Instructions to use SVECTOR-CORPORATION/Optrix-1-0257 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use SVECTOR-CORPORATION/Optrix-1-0257 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="SVECTOR-CORPORATION/Optrix-1-0257")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("SVECTOR-CORPORATION/Optrix-1-0257", dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use SVECTOR-CORPORATION/Optrix-1-0257 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "SVECTOR-CORPORATION/Optrix-1-0257"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SVECTOR-CORPORATION/Optrix-1-0257",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/SVECTOR-CORPORATION/Optrix-1-0257

SGLang

How to use SVECTOR-CORPORATION/Optrix-1-0257 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "SVECTOR-CORPORATION/Optrix-1-0257" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SVECTOR-CORPORATION/Optrix-1-0257",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "SVECTOR-CORPORATION/Optrix-1-0257" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "SVECTOR-CORPORATION/Optrix-1-0257",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use SVECTOR-CORPORATION/Optrix-1-0257 with Docker Model Runner:
```
docker model run hf.co/SVECTOR-CORPORATION/Optrix-1-0257
```

You need to agree to share your contact information to access this model

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Optrix-1-0257

Optrix-1-0257 is a base 1 billion parameter language model developed by SVECTOR for general-purpose language generation and understanding. Pretrained on a broad corpus, it provides a strong foundation for fine-tuning on tasks such as summarization, dialogue, and retrieval.

Key Features

1B parameter transformer architecture
Pretrained on a broad, diverse corpus
Optimized for efficient inference and low memory usage
Suitable for fine-tuning on a wide range of language tasks

Usage

With Hugging Face Transformers

from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("SVECTOR-CORPORATION/Optrix-1-0257")
model = AutoModelForCausalLM.from_pretrained("SVECTOR-CORPORATION/Optrix-1-0257")

inputs = tokenizer("What is AI?", return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=64)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

With llama.cpp (GGUF Format)

./main -m Optrix-1-0257 -p "What is AI?"

Ensure the model is available in GGUF format before inference.

Model Specifications

Developer: SVECTOR
Architecture: Custom transformer with Grouped-Query Attention (GQA)
Embedding Dimension: 2048
Layers: 16
Attention Heads: 32
Vocabulary Size: 128,256
Max Position Embeddings: 131,072
Positional Encoding: Rotary with dynamic scaling
Activation Function: GELU
Output Head: Tied linear projection
Languages: English, German, French, Spanish, Hindi, Portuguese, Thai, Italian, and others
Release Date: June 27, 2025

Architecture Overview

Embedding Layer: nn.Embedding(vocab_size, hidden_size)
Transformer Block (×16):
- nn.MultiheadAttention(batch_first=True)
- 2-layer MLP with GELU activation
- LayerNorm (pre-attention and pre-MLP)
Final LayerNorm
Output Layer: nn.Linear(hidden_size, vocab_size, bias=False)
Causal Masking: Left-to-right for autoregressive generation
Rotary Embeddings: Applied with dynamic scaling

Example Configuration

{
  "architectures": ["OptrixForCausalLM"],
  "hidden_size": 2048,
  "num_hidden_layers": 16,
  "num_attention_heads": 32,
  "vocab_size": 128256,
  "max_position_embeddings": 131072,
  "model_type": "Optrix-1-0257"
}

License & Contact

Use of this model is governed by the SVECTOR License. For inquiries, please contact SVECTOR.

Downloads last month: -

Safetensors

Model size

1B params

Tensor type

BF16