---
library_name: transformers
tags:
- time series
- multimodal
- TimeSeries-Text-to-Text
license: apache-2.0
---

# Mists-7B-v0.1-not-trained

Mists(**Mis**tral **T**ime **S**eries) model is a multimodal model that combines language and time series model.  
This model is based on the following models:  
 - [mistralai/Mistral-7B-Instruct-v0.3](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.3)
 - [AutonLab/MOMENT-1-large](https://huggingface.co/AutonLab/MOMENT-1-large)

This is an experimental model.  
It has some limitations and is not suitable for use at this time.

## How to load model

```Python
!pip install git+https://github.com/Hajime-Y/moment.git
!pip install -U transformers
!git clone https://github.com/Hajime-Y/Mists.git
```

```Python
import torch

from Mists.configuration_mists import MistsConfig
from Mists.modeling_mists import MistsForConditionalGeneration
from Mists.processing_mists import MistsProcessor

model_id = "HachiML/Mists-7B-v0.1-not-trained"
model = MistsForConditionalGeneration.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    low_cpu_mem_usage=True,
).to("cuda")
processor = MistsProcessor.from_pretrained(model_id)
```

```Python
import pandas as pd

hist_ndaq_512 = pd.DataFrame("nasdaq_price_history.csv")
time_series_data = torch.tensor(hist_ndaq_512[["Open", "High", "Low", "Close", "Volume"]].values, dtype=torch.float)
time_series_data = time_series_data.t().unsqueeze(0)

prompt = "USER: <time_series>\nWhat are the features of this data?\nASSISTANT:"
inputs = processor(prompt, time_series_data, return_tensors='pt').to(torch.float32)

output = model.generate(**inputs, max_new_tokens=200, do_sample=False)
print(processor.decode(output[0], skip_special_tokens=True))
```