diff --git a/README.md b/README.md
index 62d78c4435b2a05beac8d5400752cd3e5c655b0e..2954c57cb3712d16b125a77c8b116c6c53c35163 100644
--- a/README.md
+++ b/README.md
@@ -1,57 +1,10 @@
----
-language:
-- en
-library_name: transformers
-pipeline_tag: text-classification
-base_model: distilbert-base-uncased
-metrics:
-- accuracy
-- f1
-tags:
-- intent-classification
-- multitask
-- iab
-- conversational-ai
-- adtech
-- calibrated-confidence
-license: apache-2.0
----
-
-# admesh/agentic-intent-classifier
-
-Production-ready intent + IAB classifier bundle for conversational traffic.
-
-Combines multitask intent modeling, supervised IAB content classification, and per-head confidence calibration to support safe monetization decisions in real time.
-
-## Links
-
-- Hugging Face: https://huggingface.co/admesh/agentic-intent-classifier
-- GitHub: https://github.com/GouniManikumar12/agentic-intent-classifier
+# Agentic Intent Classifier
 
-## What It Predicts
+`agentic-intent-classifier` is a multi-head query classification stack for conversational traffic.
 
-| Field | Description |
-|---|---|
-| `intent.type` | `commercial`, `informational`, `navigational`, `transactional`, … |
-| `intent.subtype` | `product_discovery`, `comparison`, `how_to`, … |
-| `intent.decision_phase` | `awareness`, `consideration`, `decision`, … |
-| `iab_content` | IAB Content Taxonomy 3.0 tier1 / tier2 / tier3 labels |
-| `component_confidence` | Per-head calibrated confidence with threshold flags |
-| `system_decision` | Monetization eligibility, opportunity type, policy |
-
----
-
-## Deployment Options
-
-### 0. Colab / Kaggle Quickstart (copy/paste)
-
-```python
-!pip -q install -U pip
-!pip -q install -U "torch==2.10.0" "torchvision==0.25.0" "torchaudio==2.10.0"
-!pip -q install -U "transformers>=4.36.0" "huggingface_hub>=0.20.0" "safetensors>=0.4.0"
-```
+## Quickstart (recommended): run from Hugging Face Hub
 
-Restart the runtime after installs (**Runtime → Restart runtime**) so the new Torch version is actually used.
+This is the easiest way for developers to test the full production stack (multitask intent + IAB + calibration) without training locally.
 
 ```python
 from transformers import pipeline
@@ -59,212 +12,539 @@ from transformers import pipeline
 clf = pipeline(
     "admesh-intent",
     model="admesh/agentic-intent-classifier",
-    trust_remote_code=True,  # required (custom pipeline + multi-model bundle)
+    trust_remote_code=True,
 )
 
 out = clf("Which laptop should I buy for college?")
-print(out["meta"])
 print(out["model_output"]["classification"]["intent"])
+print(out["model_output"]["classification"]["iab_content"])
+print(out["meta"])
 ```
 
----
+If you’re running in Colab/Kaggle and see Torch version conflicts, follow `COLAB_SETUP.md`.
+
+## Latency / inference timing (developer quick check)
 
-## Latency / inference timing (quick check)
+The first call includes model/code loading; measure latency after a warm-up call.
 
-The first call includes model/code loading. Warm up once, then measure:
+Single query:
 
 ```python
 import time
+from transformers import pipeline
+
+clf = pipeline("admesh-intent", model="admesh/agentic-intent-classifier", trust_remote_code=True)
 q = "Which laptop should I buy for college?"
 
 _ = clf("warm up")
 t0 = time.perf_counter()
 out = clf(q)
-print(f"latency_ms={(time.perf_counter() - t0) * 1000:.1f}")
+dt_ms = (time.perf_counter() - t0) * 1000
+
+print(f"latency_ms={dt_ms:.1f}")
+print(out["model_output"]["classification"]["intent"])
 ```
 
