Diluner/gpt54-mini-sequential-qwen3-1.7b-sft-s2-textcraft-20260920
The Diluner/gpt54-mini-sequential-qwen3-1.7b-sft-s2-textcraft-20260920 is a 1.7 billion parameter Qwen3-based causal language model, fine-tuned using Supervised Fine-Tuning (SFT) with a gpt-5.4-mini teacher. This model represents the completed 'textcraft' stage (stage 2) of a sequential training process, following an initial 'babyai' stage. It is specifically designed for text generation tasks, having undergone five epochs of textcraft training, and features a 32768 token context length.
Loading preview...
Model Overview
This model, gpt54-mini-sequential-qwen3-1.7b-sft-s2-textcraft-20260920, is a 1.7 billion parameter Qwen3-based causal language model developed by Diluner. It has been fine-tuned using Supervised Fine-Tuning (SFT) with gpt-5.4-mini as the teacher model. This particular checkpoint signifies the completion of the 'textcraft' stage (stage 2) in a sequential training pipeline, which followed an initial 'babyai' stage. The textcraft stage involved five epochs of training, with 55 optimizer updates.
Key Characteristics
- Base Model: Qwen3-1.7B architecture.
- Training Method: Supervised Fine-Tuning (SFT) using a
gpt-5.4-miniteacher. - Sequential Training: Represents the second stage ('textcraft') of a multi-stage training process, building upon a 'babyai' stage.
- Context Length: Supports a context window of 32768 tokens.
- Intermediate Checkpoint: This is an intermediate checkpoint; final evaluations are not attached and belong to the fully trained stage-3 model.
Usage Notes
This model is provided as a trained checkpoint from a specific sequential chain. It is important to note that this checkpoint does not include optimizer state, raw logs, or teacher trajectories. Users should consult the base model's license and applicable terms, as no specific license is asserted for this checkpoint.