hindi-embedding-foundational-model / hindi_embeddings.py
DeepMostInnovations's picture
Add inference script
f40b58a verified
import os
import torch
import json
import numpy as np
from torch import nn
from torch.nn import functional as F
import sentencepiece as spm
from sklearn.metrics.pairwise import cosine_similarity
from tqdm import tqdm
import matplotlib.pyplot as plt
from sklearn.manifold import TSNE
# Tokenizer wrapper class
class SentencePieceTokenizerWrapper:
def __init__(self, sp_model_path):
self.sp_model = spm.SentencePieceProcessor()
self.sp_model.Load(sp_model_path)
self.vocab_size = self.sp_model.GetPieceSize()
# Special token IDs from tokenizer training
self.pad_token_id = 0
self.bos_token_id = 1
self.eos_token_id = 2
self.unk_token_id = 3
# Set special tokens
self.pad_token = "<pad>"
self.bos_token = "<s>"
self.eos_token = "</s>"
self.unk_token = "<unk>"
self.mask_token = "<mask>"
def __call__(self, text, padding=False, truncation=False, max_length=None, return_tensors=None):
# Handle both string and list inputs
if isinstance(text, str):
# Encode a single string
ids = self.sp_model.EncodeAsIds(text)
# Handle truncation
if truncation and max_length and len(ids) > max_length:
ids = ids[:max_length]
attention_mask = [1] * len(ids)
# Handle padding
if padding and max_length:
padding_length = max(0, max_length - len(ids))
ids = ids + [self.pad_token_id] * padding_length
attention_mask = attention_mask + [0] * padding_length
result = {
'input_ids': ids,
'attention_mask': attention_mask
}
# Convert to tensors if requested
if return_tensors == 'pt':
import torch
result = {k: torch.tensor([v]) for k, v in result.items()}
return result
# Process a batch of texts
batch_encoded = [self.sp_model.EncodeAsIds(t) for t in text]
# Apply truncation if needed
if truncation and max_length:
batch_encoded = [ids[:max_length] for ids in batch_encoded]
# Create attention masks
batch_attention_mask = [[1] * len(ids) for ids in batch_encoded]
# Apply padding if needed
if padding:
if max_length:
max_len = max_length
else:
max_len = max(len(ids) for ids in batch_encoded)
# Pad sequences to max_len
batch_encoded = [ids + [self.pad_token_id] * (max_len - len(ids)) for ids in batch_encoded]
batch_attention_mask = [mask + [0] * (max_len - len(mask)) for mask in batch_attention_mask]
result = {
'input_ids': batch_encoded,
'attention_mask': batch_attention_mask
}
# Convert to tensors if requested
if return_tensors == 'pt':
import torch
result = {k: torch.tensor(v) for k, v in result.items()}
return result
# Model architecture components
class MultiHeadAttention(nn.Module):
"""Multi-headed attention mechanism"""
def __init__(self, config):
super().__init__()
self.num_attention_heads = config["num_attention_heads"]
self.attention_head_size = config["hidden_size"] // config["num_attention_heads"]
self.all_head_size = self.num_attention_heads * self.attention_head_size
# Query, Key, Value projections
self.query = nn.Linear(config["hidden_size"], self.all_head_size)
self.key = nn.Linear(config["hidden_size"], self.all_head_size)
self.value = nn.Linear(config["hidden_size"], self.all_head_size)
# Output projection
self.output = nn.Sequential(
nn.Linear(self.all_head_size, config["hidden_size"]),
nn.Dropout(config["attention_probs_dropout_prob"])
)
# Simplified relative position bias
self.max_position_embeddings = config["max_position_embeddings"]
self.relative_attention_bias = nn.Embedding(
2 * config["max_position_embeddings"] - 1,
config["num_attention_heads"]
)
def transpose_for_scores(self, x):
new_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size)
x = x.view(*new_shape)
return x.permute(0, 2, 1, 3)
def forward(self, hidden_states, attention_mask=None):
batch_size, seq_length = hidden_states.size()[:2]
