PetarKal/qwen3-4b-EM-full-finetuned-v5
PetarKal/qwen3-4b-EM-full-finetuned-v5 is a 4 billion parameter language model based on the Qwen3-4B architecture, fine-tuned by PetarKal. This model was trained using the TRL library, indicating a focus on instruction-following or specific task performance. Its fine-tuned nature suggests optimization for conversational AI or text generation tasks where adherence to prompts is crucial. The model leverages a 32768 token context length for processing extensive inputs.
Loading preview...
Overview
PetarKal/qwen3-4b-EM-full-finetuned-v5 is a 4 billion parameter language model derived from the Qwen/Qwen3-4B base model. It has undergone fine-tuning using the TRL (Transformers Reinforcement Learning) library, a framework designed for training large language models with reinforcement learning from human feedback or other methods to improve instruction following and task performance.
Key Capabilities
- Instruction Following: The fine-tuning process with TRL suggests enhanced ability to follow specific instructions and generate relevant responses.
- Text Generation: Capable of generating coherent and contextually appropriate text based on given prompts.
- Qwen3-4B Foundation: Benefits from the robust architecture and pre-training of the Qwen3-4B model.
Training Details
The model was trained using Supervised Fine-Tuning (SFT) within the TRL framework. The training process utilized specific versions of key libraries including TRL 0.29.1, Transformers 5.9.0, Pytorch 2.10.0, Datasets 4.8.5, and Tokenizers 0.22.2. Further details on the training run are available via a Weights & Biases link provided by the author.
Good For
- Applications requiring a compact yet capable language model for instruction-tuned tasks.
- Developers looking for a fine-tuned Qwen3-4B variant with improved conversational or prompt-response capabilities.