Hugging Face's logo

Octen
/
Octen-Embedding-8B

Sentence Similarity
sentence-transformers
Safetensors
English
Chinese
multilingual
qwen3
feature-extraction
embedding
text-embedding
retrieval
Model card Files
xet
Community
bflhc commited on
Commit
a8a83f7
·
1 Parent(s): e5ffd82

Add Key Highlights, Model List and Experimental Results sections to all Octen model READMEs

Browse files
Files changed (1) hide show
  1. README.md +70 -0
README.md CHANGED
@@ -20,6 +20,76 @@ base_model: Qwen/Qwen3-Embedding-8B
20
 
21
  Octen-Embedding-8B is a text embedding model designed for semantic search and retrieval tasks. This model is fine-tuned from [Qwen/Qwen3-Embedding-8B](https://huggingface.co/Qwen/Qwen3-Embedding-8B) and supports multiple languages, providing high-quality embeddings for various applications.
22
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
  ## Model Details
24
 
25
  - **Base Model**: [Qwen/Qwen3-Embedding-8B](https://huggingface.co/Qwen/Qwen3-Embedding-8B)
 
20
 
21
  Octen-Embedding-8B is a text embedding model designed for semantic search and retrieval tasks. This model is fine-tuned from [Qwen/Qwen3-Embedding-8B](https://huggingface.co/Qwen/Qwen3-Embedding-8B) and supports multiple languages, providing high-quality embeddings for various applications.
22
 
23
+ ## Key Highlights
24
+
25
+ ### 🥇 RTEB Leaderboard Champion (as of January 12, 2026)
26
+ - **Octen-Embedding-8B ranks #1 on the [RTEB Leaderboard](https://huggingface.co/spaces/mteb/leaderboard)** with Mean (Task) score of **0.8045**
27
+ - Excellent performance on both Public (0.7953) and Private (0.8157) datasets
28
+ - Demonstrates true generalization capability without overfitting to public benchmarks
29
+
30
+ ### Industry-Oriented Vertical Domain Expertise
31
+ - **Legal**: Legal document retrieval
32
+ - **Finance**: Financial reports, Q&A, and personal finance content
33
+ - **Healthcare**: Medical Q&A, clinical dialogues, and health consultations
34
+ - **Code**: Programming problems, code search, and SQL queries
35
+
36
+ ### Ultra-Long Context Support
37
+ - Supports up to **32,768 tokens** context length
38
+ - Suitable for processing long documents in legal, healthcare, and other domains
39
+ - High-dimensional embedding space for rich semantic representation
40
+
41
+ ### Multilingual Capability
42
+ - Supports **100+ languages**
43
+ - Includes various programming languages
44
+ - Strong multilingual, cross-lingual, and code retrieval capabilities
45
+
46
+ ---
47
+
48
+ ## Open Source Model List
49
+
50
+ | Model Type | Model | Size | Max Tokens | Embedding Dimensions | HuggingFace Link |
51
+ |------------|-------|------|------------|---------------------|------------------|
52
+ | Text Embedding | [Octen-Embedding-0.6B](https://huggingface.co/Octen/Octen-Embedding-0.6B) | 0.6B | 32,768 | 1024 | ✅ Available |
53
+ | Text Embedding | [Octen-Embedding-4B](https://huggingface.co/Octen/Octen-Embedding-4B) | 4.0B | 32,768 | 2560 | ✅ Available |
54
+ | Text Embedding | [Octen-Embedding-8B](https://huggingface.co/Octen/Octen-Embedding-8B) | 7.6B | 32,768 | 4096 | ✅ Available |
55
+
56
+ **Model Family Design**:
57
+ - **Octen-Embedding-8B**: Best performance, RTEB #1, for high-precision retrieval
58
+ - **Octen-Embedding-4B**: Best in 4B category, balanced performance and efficiency
59
+ - **Octen-Embedding-0.6B**: Lightweight deployment, suitable for edge devices and resource-constrained environments
60
+
61
+ ---
62
+
63
+ ## Experimental Results
64
+
65
+ ### RTEB Leaderboard (Overall Performance)
66
+
67
+ | Model | Embedding Dim | Max Tokens | Mean (Public) | Mean (Private) | Mean (Task) |
68
+ |-------|---------------|------------|---------------|----------------|-------------|
69
+ | **Octen-Embedding-8B** | **4096** | **32768** | **0.7953** | **0.8157** | **0.8045** |
70
+ | voyage-3-large | 1024 | 32000 | 0.7434 | 0.8277 | 0.7812 |
71
+ | gemini-embedding-001 | 3072 | 2048 | 0.7218 | 0.8075 | 0.7602 |
72
+ | **Octen-Embedding-4B** | **2560** | **32768** | **0.7747** | **0.7942** | **0.7834** |
73
+ | MoD-Embedding | 2560 | 32768 | 0.7642 | 0.7900 | 0.7758 |
74
+ | Qwen3-Embedding-8B | 4096 | 32768 | 0.7310 | 0.7838 | 0.7547 |
75
+ | **Octen-Embedding-0.6B** | **1024** | **32768** | **0.7241** | **-** | **-** |
76
+ | voyage-3.5 | 1024 | 32000 | 0.7139 | 0.8102 | 0.7571 |
77
+ | Cohere-embed-v4.0 | 1536 | 128000 | 0.6534 | 0.7943 | 0.7166 |
78
+ | jina-embeddings-v4 | 2048 | 32768 | 0.6652 | 0.7664 | 0.7105 |
79
+ | GritLM-7B | 4096 | 32768 | 0.6187 | 0.7385 | 0.6724 |
80
+ | text-embedding-3-large | 3072 | 8191 | 0.6110 | 0.7130 | 0.6567 |
81
+ | e5-mistral-7b-instruct | 4096 | 32768 | 0.5090 | 0.7091 | 0.5987 |
82
+ | NV-Embed-v2 | 4096 | 32768 | 0.5805 | 0.6691 | 0.6203 |
83
+ | snowflake-arctic-embed-l-v2.0 | 1024 | 8192 | 0.5395 | 0.7079 | 0.6150 |
84
+ | multilingual-e5-large-instruct | 1024 | 514 | 0.5478 | 0.6859 | 0.6097 |
85
+ | gte-multilingual-base | 768 | 8192 | 0.5291 | 0.6697 | 0.5921 |
86
+ | text-embedding-3-small | 1536 | 8191 | 0.5260 | 0.6630 | 0.5874 |
87
+ | bge-m3 | 1024 | 8194 | 0.5216 | 0.6726 | 0.5893 |
88
+ | Qwen3-Embedding-4B | 2560 | 32768 | - | 0.7711 | - |
89
+ | Qwen3-Embedding-0.6B | 1024 | 32768 | - | 0.7117 | - |
90
+
91
+ ---
92
+
93
  ## Model Details
94
 
95
  - **Base Model**: [Qwen/Qwen3-Embedding-8B](https://huggingface.co/Qwen/Qwen3-Embedding-8B)