Er1ckRg/cuphead-qwen3-1.7b-finetuned
The Er1ckRg/cuphead-qwen3-1.7b-finetuned model is a 1.7 billion parameter Qwen3-based large language model, fine-tuned by Er1ckRg. It is specifically optimized for answering factual questions about the game Cuphead and its DLC, The Delicious Last Course, in Portuguese. This model excels at providing precise information on game mechanics, characters, bosses, weapons, and cooperative play.
Loading preview...
Model Overview
This model, Er1ckRg/cuphead-qwen3-1.7b-finetuned, is a specialized version of the Qwen3-1.7B large language model. It has been fine-tuned using Supervised Fine-Tuning (SFT) with the TRL library to become an expert on the video game Cuphead and its DLC, The Delicious Last Course, with responses primarily in Portuguese.
Key Capabilities
- Cuphead Expertise: Provides accurate and factual answers regarding Cuphead's campaign, bosses, game mechanics, characters (Cuphead, Mugman, Ms. Chalice), weapons, charms, and cooperative mode.
- DLC Knowledge: Covers content from The Delicious Last Course DLC.
- Portuguese Language: Optimized for generating responses in Portuguese.
Training Details
The model was trained on the Er1ckRg/cuphead_basico dataset, which consists of factual question-and-answer pairs in Portuguese, formatted as chat conversations. Training involved 3 epochs on NVIDIA Tesla T4 hardware using FP16 precision, with loss calculated only on assistant tokens.
Limitations
- Domain Specificity: Performance is guaranteed only for questions related to Cuphead. Queries outside this domain will rely on the base Qwen3-1.7B's general knowledge, without quality assurance.
- Dataset Scope: The training dataset is of reduced size, meaning it might not cover extremely specific or rare in-game situations.
- Language Focus: Primarily designed for Portuguese; performance in other languages has not been evaluated.