yyyyFan/final_proj-stage2-best-lr1e4-r16-merged-bf16
TEXT GENERATIONConcurrency Cost:1Model Size:8BQuant:FP8Ctx Length:32kPublished:Apr 12, 2026Architecture:Transformer0.0K Cold

The yyyyFan/final_proj-stage2-best-lr1e4-r16-merged-bf16 is an 8 billion parameter causal language model, derived from the Qwen/Qwen3-8B base model. It represents a merged BF16 full-model export of a LoRA adapter, specifically from yyyyFan/final_proj-stage2-best-lr1e4-r16. This model is designed for standard Hugging Face causal language model applications, offering a context length of 32768 tokens.

Loading preview...

Overview

The yyyyFan/final_proj-stage2-best-lr1e4-r16-merged-bf16 is an 8 billion parameter causal language model, built upon the Qwen/Qwen3-8B base architecture. This model is a merged BF16 full-model export of a LoRA adapter, specifically sourced from the yyyyFan/final_proj-stage2-best-lr1e4-r16 repository. It is designed to be loaded and utilized directly as a standard Hugging Face causal language model checkpoint.

Key Characteristics

  • Base Model: Qwen/Qwen3-8B
  • Parameter Count: 8 billion parameters
  • Context Length: 32768 tokens
  • Export Type: Merged BF16, indicating a full-model export with bfloat16 precision.

Good for

  • Applications requiring a Qwen3-8B-based model with specific fine-tuning from the yyyyFan/final_proj-stage2-best-lr1e4-r16 adapter.
  • Developers looking for a readily available, merged checkpoint for deployment in Hugging Face environments.
  • Use cases benefiting from a model with a substantial 32768 token context window.