---
license: other
license_name: research-diagnostic-only
license_link: LICENSE
library_name: pytorch
language:
  - en
tags:
  - news-impact
  - multimodal
  - finance
  - text-chart-fusion
  - research-diagnostic
  - quarantined
  - tbv
  - catchem
  - not-for-production
pipeline_tag: text-classification
extra_gated_prompt: >-
  This checkpoint failed its own internal audit (see AUDIT_REPORT.md, 3 CRITICAL
  bugs). It is shared as a quarantined research artifact only. By downloading
  you acknowledge it is not validated for production use.
---

# TBV NewsImpact v1 — Research Diagnostic Checkpoint

> **⚠️ NOT FOR PRODUCTION USE — QUARANTINED RESEARCH ARTIFACT**
>
> This checkpoint failed its own internal audit (2026-05-13). It is published
> as a *diagnostic* artifact so the catchem governance pipeline and downstream
> research can reproduce the audit findings against the actual weights. The
> companion catchem project loads it only through `newsimpact_guarded_adapter`
> which **refuses** to enable it outside `research_diagnostic` mode and only
> when `governance_index.release_gate_passed == False`.
>
> If you are looking for a working news-impact model, this is not it. Read
> [AUDIT_REPORT.md](AUDIT_REPORT.md) before any other section.

---

## TL;DR

| Field | Value |
|---|---|
| Architecture | Hybrid: RoBERTa-large (frozen) + small chart Transformer + cross-attention fusion |
| Parameters | **361.6 M** total (`float32`) — 355.4 M text encoder + 6.2 M trainable task layer |
| Checkpoint size | **1.38 GB** (`final_best.pt`, 456 tensors, PyTorch serialization v3) |
| Heads | `head_sig` (2-way), `head_dir` (3-way), `head_mag` (1-d), `head_reason` (1024-d, *untrained*) |
| Reported val loss | 2.4951 (pre-merge per-shard mean — see audit C26/CRITICAL-3) |
| Training corpus | BBC + Guardian news (9.6 M rows) × multi-market OHLCV |
| Training date | 2026-05-12 (4 shards, weight-averaged) |
| **Audit verdict** | **NOT TRUSTED YET** — 3 CRITICAL bugs, 5 HIGH, 3 MEDIUM |
| Governance state | Quarantined; `release_gate_passed = False` |
| Intended use | Audit reproduction, diagnostic comparisons, ablation studies |

---

## Why this is published despite failing audit

`catchem` is an open trading-news pipeline that consumes this checkpoint **only
in research-diagnostic mode**. The audit findings are not hidden — they are
the entire point of the quarantine. Publishing the weights lets independent
researchers:

1. Reproduce the cross-attention collapse proof
   ([`cross_attention_collapse_proof.md`](cross_attention_collapse_proof.md))
2. Verify that swapping text inputs produces bit-identical outputs (the canonical
   text-is-unused test, see audit CRITICAL-1)
3. Test the same architecture *with the collapse fixed* against this baseline
4. Confirm the daily/4h chart resolution mixup at label time (CRITICAL-2)

---

## ⚠️ Known critical defects (must-read)

These are summarised here so a casual reader understands the risk before
downloading. Full evidence and line numbers in `AUDIT_REPORT.md`.

### CRITICAL-1 — Cross-attention fusion is a mathematical no-op

The fusion layer is `nn.MultiheadAttention(1024, 8, batch_first=True)` but
the forward pass passes single-token tensors `(B, 1, D)` as Q/K/V. With T=1
the softmax over a one-element score vector is identically 1.0, so attention
weights cannot modulate the value. **The text encoder's output never reaches
the prediction surface.** The 336M-parameter RoBERTa-large is decorative.

You can verify this yourself by loading the checkpoint and comparing outputs
for two different news strings paired with the same chart — they will be
bit-identical. See section 7 of `AUDIT_REPORT.md` for the assertion script.

### CRITICAL-2 — `charts_4h/` contains daily bars; equity labels silently rejected

`fetch_yfinance_daily` writes daily bars to a directory named `charts_4h/`.
`compute_impact_label` computes `pre['close'].pct_change().std()` over a 24h
pre-window — which for daily bars contains 0–1 rows → `std()` = NaN → label
rejected at `pipeline.py:1051`. **The effective training corpus is filtered to
crypto symbols only**, silently. Equities, FX, indices, and commodities
contribute zero labels despite being ingested.

### CRITICAL-3 — Reported `mean_val_loss` is pre-merge per-shard, not the merged model

`merge_shards_and_test` writes the soft-averaged state dict and reports the
*mean of per-shard pre-merge validation losses*. No re-assessment of the
merged checkpoint occurs. The actual performance of `final_best.pt` is
**unknown** in the original report. Weight averaging of independently-trained
shards typically degrades performance; the merged model may be materially
worse than 2.4951.

