Edit model card

This model is randomly initialized, using the config from mistralai/Mistral-7B-v0.1 but with smaller size.

Codes:

from optimum.intel.openvino import OVModelForCausalLM
from transformers import pipeline
from huggingface_hub import create_repo, upload_folder
import torch
import transformers
import os

model_id = 'mistralai/Mistral-7B-v0.1'
save_path = '/tmp/yujiepan/mistral-tiny-random'
repo_id = 'yujiepan/mistral-tiny-random'

config = transformers.AutoConfig.from_pretrained(model_id)
config.hidden_size = 8
config.intermediate_size = 32
config.num_attention_heads = 4
config.num_hidden_layers = 2
config.num_key_value_heads = 2
print(config)

tokenizer = transformers.AutoTokenizer.from_pretrained(model_id)
tokenizer.save_pretrained(save_path)

model = transformers.AutoModelForCausalLM.from_config(config, torch_dtype=torch.float16)
model = model.half()

pipe = pipeline('text-generation', model=model, tokenizer=tokenizer, do_sample=False, device='cuda')
print(pipe('Hello World!'))

model.save_pretrained(save_path)

ovmodel = OVModelForCausalLM.from_pretrained(save_path, export=True)
ovmodel = ovmodel.half()
ovmodel.save_pretrained(save_path)

os.system(f'ls -alh {save_path}')
create_repo(repo_id, exist_ok=True)
upload_folder(repo_id=repo_id, folder_path=save_path)
Downloads last month
7
Safetensors
Model size
514k params
Tensor type
FP16
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Collection including yujiepan/mistral-tiny-random