Jet-4B
v6.2.0 · released 2026-09-26 · focused continuation, selected step 250.
Jet-4B, part of the Jet family, returns typed decisions and probabilities from supplied options without generating free-form answers.
This is the full merged BF16 model, not an adapter, fine-tuned from Qwen3.5-4B. No separate
base-model download is required. It replaces v6.1 in quaedra/jet-4b; previous
releases remain accessible by their version tags.
Run
Linux + NVIDIA CUDA, Python 3.12. Verified with PyTorch 2.11.0+cu128, Transformers 5.17.0 and flash-linear-attention 0.5.2.
hf download quaedra/jet-4b --revision v6.2.0 --local-dir jet-4b
cd jet-4b
python -m pip install -r requirements.txt
python jet.py <<'JSON'
{"state":"I was charged twice this month.","questions":{"topic":{"type":"choice","instructions":"What is the primary issue?","criteria":{"billing":"billing or payment problem","bug":"the product is broken"}}}}
JSON
from jet import Jet
model = Jet()
result = model.decide("The item arrived broken.", {
"damaged": {"type": "noul", "instructions": "Is the item damaged?"}
})
| Type | Input | Output |
|---|---|---|
choice | 2–255 named options | Selected key and probabilities |
score | 2–10 ordered levels | Expected zero-based score, selected level, probabilities |
noul | Yes/no question | Probability of yes |
Use the included native prompt and restricted label-token readout. A generic text-generation pipeline does not implement this API. Questions are processed separately. The runtime now accepts complete prompts up to 16,384 tokens; inputs and options are never silently truncated. The longest reconstructed API-Bank input (11,495 tokens) passed the standalone runtime check, but full API-Bank accuracy has not been measured for this release. Context feasibility does not establish long-context quality across tasks.
FLA is enabled; convolution uses the PyTorch fallback. confidence is normalized
inverse entropy, not a calibrated correctness probability. Temperatures are
inherited from v6.1 and were not refitted or established as calibrated for v6.2.
Training and selection
The parent is full merged Jet-4B v6.1 at
f446b82727be57da348bb46eccf211414294ab3e. Parent shard hashes and the exact
adapter hash are in merge-provenance.json. A fresh rank/alpha-16 correction LoRA
was trained on that parent; the model was not reset to the original Qwen weights.
- 4,000 examples: 800 banking intent, 800 entity-specific financial sentiment, 400 sarcasm/literal, and 2,000 broad retention examples.
- BF16 backbone / FP32 adapter, dropout .05, microbatch 1, accumulation 4, seed 260925; two trials of 1,000 updates at peak learning rates 3e-6 and 8e-6.
- Validation-selected winner: 3e-6 at step 250. The higher-rate trial failed the selection guards and is not included in this release.
- Selection used 915 examples and an objective of 70% mean focus metrics plus 30% retention accuracy; every family must stay within two percentage points of its initial metric. The winner was frozen before final holdout evaluation.
- Banking uses the training partition of BANKING77, financial supervision uses SEntFiN, and sarcasm uses author-labeled iSarcasm training examples. Financial headlines and tweet/rephrase groups stay in one split.
Full-model holdout results
The results below were re-measured on the full merged model, after BF16 rounding. Focus holdouts exclude the previous local input corpus listed in the data audit; retention is reused. Small sample sizes mean these modest differences do not establish statistical significance or broad benchmark improvement.
| Holdout / metric | Cases | v6.1 | v6.2 merged |
|---|---|---|---|
| Banking / accuracy | 154 | 74.68% | 74.68% |
| Entity sentiment / macro-F1 | 180 | 71.16% | 71.21% |
| Sarcasm / positive-class F1 | 80 | 46.81% | 50.00% |
| Retention / accuracy | 500 | 93.40% | 93.40% |
Financial sentiment is a SEntFiN transfer holdout, not the FinEntity benchmark. Banking and sarcasm rows above are local source holdouts, not their public test benchmark scores. Exact content/group exclusions do not establish semantic or pretraining decontamination. Benchmarks informed training focus.
Decision Index 0.3 (official, 2026-10-07): 40.01, rank 52 of 113. Run by the index maintainers on the full suite (110,201 requests, 42 benchmarks; 243 requests exceed the 16,384-token prompt limit): public 42.17, same skills 40.93, new domains 34.46. Within the index's tie band of the other entries on the same 4B base. Strongest in Tools & Automation (62.9), weakest in Knowledge & Reasoning (25.4); 66.0% accurate at 80.1% mean confidence (ECE 0.141). Details · Leaderboard
The earlier 25-benchmark comparison belongs to the v6.1 candidate and must not be attributed to this release.
Merge verification
FP32 B@A corrections were added to the already merged parent and rounded to BF16. All 426 tensors loaded without missing, unexpected or mismatched keys; 248 modules received updates. Nine shards contain 8,411,510,272 parameter bytes.
On 157 fixed verification cases, 0 selected answers changed between the adapter and merged model. The largest probability difference was 4.76 percentage points.
These checks include choice, score, noul and a long API-Bank input. The full 914-case holdout passed the predeclared release guards. See evaluation.json, merge-validation.json and release-manifest.json. Merge equivalence is approximate.
Website · Docs · Source and experiment records · Training protocol · Release validation
Apache-2.0 model/runtime; source datasets retain their own licenses. Dataset rows are not redistributed here. Quality outside the evaluated domains is not established.