# Project inputs to queries, keys, and values
query_layer = self.transpose_for_scores(self.query(hidden_states))
key_layer = self.transpose_for_scores(self.key(hidden_states))
value_layer = self.transpose_for_scores(self.value(hidden_states))
# Take the dot product between query and key to get the raw attention scores
attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))
# Generate relative position matrix
position_ids = torch.arange(seq_length, dtype=torch.long, device=hidden_states.device)
relative_position = position_ids.unsqueeze(1) - position_ids.unsqueeze(0) # [seq_len, seq_len]
# Shift values to be >= 0
relative_position = relative_position + self.max_position_embeddings - 1
# Ensure indices are within bounds
relative_position = torch.clamp(relative_position, 0, 2 * self.max_position_embeddings - 2)
# Get relative position embeddings [seq_len, seq_len, num_heads]
rel_attn_bias = self.relative_attention_bias(relative_position) # [seq_len, seq_len, num_heads]
# Reshape to add to attention heads [1, num_heads, seq_len, seq_len]
rel_attn_bias = rel_attn_bias.permute(2, 0, 1).unsqueeze(0)
# Add to attention scores - now dimensions will match
attention_scores = attention_scores + rel_attn_bias
# Scale attention scores
attention_scores = attention_scores / (self.attention_head_size ** 0.5)
# Apply attention mask
if attention_mask is not None:
attention_scores = attention_scores + attention_mask
# Normalize the attention scores to probabilities
attention_probs = F.softmax(attention_scores, dim=-1)
# Apply dropout
attention_probs = F.dropout(attention_probs, p=0.1, training=self.training)
# Apply attention to values
context_layer = torch.matmul(attention_probs, value_layer)
# Reshape back to [batch_size, seq_length, hidden_size]
context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
new_shape = context_layer.size()[:-2] + (self.all_head_size,)
context_layer = context_layer.view(*new_shape)
# Final output projection
output = self.output(context_layer)
return output
class EnhancedTransformerLayer(nn.Module):
"""Advanced transformer layer with pre-layer norm and enhanced attention"""
def __init__(self, config):
super().__init__()
self.attention_pre_norm = nn.LayerNorm(config["hidden_size"], eps=config["layer_norm_eps"])
self.attention = MultiHeadAttention(config)
self.ffn_pre_norm = nn.LayerNorm(config["hidden_size"], eps=config["layer_norm_eps"])
# Feed-forward network
self.ffn = nn.Sequential(
nn.Linear(config["hidden_size"], config["intermediate_size"]),
nn.GELU(),
nn.Dropout(config["hidden_dropout_prob"]),
nn.Linear(config["intermediate_size"], config["hidden_size"]),
nn.Dropout(config["hidden_dropout_prob"])
)
def forward(self, hidden_states, attention_mask=None):
# Pre-layer norm for attention
attn_norm_hidden = self.attention_pre_norm(hidden_states)
# Self-attention
attention_output = self.attention(attn_norm_hidden, attention_mask)
# Residual connection
hidden_states = hidden_states + attention_output
# Pre-layer norm for feed-forward
ffn_norm_hidden = self.ffn_pre_norm(hidden_states)
# Feed-forward
ffn_output = self.ffn(ffn_norm_hidden)
# Residual connection
hidden_states = hidden_states + ffn_output
return hidden_states
class AdvancedTransformerModel(nn.Module):
"""Advanced Transformer model for inference"""
def __init__(self, config):
super().__init__()
self.config = config
# Embeddings
self.word_embeddings = nn.Embedding(
config["vocab_size"],
config["hidden_size"],
padding_idx=config["pad_token_id"]
)
# Position embeddings
self.position_embeddings = nn.Embedding(config["max_position_embeddings"], config["hidden_size"])
# Embedding dropout
self.embedding_dropout = nn.Dropout(config["hidden_dropout_prob"])
# Transformer layers
self.layers = nn.ModuleList([
EnhancedTransformerLayer(config) for _ in range(config["num_hidden_layers"])
])
