cstr/octen-0.6b-GGUF

octen-0.6b GGUF

GGUF format of Octen/Octen-Embedding-0.6B for use with CrispEmbed and Ollama.

Files

File	Quantization	Size
octen-0.6b-q4_k.gguf	Q4_K	0 MB
octen-0.6b-q8_0.gguf	Q8_0	0 MB
octen-0.6b.gguf	F32	0 MB

Recommended: Q8_0 for quality (cos vs HF: L2=1.0), Q4_K for size (L2=1.0).

Quick Start

CrispEmbed

./crispembed -m octen-0.6b "Hello world"
./crispembed-server -m octen-0.6b --port 8080

Ollama (with CrispStrobe fork)

echo "FROM octen-0.6b-q8_0.gguf" > Modelfile
ollama create octen-0.6b -f Modelfile
curl http://localhost:11434/api/embed -d '{"model":"octen-0.6b","input":["Hello world"]}'

Python (CrispEmbed)

from crispembed import CrispEmbed
model = CrispEmbed("octen-0.6b-q8_0.gguf")
vectors = model.encode(["Hello world", "Goodbye world"])

Model Details

Property	Value
Architecture	Qwen3
Parameters	600M
Embedding Dimension	1024
Layers	28
Pooling	last-token
Tokenizer	BPE
Language	multilingual
Q8_0 vs HuggingFace	L2=1.0
Q4_K vs HuggingFace	L2=1.0

Server API

CrispEmbed server supports four API dialects:

• POST /embed -- native
• POST /v1/embeddings -- OpenAI-compatible
• POST /api/embed -- Ollama-compatible
• POST /api/embeddings -- Ollama legacy

Credits

• Original model: Octen/Octen-Embedding-0.6B
• Inference: CrispEmbed (MIT, ggml-based)