ADRA-RL/tulu2-7b_olympiads_controlled_contamination_original
ADRA-RL/tulu2-7b_olympiads_controlled_contamination_original is a 7 billion parameter language model fine-tuned from allenai/tulu-2-7b using the TRL framework. This model was trained with Supervised Fine-Tuning (SFT) and has a context length of 4096 tokens. It is designed for general text generation tasks, building upon the capabilities of its base model.
Loading preview...
Model Overview
This model, tulu2-7b_olympiads_controlled_contamination_original, is a 7 billion parameter language model derived from the allenai/tulu-2-7b base model. It has been fine-tuned using the Transformer Reinforcement Learning (TRL) framework, specifically employing Supervised Fine-Tuning (SFT) for its training procedure.
Key Capabilities
- Text Generation: Capable of generating coherent and contextually relevant text based on given prompts.
- Instruction Following: Inherits and potentially enhances the instruction-following abilities of the
tulu-2-7bbase model through SFT. - Base Model Foundation: Leverages the robust architecture and pre-training of
allenai/tulu-2-7b.
Training Details
The model underwent Supervised Fine-Tuning (SFT) using TRL version 0.19.1, with Transformers 4.51.1 and Pytorch 2.6.0. The training process can be visualized via Weights & Biases, as indicated in the original model card. This fine-tuning aims to adapt the base model for specific performance characteristics, though the exact nature of the 'olympiads_controlled_contamination' dataset or objective is not detailed in the provided README.
Good For
- Developers looking for a fine-tuned 7B parameter model based on
tulu-2-7b. - General text generation tasks where a TRL-trained model is preferred.
- Further experimentation or fine-tuning on specific downstream applications.