# Final layer norm
self.final_layer_norm = nn.LayerNorm(config["hidden_size"], eps=config["layer_norm_eps"])
def forward(self, input_ids, attention_mask=None):
input_shape = input_ids.size()
batch_size, seq_length = input_shape
# Get position ids
position_ids = torch.arange(seq_length, dtype=torch.long, device=input_ids.device)
position_ids = position_ids.unsqueeze(0).expand(batch_size, -1)
# Get embeddings
word_embeds = self.word_embeddings(input_ids)
position_embeds = self.position_embeddings(position_ids)
# Sum embeddings
embeddings = word_embeds + position_embeds
# Apply dropout
embeddings = self.embedding_dropout(embeddings)
# Default attention mask
if attention_mask is None:
attention_mask = torch.ones(input_shape, device=input_ids.device)
# Extended attention mask for transformer layers (1 for tokens to attend to, 0 for masked tokens)
extended_attention_mask = attention_mask.unsqueeze(1).unsqueeze(2)
extended_attention_mask = (1.0 - extended_attention_mask) * -10000.0
# Apply transformer layers
hidden_states = embeddings
for layer in self.layers:
hidden_states = layer(hidden_states, extended_attention_mask)
# Final layer norm
hidden_states = self.final_layer_norm(hidden_states)
return hidden_states
class AdvancedPooling(nn.Module):
"""Advanced pooling module supporting multiple pooling strategies"""
def __init__(self, config):
super().__init__()
self.pooling_mode = config["pooling_mode"] # 'mean', 'max', 'cls', 'attention'
self.hidden_size = config["hidden_size"]
# For attention pooling
if self.pooling_mode == 'attention':
self.attention_weights = nn.Linear(config["hidden_size"], 1)
# For weighted pooling
elif self.pooling_mode == 'weighted':
self.weight_layer = nn.Linear(config["hidden_size"], 1)
def forward(self, token_embeddings, attention_mask=None):
if attention_mask is None:
attention_mask = torch.ones_like(token_embeddings[:, :, 0])
mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
if self.pooling_mode == 'cls':
# Use [CLS] token (first token)
pooled = token_embeddings[:, 0]
elif self.pooling_mode == 'max':
# Max pooling
token_embeddings = token_embeddings.clone()
# Set padding tokens to large negative value to exclude them from max
token_embeddings[mask_expanded == 0] = -1e9
pooled = torch.max(token_embeddings, dim=1)[0]
elif self.pooling_mode == 'attention':
# Attention pooling
weights = self.attention_weights(token_embeddings).squeeze(-1)
# Mask out padding tokens
weights = weights.masked_fill(attention_mask == 0, -1e9)
weights = F.softmax(weights, dim=1).unsqueeze(-1)
pooled = torch.sum(token_embeddings * weights, dim=1)
elif self.pooling_mode == 'weighted':
# Weighted average pooling
weights = torch.sigmoid(self.weight_layer(token_embeddings)).squeeze(-1)
# Apply mask
weights = weights * attention_mask
# Normalize weights
sum_weights = torch.sum(weights, dim=1, keepdim=True)
sum_weights = torch.clamp(sum_weights, min=1e-9)
weights = weights / sum_weights
# Apply weights
pooled = torch.sum(token_embeddings * weights.unsqueeze(-1), dim=1)
else: # Default to mean pooling
# Mean pooling
sum_embeddings = torch.sum(token_embeddings * mask_expanded, dim=1)
sum_mask = torch.clamp(mask_expanded.sum(1), min=1e-9)
pooled = sum_embeddings / sum_mask
# L2 normalize
pooled = F.normalize(pooled, p=2, dim=1)
return pooled
class SentenceEmbeddingModel(nn.Module):
"""Complete sentence embedding model for inference"""
def __init__(self, config):
super(SentenceEmbeddingModel, self).__init__()
self.config = config
# Create transformer model
self.transformer = AdvancedTransformerModel(config)
# Create pooling module
self.pooling = AdvancedPooling(config)
# Build projection module if needed
if "projection_dim" in config and config["projection_dim"] > 0:
self.use_projection = True
self.projection = nn.Sequential(
nn.Linear(config["hidden_size"], config["hidden_size"]),
