Downloads · 30 days
455
26% of all-time downloads
RX5950XT/silicon-based-girlfriend
silicon-based-girlfriend is a machine learning model from RX5950XT. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
🚀 重大升級:矽基女友 v2(Silicon-Based-Girlfriend v2)已正式發布! 升級至 9B 基底(Huihui-Qwen3.5-9B-abliterated),歷經 SFT 與兩輪 GRPO 強化學習訓練,全面解決跨輪重複問題,大幅提升入戲感與多輪連貫性: - 📦 開箱即用 GGUF 模型(推薦):RX5950XT/silicon-based-girlfriend-v2-GGUF - 🧬 三段 LoRA…
Downloads · 30 days
455
26% of all-time downloads
All-time downloads
1.8K
Public
Repo size
4.8 GB
Likes
5
Public
Click a slice to open those files.
.gguf4.5 GB · 94%
From the Hugging Face model README
🚀 重大升級:矽基女友 v2(Silicon-Based-Girlfriend v2)已正式發布!
升級至 9B 基底(
Huihui-Qwen3.5-9B-abliterated),歷經 SFT 與兩輪 GRPO 強化學習訓練,全面解決跨輪重複問題,大幅提升入戲感與多輪連貫性:
- 📦 開箱即用 GGUF 模型(推薦):RX5950XT/silicon-based-girlfriend-v2-GGUF
- 🧬 三段 LoRA Adapter 權重:RX5950XT/silicon-based-girlfriend-v2
- 📚 2,109 筆多輪合成語料集:RX5950XT/silicon-based-girlfriend-v2-dataset
基於 Qwen3.5-4B 的 QLoRA 微調 Adapter,訓練目標為沉浸式繁體中文角色扮演。本倉庫包含 LoRA Adapter 權重與 GGUF 格式模型。
| 項目 | 內容 |
|---|---|
| Base Model | Qwen/Qwen3.5-4B |
| Fine-tuning Method | QLoRA (4-bit NF4) |
| LoRA Rank | 32 |
| LoRA Alpha | 64 |
| LoRA Dropout | 0.05 |
| LoRA Target | All linear layers |
| Training Epochs | 5 |
| Context Length | 8192 tokens |
| Learning Rate | 1e-4 |
| LR Scheduler | Cosine |
| Optimizer | paged_adamw_8bit |
| Training Samples | 985 |
| Train Loss | 1.108 |
| Eval Loss | 1.434 |
| Hardware | NVIDIA RTX A6000 (48GB VRAM) |
| Training Time | ~19 hours |
| Framework | LLaMA-Factory |
| Chat Template | qwen3_5_nothink (non-thinking mode) |
| 檔案 | 說明 |
|---|---|
adapter_config.json | LoRA 設定檔 |
adapter_model.safetensors | LoRA 權重(248 MB) |
tokenizer_config.json | Tokenizer 設定(含 nothink chat template) |
tokenizer.json | Tokenizer |
vocab.json / merges.txt | Vocabulary |
silicon-gf-q8_0.gguf | Q8_0 量化 GGUF(4.2 GB,適用 llama.cpp / LM Studio) |
training_loss.png | 訓練 Loss 曲線 |
training_eval_loss.png | 評估 Loss 曲線 |
直接在 LM Studio 或 llama.cpp 載入 silicon-gf-q8_0.gguf,無需額外安裝。
# llama.cpp
./llama-cli -m silicon-gf-q8_0.gguf -c 8192 --temp 0.8
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base_model = "Qwen/Qwen3.5-4B"
adapter = "RX5950XTP/silicon-based-girlfriend"
tokenizer = AutoTokenizer.from_pretrained(adapter)
model = AutoModelForCausalLM.from_pretrained(base_model, device_map="auto")
model = PeftModel.from_pretrained(model, adapter)
messages = [
{"role": "user", "content": "嘿,你在幹嘛?"}
]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512, temperature=0.8, do_sample=True)
print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))
llamafactory-cli chat \
--model_name_or_path Qwen/Qwen3.5-4B \
--adapter_name_or_path RX5950XTP/silicon-based-girlfriend \
--template qwen3_5_nothink \
--finetuning_type lora
![]()
system + conversations with from/value)qwen3_5_nothink chat template,預設不啟用思考模式,回覆會直接輸出角色對話。Apache 2.0(遵循 Qwen3.5-4B 原授權)