OSGeo-Knowledge-Embedding-Server - Revision history

Ominiverdi at 18:15, 5 June 2026

2026-06-05T18:15:14Z

Ominiverdi at 17:59, 5 June 2026

2026-06-05T17:59:06Z

Ominiverdi: Ominiverdi moved page BG3-M3-Embedding-Server to OSGeo-Knowledge-Embedding-Server

2026-06-05T17:02:44Z

Ominiverdi moved page BG3-M3-Embedding-Server to OSGeo-Knowledge-Embedding-Server

Strk at 17:00, 5 June 2026

2026-06-05T17:00:48Z

Strk at 16:57, 5 June 2026

2026-06-05T16:57:03Z

Strk: Created page with "* '''BGE-M3 Embedding Server''' - Semantic search embeddings via llama.cpp ** Port: 8094 (localhost only), Model: bge-m3-Q8_0.gguf Runs in Gallery_Container The embeddin..."

2026-06-05T16:54:58Z

Created page with "* '''BGE-M3 Embedding Server''' - Semantic search embeddings via llama.cpp ** Port: 8094 (localhost only), Model: bge-m3-Q8_0.gguf Runs in Gallery_Container The embeddin..."

New page

* '''BGE-M3 Embedding Server''' - Semantic search embeddings via llama.cpp
** Port: 8094 (localhost only), Model: bge-m3-Q8_0.gguf

Runs in [[Gallery_Container]]

The embedding model running on osgeo7-gallery is BGE-M3 (bge-m3-Q8_0.gguf), a multilingual model from BAAI supporting 100+ languages. It produces 1024-dimensional vectors and runs via llama-server on port 8094 (localhost only). The Q8_0 quantization keeps it light: 606 MB on disk, ~457 MB RAM, ~12ms per query.

'''Contact:''' ominiverdi (Lorenzo Becchi) or SAC channel

[[Category:Service]]

@@ Line 1: / Line 1: @@
- OSGeo-Knowledge-Embedding-Server =
+= OSGeo-Knowledge-Embedding-Server =
-BGE-M3 embedding server for semantic search, running on [[Gallery_Container]] when active.
+BGE-M3 embedding server for semantic search, running on [[Gallery_Container]].
-It provides local embeddings for services such as [[Osgeo-library]] and other OSGeo AI/search tooling.
+It provides local embeddings for [[Osgeo-library]] and other OSGeo AI/search tooling.
 == Model ==
@@ Line 11: / Line 11: @@
 <pre>
 /home/ominiverdi/models/bge-m3-Q8_0.gguf
 </pre>
@@ Line 24: / Line 18: @@
 * Upstream: BAAI
 * Quantization: Q8_0
 * Vector size: 1024 dimensions
 * Multilingual: supports 100+ languages
@@ Line 29: / Line 24: @@
 == Endpoint ==
-Configured by [[Osgeo-library]] as:
+The service listens on localhost only:
 <pre>
@@ Line 35: / Line 30: @@
 </pre>
-Verified config files:
+Configured by [[Osgeo-library]] in:
 <pre>
 /home/ominiverdi/github/osgeo-library/config.example.toml
 </pre>
-The endpoint is intended to listen on localhost only.
+== Runtime ==
-== Runtime ==
+<pre>
-The service is expected to run through llama.cpp / <code>llama-server</code>.
+== Operations ==
 Check listener status:
@@ Line 52: / Line 80: @@
 <pre>
 ss -ltnp | grep 8094
 </pre>
@@ Line 57: / Line 91: @@
 <pre>
-pgrep -af "llama|bge|8094|embedding"
+pgrep -af "llama-server.*8094|bge-m3"
 </pre>
-At the time of the last check, the model file and client configuration were present, but no active listener on port <code>8094</code> was observed.
+Start manually:
 == Consumers ==

@@ Line 1: / Line 1: @@
-''BGE-M3 Embedding Server''' - Semantic search embeddings via llama.cpp
+ OSGeo-Knowledge-Embedding-Server =
-Runs in [[Gallery_Container]]
+BGE-M3 embedding server for semantic search, running on [[Gallery_Container]] when active.
-Only listens to localhost device, port: 8094, Model: bge-m3-Q8_0.gguf
+It provides local embeddings for services such as [[Osgeo-library]] and other OSGeo AI/search tooling.
-The embedding model running on osgeo7-gallery is BGE-M3 (bge-m3-Q8_0.gguf), a multilingual model from BAAI supporting 100+ languages. It produces 1024-dimensional vectors and runs via llama-server on port 8094 (localhost only). The Q8_0 quantization keeps it light: 606 MB on disk, ~457 MB RAM, ~12ms per query.
+== Model ==
-'''Contact:''' ominiverdi (Lorenzo Becchi) or SAC channel
+Verified model file on <code>osgeo7-gallery</code>:
 [[Category:Services]]
 [[Category:AI-Services]]

← Older revision		Revision as of 16:57, 5 June 2026
Line 8:		Line 8:
	'''Contact:''' ominiverdi (Lorenzo Becchi) or SAC channel		'''Contact:''' ominiverdi (Lorenzo Becchi) or SAC channel

−	[[Category:~~Service~~]]	+	[[Category:Services]]
		+	[[Category:AI-Services]]

← Older revision	Revision as of 17:02, 5 June 2026
(No difference)