nn.GELU(),
nn.Linear(config["hidden_size"], config["projection_dim"]),
nn.LayerNorm(config["projection_dim"], eps=config["layer_norm_eps"])
)
else:
self.use_projection = False
def forward(self, input_ids, attention_mask=None):
# Get token embeddings from transformer
token_embeddings = self.transformer(input_ids, attention_mask)
# Pool token embeddings
pooled_output = self.pooling(token_embeddings, attention_mask)
# Apply projection if enabled
if self.use_projection:
pooled_output = self.projection(pooled_output)
pooled_output = F.normalize(pooled_output, p=2, dim=1)
return pooled_output
class HindiEmbedder:
def __init__(self, model_path="/home/ubuntu/output/hindi-embeddings-custom-tokenizer/final"):
"""
Initialize the Hindi sentence embedder.
Args:
model_path: Path to the model directory
"""
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print(f"Using device: {self.device}")
# Load tokenizer - look for it in the model directory
tokenizer_path = os.path.join(model_path, "tokenizer.model")
if not os.path.exists(tokenizer_path):
raise FileNotFoundError(f"Could not find tokenizer at {tokenizer_path}")
self.tokenizer = SentencePieceTokenizerWrapper(tokenizer_path)
print(f"Loaded tokenizer from {tokenizer_path} with vocabulary size: {self.tokenizer.vocab_size}")
# Load model config
config_path = os.path.join(model_path, "config.json")
with open(config_path, "r") as f:
self.config = json.load(f)
print(f"Loaded model config with hidden_size={self.config['hidden_size']}")
# Load model
model_pt_path = os.path.join(model_path, "embedding_model.pt")
try:
# Support both PyTorch 2.6+ and older versions
try:
checkpoint = torch.load(model_pt_path, map_location=self.device, weights_only=False)
print("Loaded model using PyTorch 2.6+ style loading")
except TypeError:
checkpoint = torch.load(model_pt_path, map_location=self.device)
print("Loaded model using older PyTorch style loading")
# Create model
self.model = SentenceEmbeddingModel(self.config)
# Load state dict
if "model_state_dict" in checkpoint:
state_dict = checkpoint["model_state_dict"]
else:
state_dict = checkpoint
missing_keys, unexpected_keys = self.model.load_state_dict(state_dict, strict=False)
print(f"Loaded model with {len(missing_keys)} missing keys and {len(unexpected_keys)} unexpected keys")
# Move to device
self.model.to(self.device)
self.model.eval()
print("Model loaded successfully and placed in evaluation mode")
except Exception as e:
print(f"Error loading model: {e}")
raise RuntimeError(f"Failed to load the model: {e}")
def encode(self, sentences, batch_size=32, normalize=True):
"""
Encode sentences to embeddings.
Args:
sentences: A string or list of strings to encode
batch_size: Batch size for encoding
normalize: Whether to normalize the embeddings
Returns:
Numpy array of embeddings
"""
# Handle single sentence
if isinstance(sentences, str):
sentences = [sentences]
all_embeddings = []
# Process in batches
with torch.no_grad():
for i in range(0, len(sentences), batch_size):
batch = sentences[i:i+batch_size]
# Tokenize
inputs = self.tokenizer(
batch,
padding=True,
truncation=True,
max_length=self.config.get("max_position_embeddings", 128),
return_tensors="pt"
)
# Move to device
input_ids = inputs["input_ids"].to(self.device)
attention_mask = inputs["attention_mask"].to(self.device)
# Get embeddings
embeddings = self.model(input_ids, attention_mask)
# Move to CPU and convert to numpy
all_embeddings.append(embeddings.cpu().numpy())
# Concatenate all embeddings
all_embeddings = np.vstack(all_embeddings)
# Normalize if requested
if normalize:
all_embeddings = all_embeddings / np.linalg.norm(all_embeddings, axis=1, keepdims=True)
return all_embeddings
def compute_similarity(self, texts1, texts2=None):
"""
Compute cosine similarity between texts.
