YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
GPT-OSS mini (RoPE + GQA + MoE + SwiGLU)
This is a mini version of the GPT-OSS model trained on a Thai chatbot conversation dataset.
Model Details
- Architecture: GPT-OSS mini with RoPE, GQA, MoE, and SwiGLU
- Vocab Size: 48000
- Hidden Size: 512
- Number of Layers: 6
- Number of Attention Heads: 8
- Number of Key-Value Heads: 2
- Intermediate Size: 1536
- Number of Experts (MoE): 8
- Top-k (MoE): 2
- Max Position Embeddings: 8192
Training Details
- Dataset: ZombitX64/ThaiChatbotConversation (or fallback mini-dataset)
- Training Epochs: 3
- Learning Rate: 3e-4
- Optimizer: AdamW
- Gradient Accumulation Steps: 8
- Batch Size: 2 per device
- Gradient Checkpointing: Enabled
- Mixed Precision: FP32
Usage
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support