GLM-5.2 is live. #1 throughput on OpenRouter, pay-per-token on FriendliAI. Try it today ➜
mikuhhn1239
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
mikuhhn1239
Model Tree
Base
mikuhhn1239/qwen3-8b-novel-base-sft
Adapter
this model
Input Modalities
Text
Output Modalities
Text
Supported Functionality
Dedicated Endpoints
Container
{"best_candidate": "角色名", "uncertain": true/false}markdown
输入:候选: 林秋、苏晚、陈遥上下文:[1] "你今天又迟到了。"苏晚把笔放下。[2] 我把书包丢到桌上，没接话。[3] 陈遥坐在后排憋笑。[4] "路上堵车。" 【目标对话】输出:{"best_candidate": "林秋", "uncertain": true}← 第2句"我把书包丢到桌上"暗示"我"说了[4]，陈遥在憋笑排除，苏晚说了[1]；证据不足，不确定
python
from transformers import AutoModelForCausalLMfrom peft import PeftModelbase = AutoModelForCausalLM.from_pretrained("mikuhhn1239/qwen3-8b-novel-base-sft",torch_dtype="auto", device_map="auto",)model = PeftModel.from_pretrained(base, "mikuhhn1239/qwen3-8b-attribution-assist-lora")
markdown
基座: Qwen3-8B-Novel-Base-SFT (Stage1 全参 SFT, 72K 小说续写数据)方法: LoRA (r=64, α=128, dropout=0.05)数据: 495 条 (465 train / 30 val / 30 test)框架: transformers Trainer + PEFT优化器: AdamW (adamw_torch_fused), cosine schedule, warmup=5%epoch: 5 | LR: 1e-4 | batch: 1×16(accum) | bf16 | max_length: 4096
| 版本 | 数据量 | epochs | LR | JSON解析 | 最佳候选准确率 | 说明 |
|---|---|---|---|---|---|---|
| 零基座 | — | — | — | 0% | 0% | Qwen3-8B 原始完全不会 |
| +Stage1 | — | — | — | 3.3% | 0% | 读完 669 本也不会 |
| v1 | 66 | 3 | 2e-4 | 100% | 14.3% | 端到端（排序+打分） |
| v2 | 240 |
| 3 |
| 2e-4 |
| 100% |
| 33.3% |
| 只选最佳+不确定 |
| v3.1 | 240 | 5 | 1e-4 | 100% | 43.3% | Clean 标注 |
| v3.2 ⭐ | 465 | 5 | 1e-4 | 100% | 86.7% | 候选清洁 + 数据增强 +94% |