mempool Qwen3 0.6B Logits Orchestrator v1
This repository contains a usable checkpoint for the mempool
Qwen 0.6B logits-head orchestrator path with a deterministic held-out gate.
The checkpoint stores only the trained routing heads, not the Qwen base model weights. Load the base model separately and attach the heads.
- Checkpoint:
qwen_logits_heads.pt - Training rows:
53 - Final training loss:
2.474162220954895
Worker labels:
ollama-cloud-deepseek-v4-proollama-cloud-glm-5.2ollama-cloud-kimi-k2.7-codeollama-cloud-qwen3-coder-480b
Train-row evaluation:
- Worker accuracy:
0.8113207547169812 - Workflow accuracy:
0.8867924528301887 - Mean worker loss:
1.2073683018954295 - Mean workflow loss:
0.23352967146432624
Held-out evaluation:
- Worker accuracy:
0.6923076923076923 - Workflow accuracy:
0.7692307692307693 - Mean worker loss:
1.3018739131780772 - Mean workflow loss:
0.6372018387684455
A sample router prediction is included at sample_prediction.json.
This is a research artifact. It is usable for routing experiments, but it is not yet a promoted production policy.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("blazeofchi/mempool-qwen3-0p6b-logits-orchestrator-v1", dtype="auto")