anthonym21/qwen3-0.6b-rlcd-decision
text-classification model by anthonym21
Older snapshot: check the original source for current information.
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
A decision-only model: it reads one state and a list of typed questions, runs one forward pass, and returns a probability distribution over the options each question declares. It does not generate text. It is Qwen3-0.6B-Base after a short supervised warmup and 500 steps of reinforcement learning for calibrated decisions (RLCD), a REINFORCE loop under bandit feedback whose reward is the outcome minus the probability the model put on the option it chose. On the held-out test split it is as calibrated as the warmup it started from (ECE 0.021) and six accuracy points better (0.807 against 0.746); the same loop with a plain outcome reward ends at ECE 0.216 or worse. Code, training, evaluation and the loader live at github.com/anthony-maio/eve-rlcd. This repository is the export of runs/q-rlcd from that project, and it needs the rlcd package from that repository to load; it is not a stock transformers checkpoint with…
Read the full model card ↗ · Preview checked 2026-11-01T21:35:46.061Z
Inside the original model card — Document outline
- Qwen3-0.6B RLCD decision model
- What is in the repository
- The three primitives
- How it was trained
- Results
- Known-posterior probe
- Limitations
- Intended use and out-of-scope use
- Reproduce
- Credits
- Citation
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Documentation belongs to its respective authors. Reported project/model license: apache-2.0. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
anthonym21/qwen3-0.6b-rlcd-decision is a text-classification model repository published by anthonym21 on Hugging Face. A library was not reported.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- text-classification
- Library
- Not reported
- Recent downloads (30 days)
- 0
- Cumulative likes
- 0
- Hugging Face trending score
- 0
- Reported safetensors parameters
- 596,049,920
- Architecture
- Qwen3Model
- License
- apache-2.0
- Access
- Not gated by Hugging Face
- Created
- 2026-09-19T06:15:20.000Z
- Last modified
- 2026-09-19T06:22:52.000Z
Compatibility and lineage
Reported languages: en
Reported base models: Qwen/Qwen3-0.6B-Base
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-09-19T06:35:16.985Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.