checkpoints-mistral-300M
This model is a fine-tuned version of None on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.205
Model description
More information needed
Training and evaluation data
***** train metrics *****
epoch = 13.91 train_loss = 2.205
***** eval metrics *****
epoch = 13.91 eval_loss = 2.4 perplexity = 11.0228
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 6
- eval_batch_size: 6
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradient_accumulation_steps: 16
- total_train_batch_size: 192
- total_eval_batch_size: 12
- optimizer: Adam with betas=(0.9,0.95) and epsilon=0.0001
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 4
- num_epochs: 6
- mixed_precision_training: Native AMP
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.5
- Tokenizers 0.14.1
Usage
from transformers import pipeline
pipe = pipeline("text-generation", model="ayousanz/japanese-mistral-0.3b-base")
from transformers import AutoModelForCausalLM, AutoTokenizer, TextStreamer
import torch
MODEL_NAME = "ayousanz/japanese-mistral-0.3b-base"
torch.set_float32_matmul_precision('high')
DEVICE = "cuda"
if torch.cuda.is_available():
print("cuda")
DEVICE = "cuda"
else:
print("cpu")
DEVICE = "cpu"
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME,use_fast=False)
model = AutoModelForCausalLM.from_pretrained(
MODEL_NAME,
trust_remote_code=True,
).to(DEVICE)
prompt = "大規模言語モデルとは、"
inputs = tokenizer(prompt, add_special_tokens=False,return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model.generate(
inputs["input_ids"],
max_new_tokens=256,
do_sample=True,
early_stopping=False,
top_p=0.95,
top_k=50,
temperature=0.9,
no_repeat_ngram_size=2,
num_beams=3
)
outputs_txt = tokenizer.decode(outputs[0])
print(outputs_txt)
- Downloads last month
- 12
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.