GPT-2 — ONNX
ONNX export of GPT-2 (124M parameters) with KV cache support for efficient autoregressive generation.
Converted for use with inference4j, an inference-only AI library for Java.
Original Source
- Repository: OpenAI
- License: MIT
Usage with inference4j
try (Gpt2TextGenerator gen = Gpt2TextGenerator.builder().build()) {
GenerationResult result = gen.generate("Once upon a time");
System.out.println(result.text());
}
Model Details
| Property | Value |
|---|---|
| Architecture | GPT-2 (124M parameters, 12 layers, 768 hidden, 12 heads) |
| Task | Text generation |
| Context length | 1024 tokens |
| Vocabulary | 50257 tokens (BPE) |
| Original framework | PyTorch (transformers) |
| Export method | Hugging Face Optimum (with KV cache) |
License
This model is licensed under the MIT License. Original model by OpenAI.