metadata
library_name: transformers
license: apache-2.0
tags:
- jamba
- mamba
- moe
Mini-Jamba-v2
[Experimental Version] We initialized the model according to Jamba, but with much smaller parameters. It was then trained using about 1B of python code, and has the simplest python code generation capabilities.
Usage
Here give some examples of how to use our model:
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
prompt = '''def min(arr):
"""
Returns the minimum value from the list `arr`.
Parameters:
- arr (list): A list of numerical values.
Returns:
- The minimum value in `arr`.
"""
'''
tokenizer = AutoTokenizer.from_pretrained(
"TechxGenus/Mini-Jamba-v2",
trust_remote_code=True,
)
tokenizer.pad_token = tokenizer.eos_token
model = AutoModelForCausalLM.from_pretrained(
"TechxGenus/Mini-Jamba-v2",
torch_dtype=torch.float16,
device_map="auto",
trust_remote_code=True,
)
inputs = tokenizer.encode(prompt, return_tensors="pt")
outputs = model.generate(
input_ids=inputs.to(model.device),
max_new_tokens=64,
do_sample=False,
)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
Note
Model may sometimes make errors, produce misleading contents, or struggle to manage tasks that are not related to coding. It has undergone very limited testing. Additional safety testing should be performed before any real-world deployments.