← All Models | GPT - OpenAI's Generative Pre-trained Transformers:

openai/whisper-base

📊 Model Parameters

Total Parameters 78,558,208
Context Length 2,048
Hidden Size 512
Layers 6
Attention Heads 8
KV Heads 8

💾 Memory Requirements

FP32 (Full) 299.7 MB
FP16 (Half) 149.8 MB
INT8 (Quantized) 74.9 MB
INT4 (Quantized) 37.5 MB

🔑 KV Cache (Inference)

Per Token (FP16) 12.29 KB
Max Context FP32 48.0 MB
Max Context FP16 24.0 MB
Max Context INT8 12.0 MB

⚙️ Model Configuration

Core Architecture

Vocabulary Size51,865
Number of Layers6

Attention Configuration

Attention Dropout0%
Tied EmbeddingsYes

Activation & Normalization

Activation Functiongelu

Special Tokens

BOS Token ID50,257
Pad Token ID50,257
EOS Token ID50257

Data Type

Model Dtypefloat32
Layer Types:
Attention
MLP/FFN
Normalization
Embedding