Args:
texts1: First set of texts
texts2: Second set of texts. If None, compute similarity matrix within texts1.
Returns:
Similarity scores
"""
# Convert single strings to lists for consistent handling
if isinstance(texts1, str):
texts1 = [texts1]
if texts2 is not None and isinstance(texts2, str):
texts2 = [texts2]
embeddings1 = self.encode(texts1)
if texts2 is None:
# Compute similarity matrix within texts1
similarities = cosine_similarity(embeddings1)
return similarities
else:
# Compute similarity between texts1 and texts2
embeddings2 = self.encode(texts2)
if len(texts1) == len(texts2):
# Compute pairwise similarity when the number of texts match
similarities = np.array([
cosine_similarity([e1], [e2])[0][0]
for e1, e2 in zip(embeddings1, embeddings2)
])
# If there's just one pair, return a scalar
if len(similarities) == 1:
return similarities[0]
return similarities
else:
# Return full similarity matrix
return cosine_similarity(embeddings1, embeddings2)
def search(self, query, documents, top_k=5):
"""
Search for similar documents to a query.
Args:
query: The query text
documents: List of documents to search
top_k: Number of top results to return
Returns:
List of dictionaries with document and score
"""
# Get embeddings
query_embedding = self.encode([query])[0]
document_embeddings = self.encode(documents)
# Compute similarities
similarities = np.dot(document_embeddings, query_embedding)
# Get top indices
top_indices = np.argsort(similarities)[-top_k:][::-1]
# Return results
results = []
for idx in top_indices:
results.append({
"document": documents[idx],
"score": float(similarities[idx])
})
return results
def evaluate_similarity_samples(self):
"""Evaluate model on some standard similarity examples for Hindi"""
test_pairs = [
(
"मुझे हिंदी में पढ़ना बहुत पसंद है।",
"मैं हिंदी किताबें बहुत पसंद करता हूँ।"
),
(
"आज मौसम बहुत अच्छा है।",
"आज बारिश हो रही है।"
),
(
"भारत एक विशाल देश है।",
"भारत में कई भाषाएँ बोली जाती हैं।"
),
(
"कंप्यूटर विज्ञान एक रोचक विषय है।",
"मैं कंप्यूटर साइंस का छात्र हूँ।"
),
(
"मैं रोज सुबह योग करता हूँ।",
"स्वस्थ रहने के लिए व्यायाम जरूरी है।"
),
# Add contrasting pairs to test discrimination
(
"मुझे हिंदी में पढ़ना बहुत पसंद है।",
"क्रिकेट भारत में सबसे लोकप्रिय खेल है।"
),
(
"आज मौसम बहुत अच्छा है।",
"भारतीय व्यंजन दुनिया भर में मशहूर हैं।"
),
(
"कंप्यूटर विज्ञान एक रोचक विषय है।",
"हिमालय दुनिया का सबसे ऊंचा पर्वत है।"
)
]
print("Evaluating model on standard similarity samples:")
for i, (text1, text2) in enumerate(test_pairs):
similarity = self.compute_similarity([text1], [text2])[0]
print(f"\nPair {i+1}:")
print(f" Sentence 1: {text1}")
print(f" Sentence 2: {text2}")
print(f" Similarity: {similarity:.4f}")
return
def visualize_embeddings(self, sentences, labels=None, output_path="hindi_embeddings_visualization.png"):
"""
Create a t-SNE visualization of the embeddings.
