Downloads · 30 days
12
20% of all-time downloads
mistral-sat
nyanxyz/mistral-sat
mistral-sat is a text generation model from nyanxyz. Use it when you need the model to write or continue text. It is set up for transformers. The card lists the license as other.
This model was trained using AutoTrain. For more information, please visit AutoTrain.
All-time downloads
59
Public
Parameters
7.2B
14.6 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors14.5 GB · 99%
At a glance
- Task
- Text Generation
- Library
- transformers
- License
- other
- Model type
- mistral
- Access
- Public
- Created
- Dec 6, 2023
- Updated
- Dec 6, 2023
- SHA
- b8f591df
Try a prompt
Model card
From the Hugging Face model README
Model Trained Using AutoTrain
This model was trained using AutoTrain. For more information, please visit AutoTrain.
Usage
from transformers import AutoModelForCausalLM, AutoTokenizer
model_path = "PATH_TO_THIS_REPO"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
model_path,
device_map="auto",
torch_dtype='auto'
).eval()
# Prompt content: "hi"
messages = [
{"role": "user", "content": "hi"}
]
input_ids = tokenizer.apply_chat_template(conversation=messages, tokenize=True, add_generation_prompt=True, return_tensors='pt')
output_ids = model.generate(input_ids.to('cuda'))
response = tokenizer.decode(output_ids[0][input_ids.shape[1]:], skip_special_tokens=True)
# Model response: "Hello! How can I assist you today?"
print(response)
- Task
- Text Generation
- Library
- transformers
- Type
- mistral
- License
- other
- Created
- Dec 6, 2023
- Updated
- Dec 6, 2023