|
import torch |
|
from transformers import AutoTokenizer, AutoModelForSequenceClassification, ModernBertConfig |
|
|
|
from typing import Dict, Any |
|
import yaml |
|
|
|
class SentimentInference: |
|
def __init__(self, config_path: str = "config.yaml"): |
|
"""Load configuration and initialize model and tokenizer from Hugging Face Hub.""" |
|
with open(config_path, 'r') as f: |
|
config_data = yaml.safe_load(f) |
|
|
|
model_yaml_cfg = config_data.get('model', {}) |
|
inference_yaml_cfg = config_data.get('inference', {}) |
|
|
|
model_hf_repo_id = model_yaml_cfg.get('name_or_path') |
|
if not model_hf_repo_id: |
|
raise ValueError("model.name_or_path must be specified in config.yaml (e.g., 'username/model_name')") |
|
|
|
tokenizer_hf_repo_id = model_yaml_cfg.get('tokenizer_name_or_path', model_hf_repo_id) |
|
|
|
self.max_length = inference_yaml_cfg.get('max_length', model_yaml_cfg.get('max_length', 512)) |
|
|
|
print(f"Loading tokenizer from: {tokenizer_hf_repo_id}") |
|
self.tokenizer = AutoTokenizer.from_pretrained(tokenizer_hf_repo_id) |
|
|
|
print(f"Loading base ModernBertConfig from: {model_hf_repo_id}") |
|
|
|
|
|
|
|
loaded_config = ModernBertConfig.from_pretrained(model_hf_repo_id) |
|
|
|
|
|
|
|
loaded_config.pooling_strategy = model_yaml_cfg.get('pooling_strategy', 'mean') |
|
loaded_config.num_weighted_layers = model_yaml_cfg.get('num_weighted_layers', 4) |
|
loaded_config.classifier_dropout = model_yaml_cfg.get('dropout') |
|
|
|
|
|
loaded_config.num_labels = model_yaml_cfg.get('num_labels', 1) |
|
|
|
|
|
|
|
|
|
|
|
print(f"Instantiating and loading model weights for {model_hf_repo_id}...") |
|
|
|
|
|
self.model = AutoModelForSequenceClassification.from_pretrained( |
|
model_hf_repo_id, |
|
config=loaded_config, |
|
trust_remote_code=True |
|
) |
|
self.model.eval() |
|
print(f"Model {model_hf_repo_id} loaded successfully from Hugging Face Hub.") |
|
|
|
def predict(self, text: str) -> Dict[str, Any]: |
|
inputs = self.tokenizer(text, return_tensors="pt", truncation=True, max_length=self.max_length, padding=True) |
|
with torch.no_grad(): |
|
outputs = self.model(input_ids=inputs['input_ids'], attention_mask=inputs['attention_mask']) |
|
logits = outputs.get("logits") |
|
if logits is None: |
|
raise ValueError("Model output did not contain 'logits'. Check model's forward pass.") |
|
prob = torch.sigmoid(logits).item() |
|
return {"sentiment": "positive" if prob > 0.5 else "negative", "confidence": prob} |