Args:
sentences: List of sentences to visualize
labels: Optional list of labels for the points
output_path: Path to save the visualization
Returns:
Path to the saved visualization
"""
# Encode sentences
embeddings = self.encode(sentences)
# Apply t-SNE
tsne = TSNE(n_components=2, random_state=42, perplexity=min(30, len(embeddings)-1))
reduced_embeddings = tsne.fit_transform(embeddings)
# Create plot
plt.figure(figsize=(12, 10))
# Plot points
scatter = plt.scatter(
reduced_embeddings[:, 0],
reduced_embeddings[:, 1],
c=range(len(reduced_embeddings)),
cmap='viridis',
alpha=0.8,
s=100
)
# Add labels if provided
if labels:
for i, label in enumerate(labels):
plt.annotate(
label,
(reduced_embeddings[i, 0], reduced_embeddings[i, 1]),
fontsize=10,
alpha=0.7
)
plt.title("t-SNE Visualization of Hindi Sentence Embeddings", fontsize=16)
plt.xlabel("Dimension 1", fontsize=12)
plt.ylabel("Dimension 2", fontsize=12)
plt.colorbar(scatter, label="Sentence Index")
plt.grid(alpha=0.3)
# Save the figure
plt.tight_layout()
plt.savefig(output_path, dpi=300, bbox_inches='tight')
plt.close()
print(f"Visualization saved to {output_path}")
return output_path
def main():
# Create embedder
embedder = HindiEmbedder()
# Run sample evaluation
embedder.evaluate_similarity_samples()
# Example of semantic search
print("\nSemantic Search Example:")
query = "भारत की संस्कृति"
documents = [
"भारतीय संस्कृति दुनिया की सबसे प्राचीन संस्कृतियों में से एक है।",
"भारत की आबादी 1.3 अरब से अधिक है।",
"हिमालय पर्वत श्रृंखला भारत के उत्तर में स्थित है।",
"भारतीय व्यंजन में मसालों का प्रयोग किया जाता है।",
"भारत में 22 आधिकारिक भाषाएँ हैं।",
"संस्कृति लोगों के रहन-सहन का तरीका है।",
"भारत के विभिन्न राज्यों की अपनी अलग संस्कृति है।",
"रामायण और महाभारत भारतीय संस्कृति के महत्वपूर्ण हिस्से हैं।",
]
results = embedder.search(query, documents)
print(f"Query: {query}")
print("Top results:")
for i, result in enumerate(results):
print(f"{i+1}. Score: {result['score']:.4f}")
print(f" {result['document']}")
# Create visualization example
print("\nCreating embedding visualization...")
visualization_sentences = [
"मुझे हिंदी में पढ़ना बहुत पसंद है।",
"मैं हिंदी किताबें बहुत पसंद करता हूँ।",
"आज मौसम बहुत अच्छा है।",
"आज बारिश हो रही है।",
"भारत एक विशाल देश है।",
"भारत में कई भाषाएँ बोली जाती हैं।",
"कंप्यूटर विज्ञान एक रोचक विषय है।",
"मैं कंप्यूटर साइंस का छात्र हूँ।",
"क्रिकेट भारत में सबसे लोकप्रिय खेल है।",
"भारतीय व्यंजन दुनिया भर में मशहूर हैं।",
"हिमालय दुनिया का सबसे ऊंचा पर्वत है।",
"गंगा भारत की सबसे पवित्र नदी है।",
"दिल्ली भारत की राजधानी है।",
"मुंबई भारत का आर्थिक केंद्र है।",
"तमिल, तेलुगु, कन्नड़ और मलयालम दक्षिण भारत की प्रमुख भाषाएँ हैं।"
]
labels = ["पढ़ना", "किताबें", "मौसम", "बारिश", "भारत", "भाषाएँ", "कंप्यूटर",
"छात्र", "क्रिकेट", "व्यंजन", "हिमालय", "गंगा", "दिल्ली", "मुंबई", "भाषाएँ"]
embedder.visualize_embeddings(visualization_sentences, labels)
if __name__ == "__main__":
main()