Xinging/llama2-7b_sft_0.3_ratio_alpaca_gpt4_proj_by_tydiqa_ntrain_49400_default

TEXT GENERATIONConcurrency Cost:1Model Size:7BQuant:FP8Ctx Length:4kPublished:Jan 21, 2025License:otherArchitecture:Transformer Cold

The Xinging/llama2-7b_sft_0.3_ratio_alpaca_gpt4_proj_by_tydiqa_ntrain_49400_default is a 7 billion parameter language model fine-tuned from Meta's Llama-2-7b-hf. This model was specifically fine-tuned on a dataset derived from Alpaca, GPT-4 projections, and TyDiQA, suggesting an optimization for instruction-following and question-answering tasks. It is designed for general language generation and understanding, leveraging its Llama 2 base with specialized instruction tuning.

Loading preview...

Overview

This model, llama2-7b_sft_0.3_ratio_alpaca_gpt4_proj_by_tydiqa_ntrain_49400_default, is a 7 billion parameter language model built upon the foundational Meta Llama-2-7b-hf architecture. It has undergone supervised fine-tuning (SFT) using a unique dataset that combines elements from Alpaca, GPT-4 projections, and the TyDiQA dataset. This specific training regimen aims to enhance its capabilities in instruction-following and diverse question-answering scenarios.

Key Training Details

  • Base Model: meta-llama/Llama-2-7b-hf
  • Fine-tuning Dataset: A composite dataset referred to as 0.3_ratio_alpaca_gpt4_proj_by_tydiqa_ntrain_49400.
  • Hyperparameters:
    • Learning Rate: 2e-05
    • Batch Size: 32 (train), 8 (eval)
    • Optimizer: AdamW with specific betas and epsilon
    • LR Scheduler: Cosine with 0.03 warmup ratio
    • Epochs: 3.0

Potential Use Cases

Given its fine-tuning on instruction-following and question-answering data, this model is likely suitable for:

  • Instruction-following tasks: Generating responses based on explicit instructions.
  • General question answering: Providing informative answers to a wide range of queries.
  • Text generation: Creating coherent and contextually relevant text.