, and Llama-3-8b-Instruct was used for generation of both RAG and Raw outputs.

Data Fields

1.id - The id for the question/response.
2.question - The queston used for generating response from the LLM.
3.generation - The response generated using appropriate context from retrieved files
4.generation_time - Time taken to generate the response from the LLM
5.completion_tokens - Number of tokens in the generated response
6.prompt tokens - Number of tokens in the prompt
7.total tokens - Total number of tokens in prompt + generation