-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathseed.example.surql
More file actions
34 lines (26 loc) · 3.04 KB
/
Copy pathseed.example.surql
File metadata and controls
34 lines (26 loc) · 3.04 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
-- Example seed data for surrealdb-prototype
-- Replace with your own memories and concepts
-- Concepts
CREATE concept:project_alpha SET name = "Project Alpha", description = "Primary product — a CLI tool for local semantic search.";
CREATE concept:blog SET name = "Blog", description = "Static site for writing and project updates.";
CREATE concept:embedding SET name = "Embeddings", description = "Vector representations for semantic similarity search.";
CREATE concept:database SET name = "Database", description = "SurrealDB multi-model backend for memory storage.";
-- Memory entries
CREATE memory:setup SET kind = "daily_log", title = "Initial Setup", content = "Installed SurrealDB v2.6.1 on localhost. Created schema with memory, concept, and graph edge tables. BM25 full-text search working.", source = "daily-notes/day-one.md", tags = ["setup", "surrealdb"], created_at = d"2026-01-15T10:00:00Z";
CREATE memory:model_selection SET kind = "decision", title = "Embedding Model Selection", content = "Evaluated mxbai-embed-large, bge-large, and Qwen3-Embedding-4B. Selected Qwen3-Embedding-4B for best MTEB scores within 12GB VRAM budget. Apache 2.0 license, safetensors format.", source = "daily-notes/day-two.md", tags = ["embeddings", "gpu", "decision"], created_at = d"2026-01-16T14:00:00Z";
CREATE memory:vector_search SET kind = "lesson", title = "Vector Search Working", content = "Wired embedding server to SurrealDB. Batch embedded all memory entries. Cosine similarity search returns relevant results. ~14 entries/second on GPU.", source = "daily-notes/day-three.md", tags = ["search", "embeddings", "milestone"], created_at = d"2026-01-17T16:00:00Z";
CREATE memory:api_compat SET kind = "daily_log", title = "OpenAI-Compatible API Server", content = "Built a FastAPI server that exposes an OpenAI-compatible /v1/embeddings endpoint backed by the local GPU model. Any tool expecting the OpenAI embedding API can now use local inference.", source = "daily-notes/day-three.md", tags = ["api", "fastapi", "embeddings"], created_at = d"2026-01-17T18:00:00Z";
-- Graph: memory -> references -> concept
RELATE memory:setup->references->concept:database SET context = "SurrealDB is the storage backend";
RELATE memory:model_selection->references->concept:embedding SET context = "Choosing the embedding model";
RELATE memory:vector_search->references->concept:embedding SET context = "Embeddings powering search";
RELATE memory:vector_search->references->concept:database SET context = "SurrealDB vector index";
RELATE memory:api_compat->references->concept:embedding SET context = "Local GPU embedding API";
-- Graph: concept -> related_to -> concept
RELATE concept:embedding->related_to->concept:database SET relation = "stored_in";
RELATE concept:project_alpha->related_to->concept:database SET relation = "built_on";
RELATE concept:blog->related_to->concept:project_alpha SET relation = "documents";
-- Temporal chain
RELATE memory:setup->follows->memory:model_selection;
RELATE memory:model_selection->follows->memory:vector_search;
RELATE memory:vector_search->follows->memory:api_compat;