File size: 3,075 Bytes
e444a3b 09ecd43 e444a3b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
---
license: apache-2.0
base_model:
- deepseek-ai/deepseek-llm-7b-chat
pipeline_tag: text-generation
tags:
- silk
- eco-friendly
- sustainable
- agriculture
- deepseek
- llama
- fine-tuned
- zh
- chinese
---
# EcoSilkModel
## Model Overview
**EcoSilkModel** is a fine-tuned language model based on [DeepSeek-LLM-7B](https://huggingface.co/deepseek-ai/deepseek-llm-7b-base), specifically designed for sustainable agriculture, silk production, and eco-friendly practices. The model excels in the following tasks:
- **Silk Production Guidance**: Provides recommendations for sustainable silk farming and sericulture.
- **Eco-friendly Practices**: Offers suggestions for environmentally friendly agricultural practices.
- **Chinese Language Support**: Focuses on Chinese tasks while also supporting both Chinese and English.
This model is part of the **EcoSilk Project**, aimed at promoting sustainable development in the silk industry.
## Applicable Scenarios
The model is suitable for the following scenarios:
- **Researchers**: Studying sustainable agriculture and silk production.
- **Farmers**: Seeking guidance on eco-friendly agricultural practices.
- **Educators**: Teaching sustainable agricultural practices.
- **Developers**: Building applications for the silk and agricultural industries.
## Training Data
The model was fine-tuned on the following datasets:
1. **zhanxu/ecosilk-chat**: A Chinese instruction-following dataset used for fine-tuning language models.（https://huggingface.co/datasets/zhanxu/ecosilk-chat）
The dataset is derived from professional literature across various fields. It uses open-source tools to automatically annotate question-answer datasets, which are then manually cleaned and filtered.
For more details, see [GitHub - ConardLi/easy-dataset: A powerful tool for creating fine-tuning datasets for LLM](https://github.com/ConardLi/easy-dataset).
## Fine-tuning Details
- **Base Model**: [DeepSeek-LLM-7B](https://huggingface.co/deepseek-ai/deepseek-llm-7b-base)
- **Fine-tuning Method**: LoRA (Low-Rank Adaptation)
- **Template**: `llama3`
- **Languages**: Chinese (`zh`) and English (`en`)
- **Truncation Length**: 1024 tokens
# How to Use Our Model
Here are some examples of how to use our model.
## Chat Completion
```python
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, GenerationConfig
model_name = "deepseek-ai/deepseek-llm-7b-chat"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
model.generation_config = GenerationConfig.from_pretrained(model_name)
model.generation_config.pad_token_id = model.generation_config.eos_token_id
messages = [
{"role": "user", "content": "Who are you?"}
]
input_tensor = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt")
outputs = model.generate(input_tensor.to(model.device), max_new_tokens=100)
result = tokenizer.decode(outputs[0][input_tensor.shape[1]:], skip_special_tokens=True)
print(result) |