Apriel 5B Instruct (Llamafied)

An approximate conversion of Apriel 5B Instruct to the Llama format.

  • No trust_remote_code required
  • Compatible w/ popular fine-tuning frameworks
  • Easier to run

Usage:

from transformers import pipeline
import torch

device = 'cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu'

pipe = pipeline('text-generation', model='mrfakename/Apriel-5B-Instruct-llamafied', device=device)

print(pipe(pipe.tokenizer.apply_chat_template([{'role': 'user', 'content': 'Hello'}], tokenize=False, add_generation_prompt=True))[0]['generated_text'])

License: MIT (same as the original Apriel 5B model)

Downloads last month
63
Safetensors
Model size
4.83B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for mrfakename/Apriel-5B-Instruct-llamafied

Finetuned
(2)
this model
Quantizations
4 models

Collection including mrfakename/Apriel-5B-Instruct-llamafied