67 lines
2.1 KiB
Markdown
67 lines
2.1 KiB
Markdown
---
|
|
license: mit
|
|
datasets:
|
|
- mlabonne/guanaco-llama2-1k
|
|
language:
|
|
- en
|
|
base_model:
|
|
- NousResearch/Llama-2-7b-chat-hf
|
|
pipeline_tag: text-generation
|
|
library_name: transformers
|
|
finetuned_model: true
|
|
model_type: causal-lm
|
|
finetuned_task: instruction-following
|
|
tags:
|
|
- instruction-following
|
|
- text-generation
|
|
- fine-tuned
|
|
- llama2
|
|
- causal-language-model
|
|
- QLoRa
|
|
- 4-bit-quantization
|
|
- low-memory
|
|
- training-optimized
|
|
metrics:
|
|
- accuracy
|
|
- loss
|
|
---
|
|
|
|
# Llama-2-7B-Chat Fine-Tuned Model
|
|
|
|
This model is a fine-tuned version of **Llama-2-7B-Chat** model, optimized for instruction-following tasks. It has been trained on the `mlabonne/guanaco-llama2-1k` dataset and is optimized for efficient text generation across various NLP tasks, including question answering, summarization, and text completion.
|
|
|
|
## Model Details
|
|
- **Base Model**: NousResearch/Llama-2-7b-chat-hf
|
|
- **Fine-Tuning Task**: Instruction-following
|
|
- **Training Dataset**: mlabonne/guanaco-llama2-1k
|
|
- **Optimized For**: Text generation, question answering, summarization, and more.
|
|
- **Fine-Tuned Parameters**:
|
|
- **LoRA** (Low-Rank Adaption) applied for efficient training with smaller parameter updates.
|
|
- Quantized to **4-bit** for memory efficiency and better GPU utilization.
|
|
- Training includes **gradient accumulation**, **gradient checkpointing**, and **weight decay** to prevent overfitting and enhance memory efficiency.
|
|
|
|
## Usage
|
|
|
|
You can use this fine-tuned model with the Hugging Face `transformers` library. Below is an example of how to load and use the model for text generation.
|
|
|
|
```python
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
|
|
# Load pre-trained model and tokenizer
|
|
tokenizer = AutoTokenizer.from_pretrained("https://huggingface.co/devshaheen/llama-2-7b-chat-finetune")
|
|
model = AutoModelForCausalLM.from_pretrained("https://huggingface.co/devshaheen/llama-2-7b-chat-finetune")
|
|
|
|
# Example text generation
|
|
input_text = "What is the capital of France?"
|
|
inputs = tokenizer(input_text, return_tensors="pt")
|
|
outputs = model.generate(**inputs)
|
|
generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
|
|
|
|
print(generated_text)
|
|
|
|
|
|
|
|
|
|
|
|
|