Qwen3-4B ReST Fused — model card

base_model: Qwen/Qwen3-4B-Instruct-2507
language:
  - en
library_name: mlx
license: other
pipeline_tag: text-generation
tags:
  - posttrainllm
  - mlx
  - tool-calling
  - function-calling
  - agentic
  - rest
  - research-specialist

Qwen3-4B ReST Fused

Model description

A research specialist built by PostTrainLLM — a Mac-local LLM factory that post-trains small models for narrow agentic tasks. This model preserves the best measured Qwen3-4B agentic candidate produced by a teacher-free ReST (Reinforced Self-Training) loop. One ReST iteration recovered the out-of-domain breadth lost by narrow distillation while retaining the saturated file-operations gate.

It is a research specialist package, not the default Pace planner. The model speaks BFCL/OpenAI-style tool calls; Pace uses a different intent envelope and requires its own ship gate.

Base model: Qwen/Qwen3-4B-Instruct-2507
Precision: bf16
Format: fused HF/MLX safetensors directory (8 files, ~8 GB)
Architecture: Qwen3ForCausalLM
Compatibility: mlx_lm, posttrainllm serve --hf-dir

Eval results

All numbers are historical results recorded on 2026-06-17. Source: specialists/qwen3-4b-rest-fused/eval_report.json.

Suite	Stock 4B	ReST 4B	Delta	n
File-ops hard gate	0.58	1.00	+0.42	12
Out-of-domain breadth	0.596	0.65	+0.054	52

The breadth suite contains 52 held-out TradingBot, VehicleControlAPI, and TravelAPI tasks. The depth suite contains 12 file-ops tasks. ReST recovered breadth (65% vs stock 59.6%) while keeping the file-ops hard gate at 100% — the negative transfer that plagued the distilled-only variant was reversed.

Evidence quality caveat: the raw timing, memory, throughput, and trace artifacts from the historical run were not preserved. This package does not claim current latency, RAM, tok/s, or exact qualitative failure counts. A rerun is intentionally not implied by this metadata promotion.

Training recipe summary

Target: same GorillaFileSystem file-ops depth anchor as the distilled specialist, plus breadth recovery across TradingBot, VehicleControlAPI, and TravelAPI backends.
Method: teacher-free ReST — one iteration over checker-passing interleaved trajectories. No paid model API was used (training cost: $0). The checker validated tool-call correctness; passing trajectories were folded back into the training set.
Depth anchor: file-ops gold depth was preserved as an anchor so ReST could not regress the saturated 100% hard gate while recovering breadth.
Student: Qwen3-4B-Instruct-2507 bf16, fused into a single HF/MLX safetensors directory.
Decision: ship as a research specialist — do not use as the Pace default planner without re-distillation and a product-specific ship gate.

Limitations

Breadth suite not frontier-validated. The 65% breadth score is the rounded historical result. The 52-task slice was directly compared with stock but has not been ceiling-validated against a frontier model.
Raw predictions unavailable. Historical trace artifacts were not preserved, so a current qualitative failure review is not possible from this package alone.
No performance numbers. Latency, RAM, tok/s, and training duration were not recorded. Do not infer Mac-runtime performance from this artifact.
Not a Pace planner. Pace requires a different intent envelope and product-specific ship gate. Re-distill and evaluate on Pace’s intent envelope before considering any downstream promotion.
Multi-GB artifact. Weights are ~8 GB and live on Hugging Face Hub, not in git.

How to use

# Metadata check (no weights loaded)
python mlx_load.py

# Load weights into MLX arrays
python mlx_load.py --load

Serve with the PostTrainLLM CLI:

posttrainllm serve --hf-dir ./qwen3-4b-rest-fused

Or load with mlx_lm:

from mlx_lm import load, generate
model, tokenizer = load("posttrainllm/qwen3-4b-rest-fused")

Upload command

This model is already published. To re-upload or update metadata:

# Stage the public metadata surface (no token needed)
python3 scripts/plan_hf_artifact_upload.py \
  specialists/qwen3-4b-rest-fused \
  --repo-id posttrainllm/qwen3-4b-rest-fused

# Upload to Hugging Face Hub (requires HF login)
huggingface-cli upload posttrainllm/qwen3-4b-rest-fused \
  dist/hf-artifacts/qwen3-4b-rest-fused \
  --repo-type model