73 lines
1.9 KiB
Markdown
73 lines
1.9 KiB
Markdown
---
|
|
language:
|
|
- en
|
|
license: apache-2.0
|
|
tags:
|
|
- text-generation-inference
|
|
- transformers
|
|
- unsloth
|
|
- llama3
|
|
- trl
|
|
- hinglish
|
|
base_model: unsloth/llama-3-8b
|
|
datasets:
|
|
- cmu_hinglish_dog
|
|
---
|
|
|
|
# Inference:
|
|
|
|
```
|
|
!pip install -q "unsloth[colab-new] @ git+https://github.com/unslothai/unsloth.git"
|
|
!pip install -q --no-deps "xformers<0.0.26" trl peft accelerate bitsandbytes
|
|
```
|
|
|
|
```python
|
|
from unsloth import FastLanguageModel
|
|
import torch
|
|
max_seq_length = 512
|
|
dtype = None # None for auto detection. Float16 for Tesla T4, V100, Bfloat16 for Ampere+
|
|
load_in_4bit = True # Use 4bit quantization to reduce memory usage. Can be False.
|
|
|
|
model, tokenizer = FastLanguageModel.from_pretrained(
|
|
model_name = "Hinglish-Project/llama-3-8b-English-to-Hinglish",
|
|
max_seq_length = max_seq_length,
|
|
dtype = dtype,
|
|
load_in_4bit = load_in_4bit,
|
|
)
|
|
```
|
|
|
|
```python
|
|
def pipe(prompt):
|
|
alpaca_prompt = """### Instrucion: Translate given text to Hinglish Text:
|
|
|
|
### Input:
|
|
{}
|
|
|
|
### Response:
|
|
"""
|
|
|
|
inputs = tokenizer(
|
|
[
|
|
alpaca_prompt.format(prompt),
|
|
], return_tensors = "pt").to("cuda")
|
|
|
|
outputs = model.generate(**inputs, max_new_tokens = 2048, use_cache = True)
|
|
raw_text = tokenizer.batch_decode(outputs)[0]
|
|
return raw_text.split("### Response:\n")[1].split("<|end_of_text|>")[0]
|
|
```
|
|
|
|
```python
|
|
text = "This is a fine-tuned Hinglish translation model using Llama 3."
|
|
pipe(text)
|
|
## yeh ek fine-tuned Hinglish translation model hai jisme Llama 3 ka use kiya gaya hai.
|
|
|
|
```
|
|
# Uploaded model
|
|
|
|
- **Developed by:** Hinglish-Project
|
|
- **License:** apache-2.0
|
|
- **Finetuned from model :** unsloth/llama-8b-bnb-4bit
|
|
|
|
This Llama3 model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
|
|
|
|
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth) |