Instructions to use ceilf6/frontagent-planner-7B-lora with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ceilf6/frontagent-planner-7B-lora with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("unsloth/qwen2.5-coder-7b-bnb-4bit")
model = PeftModel.from_pretrained(base_model, "ceilf6/frontagent-planner-7B-lora")

Transformers

How to use ceilf6/frontagent-planner-7B-lora with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="ceilf6/frontagent-planner-7B-lora")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("ceilf6/frontagent-planner-7B-lora", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ceilf6/frontagent-planner-7B-lora with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ceilf6/frontagent-planner-7B-lora"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ceilf6/frontagent-planner-7B-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ceilf6/frontagent-planner-7B-lora

SGLang

How to use ceilf6/frontagent-planner-7B-lora with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "ceilf6/frontagent-planner-7B-lora" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ceilf6/frontagent-planner-7B-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "ceilf6/frontagent-planner-7B-lora" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ceilf6/frontagent-planner-7B-lora",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio new

How to use ceilf6/frontagent-planner-7B-lora with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for ceilf6/frontagent-planner-7B-lora to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for ceilf6/frontagent-planner-7B-lora to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for ceilf6/frontagent-planner-7B-lora to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="ceilf6/frontagent-planner-7B-lora",
    max_seq_length=2048,
)

Docker Model Runner
How to use ceilf6/frontagent-planner-7B-lora with Docker Model Runner:
```
docker model run hf.co/ceilf6/frontagent-planner-7B-lora
```

frontagent-planner-7B-lora / README.md

ceilf6

Update README.md

cabcb95 verified 21 days ago

preview code

raw

history blame contribute delete

4.7 kB

	---
	base_model: unsloth/qwen2.5-coder-7b-bnb-4bit
	library_name: peft
	pipeline_tag: text-generation
	tags:
	- base_model:adapter:unsloth/qwen2.5-coder-7b-bnb-4bit
	- lora
	- sft
	- transformers
	- trl
	- unsloth
	- frontend
	- agent
	- planner
	- code-generation
	language:
	- zh
	- en
	---

	# FrontAgent Planner 7B (LoRA Adapter)

	基于 Qwen2.5-Coder-7B 微调的前端任务规划 LoRA adapter，从 [FrontAgent](https://github.com/ceilf6/FrontAgent) 的 Planner 阶段蒸馏而来，能够根据自然语言任务描述生成结构化的前端开发执行计划。

	## Model Details

	### Model Description

	- Developed by: ceilf6
	- Model type: LoRA adapter for causal language model
	- Language(s): 中文, English
	- License: Apache 2.0 (同基座模型)
	- Finetuned from model: unsloth/qwen2.5-coder-7b-bnb-4bit (Qwen/Qwen2.5-Coder-7B)

	### Model Sources

	- Repository: https://github.com/ceilf6/FrontAgent
	- 基座模型: https://huggingface.co/Qwen/Qwen2.5-Coder-7B

	## Uses

	### Direct Use

	输入一个前端开发任务描述和项目上下文，模型输出结构化的 JSON 执行计划，包含：
	- 按阶段组织的步骤列表（阶段1-分析到阶段7-仓库管理）
	- 每个步骤的 description、action、phase 字段
	- 风险分析 (risks) 和备选方案 (alternatives)

	### Out-of-Scope Use

	- 不适用于非前端工程的通用规划任务
	- 不应作为生产环境的自动化执行引擎，生成的计划需人工审核

	## Bias, Risks, and Limitations

	- 训练数据为合成数据（Claude API 生成），可能无法覆盖所有真实场景
	- 7B 模型在复杂多步骤任务上的推理能力有限
	- 输出的 JSON 结构可能不完全稳定，建议后处理校验

	## How to Get Started with the Model

	```python
	from transformers import AutoModelForCausalLM, AutoTokenizer
	from peft import PeftModel

	base_model = "Qwen/Qwen2.5-Coder-7B"
	adapter = "ceilf6/frontagent-planner-7B-lora"

	tokenizer = AutoTokenizer.from_pretrained(base_model)
	model = AutoModelForCausalLM.from_pretrained(base_model, torch_dtype="auto", device_map="auto")
	model = PeftModel.from_pretrained(model, adapter)

	messages = [
	{"role": "system", "content": "你是一个资深前端工程师和项目规划专家。请根据以下任务描述和项目上下文，生成一个结构化的执行计划。计划应按阶段组织（阶段1-分析、阶段2-创建、阶段3-安装、阶段4-验证、阶段5-启动、阶段6-浏览器验证、阶段7-仓库管理），每个步骤包含 description（描述）、action（动作类型）、phase（所属阶段）。同时提供 risks（潜在风险）和 alternatives（备选方案）。\n\n可用的动作类型: read_file, list_directory, create_file, apply_patch, search_code, get_ast, run_command, browser_navigate, browser_screenshot, get_page_structure, browser_click, browser_type"},
	{"role": "user", "content": "任务：创建一个用户登录页面，包含邮箱和密码输入框，支持表单验证\n\n项目上下文：\nReact 18 + TypeScript + Ant Design 5"},
	]

	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=1536, temperature=0.7, top_p=0.9)
	response = tokenizer.decode(outputs[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True)
	print(response)
	```

	## Training Details

	### Training Data

	由 Claude API 基于 FrontAgent-app 的 Planner 系统提示词合成的 ~100 条前端任务规划数据，覆盖创建、修改、分析三类任务场景，Alpaca 格式 (instruction/input/output)。

	### Training Hyperparameters

	- Training framework: Unsloth 2x fast finetuning
	- Base model: Qwen/Qwen2.5-Coder-7B (4-bit quantized)
	- Method: QLoRA (4-bit) + LoRA SFT
	- LoRA rank: 16
	- LoRA alpha: 32
	- Target modules: q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
	- Learning rate: 1e-4
	- Epochs: 5
	- Batch size: 2
	- Gradient accumulation: 4
	- Max sequence length: 1024
	- Optimizer: AdamW 8-bit
	- LR scheduler: Cosine
	- Warmup ratio: 0.05

	### Speeds, Sizes, Times

	- Training hardware: Google Colab T4 GPU (16GB VRAM)
	- Training time: ~45-90 分钟
	- Adapter size: ~50MB
	- 可训练参数: 80,740,352 / 4,433,712,640 (1.82%)

	## Evaluation

	### Metrics

	- JSON 合法率: 模型输出是否为合法 JSON
	- 完整计划率: 是否包含 phases/steps/risks/alternatives
	- 步骤数: 每个计划包含的步骤数量

	## Framework versions

	- PEFT 0.19.1
	- Transformers 5.5.0
	- Unsloth 2026.5.2
	- TRL (SFTTrainer)
	- Torch 2.10.0+cu128