---

## Architecture summary (from `model_analysis.json`)

```text
final_best.pt
├── text_model        391 tensors  ->  355.36 M params  (1.36 GB)  -- roberta-large (frozen at train time)
├── chart_enc          53 tensors  ->    0.99 M params  (~4 MB)    -- 4-layer Transformer over (B, 128, 5) OHLCV
├── fuse                                                           -- nn.MultiheadAttention(1024, 8)  <- collapses (see CRITICAL-1)
├── head_sig                                                       -- Linear(1024, 2)   significance
├── head_dir                                                       -- Linear(1024, 3)   direction
├── head_mag                                                       -- Linear(1024, 1)   magnitude
└── head_reason                                                    -- Linear(1024, 1024) reason embedding (NEVER receives loss signal)
```

| Param dtype | Count |
|---|---|
| `float32` | 361,605,382 |

---

## Files in this repository

| File | Size | Purpose |
|---|---|---|
| `final_best.pt` | 1.38 GB | Checkpoint loadable via `torch.load` |
| `final_report.json` | 302 B | Per-shard val losses + final.pt path from training session |
| `model_analysis.json` | 146 KB | Full tensor-level breakdown (shapes, dtypes, storage offsets) |
| `SHA256SUMS` | 80 B | Integrity hash for `final_best.pt` |
| `AUDIT_REPORT.md` | 39 KB | Full 2026-05-13 audit — 12 sections, 43 claim verifications |
| `cross_attention_collapse_proof.md` | 4.2 KB | Standalone mathematical proof of CRITICAL-1 |

```text
ccf6c8b8f80d2f1264dedae93398a7c70dbd100a95701341f8447c1b1c585141  final_best.pt
```

Verify locally:

```bash
shasum -a 256 -c SHA256SUMS
# final_best.pt: OK
```

---

## Intended use

This artifact exists for **research diagnostics only**:

- ✅ Reproducing the cross-attention collapse audit
- ✅ Benchmark baseline for "chart-only" performance
- ✅ Ablation studies (replace `forward` to enable fusion, then compare)
- ✅ Recurrence-test target for the catchem governance pipeline
- ✅ Teaching example of how a 360M-parameter model can be mathematically
  equivalent to a 6M-parameter chart classifier wearing a costume

It is **not** intended for:

- ❌ Live trading signals
- ❌ News-impact production scoring
- ❌ Down-stream fine-tuning without fixing CRITICAL-1 first
- ❌ Comparison against properly trained baselines without first reading the
  audit (the reported metric is materially misleading)

---

## How catchem uses it

The `catchem` project at <https://github.com/nazmiefearmutcu/catchem> contains
the `NewsImpactGuardedAdapter` that is the *only* sanctioned integration.
The adapter loads governance metadata read-only and refuses to instantiate
unless **three independent guards** hold:

1. catchem is configured for `research_diagnostic` mode (production_safe is
   refused outright),
2. the `guards.newsimpact_diagnostic_enabled` flag is `True`,
3. `governance_index.json` shows `release_gate_passed = False` (the expected
   quarantined state).

The adapter **does not** load `final_best.pt`; it only reads
`governance_index.json` and emits a clearly-labeled
`newsimpact_diagnostic_v0` payload alongside each catchem record.

---

## Suggested next steps before any production use

In priority order (from `AUDIT_REPORT.md` § 10):

1. **Fix the cross-attention collapse.** Either concat-then-project, or have
   `ChartEncoder.forward` return a `(B, T, D)` sequence (drop the `.mean(dim=1)`
   pool) and use cross-attention with chart as key/value sequence vs text-pool
   as a single query token. Add the `test_text_input_affects_output` pytest.
2. **Fix `compute_impact_label`** to pick one bar resolution per symbol and
   fail loudly on mismatch. Fix the vol-window ↔ return-window dimensional
   mismatch (use `pre_vol * √12` for 48h horizon over 4h bars).
3. **Add a held-out global test set** and have `merge_shards_and_test`
   assess the merged checkpoint on it. Report both pre-merge mean and
   merged-on-holdout.
4. **Set seeds at the top of every runner** (`torch.manual_seed`,
   `random.seed`, `np.random.seed`). Run each config 3× with seeds
   {42, 1337, 7} and report mean ± std.
5. **Either remove `head_reason` or train it** against a meaningful target
   (sentence-encoder embedding of a per-pair rationale).

---

## Citation

```bibtex
@misc{tbv_newsimpact_v1_2026,
  author       = {{Nazmi Efe Armutçu}},
  title        = {{TBV NewsImpact v1 (research-diagnostic, quarantined)}},
  year         = {2026},
  publisher    = {{Hugging Face}},
  howpublished = {\url{https://huggingface.co/AylinMaylinn/news-impact-v1}},
  note         = {Failed internal audit 2026-05-13; published as quarantined
                  research artifact. See AUDIT\_REPORT.md.}
}
```

---

## License

Custom **research-diagnostic-only** license. Summary: free to download and
study; not licensed for any production deployment, financial decision
making, or redistribution as a "validated" news-impact model. If you build
a corrected derivative, please cite the audit findings.

---

*Original training: 2026-05-12 (Google Drive). Audit: 2026-05-13. Published
to Hugging Face under catchem's governance pipeline: 2026-05-27.*