Mark/mesh-seed

Fork 0

[Priority 2] Ollama on unstable as formal mesh capability #5

New issue

Closed

opened 2026-03-10 20:45:24 +00:00 by Mark · 2 comments

Mark commented

2026-03-10 20:45:24 +00:00

Owner

Setup

Capability ID: llm_inference_local
Models: gemma3:27b, qwen2.5, llama3.2
Advertise to mesh via capability.advertised
Worker template with Ollama client

Economics

Cost: $0 (local hardware)
Latency: variable by model
Priority: use before paid APIs

See

DAY0_TECH_STACK.md (unstable specs)
IMMEDIATE_INTEGRATION.md (Action 3)

Register Ollama (on unstable) as mesh capability for LLM inference. ## Setup - Capability ID: llm_inference_local - Models: gemma3:27b, qwen2.5, llama3.2 - Advertise to mesh via capability.advertised - Worker template with Ollama client ## Economics - Cost: $0 (local hardware) - Latency: variable by model - Priority: use before paid APIs ## See - DAY0_TECH_STACK.md (unstable specs) - IMMEDIATE_INTEGRATION.md (Action 3)

Mark referenced this issue

2026-03-14 18:59:03 +00:00

[Phase 0] Foundation - MVP Completion #27

Mark commented

2026-03-14 19:00:07 +00:00

Author

Owner

Phase Tracking: This issue is part of Phase 0: Foundation

Status: 🟡 IN PROGRESS - Ollama client integrated, needs deployment

Dependencies:

Depends on: Ollama installed on unstable
Blocks: #12 (Real LLM execution)

Implementation:

runtime/capability_execution.py has Ollama integration
Model selection: gemma3:12b → qwen3.5-9b → qwen3.5-4b
Automatic model pulling with progress tracking
Falls back to stubs if Ollama unavailable

**Phase Tracking:** This issue is part of [Phase 0: Foundation](https://forgejo.tail018ac4.ts.net/Mark/mesh-seed/issues/27) **Status:** 🟡 IN PROGRESS - Ollama client integrated, needs deployment **Dependencies:** - Depends on: Ollama installed on unstable - Blocks: #12 (Real LLM execution) **Implementation:** - `runtime/capability_execution.py` has Ollama integration - Model selection: gemma3:12b → qwen3.5-9b → qwen3.5-4b - Automatic model pulling with progress tracking - Falls back to stubs if Ollama unavailable

Mark commented

2026-03-15 01:07:12 +00:00

Author

Owner

✅ IMPLEMENTED - Ollama Integration Complete

Status: Complete and operational

Implementation:

Ollama running on unstable.tail018ac4.ts.net:11434
24 models available including:
- qwen2.5:3b (primary for writer)
- qwen2.5:0.5b (fast fallback)
- gemma3:4b, gemma3:12b, gemma3:27b
- And 20+ more

Integration:

runtime/capability_execution.py - Ollama client integration
Automatic model selection based on intent
Fallback to stub generation if Ollama unavailable
Resource consumption tracking in ledger

Capability Registration:

Models advertised as llm_inference_local
Zero cost (local hardware)
Used before paid APIs

Tests: Working in production tests

Closing as implemented.

## ✅ IMPLEMENTED - Ollama Integration Complete **Status:** Complete and operational **Implementation:** - Ollama running on unstable.tail018ac4.ts.net:11434 - 24 models available including: - qwen2.5:3b (primary for writer) - qwen2.5:0.5b (fast fallback) - gemma3:4b, gemma3:12b, gemma3:27b - And 20+ more **Integration:** - `runtime/capability_execution.py` - Ollama client integration - Automatic model selection based on intent - Fallback to stub generation if Ollama unavailable - Resource consumption tracking in ledger **Capability Registration:** - Models advertised as `llm_inference_local` - Zero cost (local hardware) - Used before paid APIs **Tests:** Working in production tests Closing as implemented.

Mark closed this issue

2026-03-15 01:07:25 +00:00

~~Mark referenced this issue 2026-03-15 01:08:12 +00:00~~

[Meta] Project Status - March 2026 #32

Mark referenced this issue

2026-06-16 05:22:55 +00:00

EMBED-004: Conversational Memory with Embeddings #127

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

Mark/mesh-seed#5

No description provided.

Rows
Columns