-### 1. `transformers.pipeline()` — anywhere (Python)
+Warm p50 / p95 over 20 runs:
 
 ```python
-from transformers import pipeline
+import time, statistics
 
-clf = pipeline(
-    "admesh-intent",
-    model="admesh/agentic-intent-classifier",
-    trust_remote_code=True,
+times = []
+for _ in range(20):
+    t0 = time.perf_counter()
+    _ = clf(q)
+    times.append((time.perf_counter() - t0) * 1000)
+
+times_sorted = sorted(times)
+print(f"p50={statistics.median(times):.1f}ms p95={times_sorted[int(0.95*len(times))-1]:.1f}ms mean={statistics.mean(times):.1f}ms")
+```
+
+It currently produces:
+
+- `intent.type`
+- `intent.subtype`
+- `intent.decision_phase`
+- `iab_content`
+- calibrated confidence per head
+- combined fallback / policy / opportunity decisions
+
+The repo is beyond the original v0.1 baseline. It now includes:
+
+- shared config and label ownership
+- reusable model runtime
+- calibrated confidence and threshold gating
+- combined inference with fallback/policy logic
+- request/response validation in the demo API
+- repeatable evaluation and regression suites
+- full-TSV IAB taxonomy retrieval support through tier4
+- a local embedding index for taxonomy-node retrieval over IAB content paths
+- a separate synthetic full-intent-taxonomy augmentation dataset for non-IAB heads
+- a dedicated intent-type difficulty dataset and held-out benchmark with `easy`, `medium`, and `hard` cases
+- a dedicated decision-phase difficulty dataset and held-out benchmark with `easy`, `medium`, and `hard` cases
+
+Generated model weights are intentionally not committed.
+
+## Current Taxonomy
+
+### `intent.type`
+
+- `informational`
+- `exploratory`
+- `commercial`
+- `transactional`
+- `support`
+- `personal_reflection`
+- `creative_generation`
+- `chit_chat`
+- `ambiguous`
+- `prohibited`
+
+### `intent.decision_phase`
+
+- `awareness`
+- `research`
+- `consideration`
+- `decision`
+- `action`
+- `post_purchase`
+- `support`
+
+### `intent.subtype`
+
+- `education`
+- `product_discovery`
+- `comparison`
+- `evaluation`
+- `deal_seeking`
+- `provider_selection`
+- `signup`
+- `purchase`
+- `booking`
+- `download`
+- `contact_sales`
+- `task_execution`
+- `onboarding_setup`
+- `troubleshooting`
+- `account_help`
+- `billing_help`
+- `follow_up`
+- `emotional_reflection`
+
+### `iab_content`
+
+- candidates are derived from every row in [data/iab-content/Content Taxonomy 3.0.tsv](data/iab-content/Content%20Taxonomy%203.0.tsv)
+- retrieval output supports `tier1`, `tier2`, `tier3`, and optional `tier4`
+
+## What The System Does
+
+- runs three classifier heads:
+  - `intent_type`
+  - `intent_subtype`
+  - `decision_phase`
+- resolves `iab_content` through a local embedding index over taxonomy nodes plus generic label/path reranking
+- applies calibration artifacts when present
+- computes `commercial_score`
+- applies fallback when confidence is too weak or policy-safe blocking is required
+- emits a schema-validated combined envelope
+
+## What The System Does Not Do
+
+- it is not a multi-turn memory system
+- it is not a production-optimized low-latency serving path
+- it is not yet trained on large real-traffic human-labeled intent data
+- combined decision logic is still heuristic, even though it is materially stronger than the original baseline
+
+## Project Layout
+
+- [config.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/config.py): labels, thresholds, artifact paths, model paths
+- [model_runtime.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/model_runtime.py): shared calibrated inference runtime
+- [combined_inference.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/combined_inference.py): composed system response
+- [inference_intent_type.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/inference_intent_type.py): direct `intent_type` inference entrypoint
+- [inference_iab_classifier.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/inference_iab_classifier.py): direct supervised `iab_content` inference entrypoint
+- [schemas.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/schemas.py): request/response validation
+- [demo_api.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/demo_api.py): local validated API
+- [iab_taxonomy.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/iab_taxonomy.py): full taxonomy parser/index
+- [iab_classifier.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/iab_classifier.py): supervised IAB runtime with taxonomy-aware parent fallback
+- [iab_retrieval.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/iab_retrieval.py): optional shadow retrieval baseline
+- [training/build_full_intent_taxonomy_dataset.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/build_full_intent_taxonomy_dataset.py): separate synthetic intent augmentation dataset
+- [training/build_intent_type_difficulty_dataset.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/build_intent_type_difficulty_dataset.py): extra `intent_type` augmentation plus held-out difficulty benchmark
+- [training/build_decision_phase_difficulty_dataset.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/build_decision_phase_difficulty_dataset.py): extra `decision_phase` augmentation plus held-out difficulty benchmark
+- [training/build_subtype_difficulty_dataset.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/build_subtype_difficulty_dataset.py): extra `intent_subtype` augmentation plus held-out difficulty benchmark
+- [training/build_subtype_dataset.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/build_subtype_dataset.py): subtype dataset generation from existing corpora
+- [training/train_iab.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/train_iab.py): train the supervised IAB classifier head
+- [training/build_iab_taxonomy_embeddings.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/build_iab_taxonomy_embeddings.py): build local IAB node embedding artifacts
+- [training/run_full_training_pipeline.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/training/run_full_training_pipeline.py): full multi-head training/calibration/eval pipeline
+- [evaluation/run_evaluation.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/evaluation/run_evaluation.py): repeatable benchmark runner
+- [evaluation/run_regression_suite.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/evaluation/run_regression_suite.py): known-failure regression runner
+- [evaluation/run_iab_mapping_suite.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/evaluation/run_iab_mapping_suite.py): IAB behavior-lock regression runner
+- [evaluation/run_iab_quality_suite.py](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/evaluation/run_iab_quality_suite.py): curated IAB quality-target runner
+- [known_limitations.md](/Users/manikumargouni/Desktop/AdMesh/protocol/agentic-intent-classifier/known_limitations.md): current gaps and caveats
+
+## Quickstart: Run From Hugging Face
+
+Download the trained bundle and run inference in three lines — no local training required.
+
+```python
+import sys
+from huggingface_hub import snapshot_download
+
+# Download the full bundle (models + calibration + code)
+local_dir = snapshot_download(
+    repo_id="admesh/agentic-intent-classifier",
+    repo_type="model",
 )
+sys.path.insert(0, local_dir)
+
+# Import and instantiate
+from pipeline import AdmeshIntentPipeline
+clf = AdmeshIntentPipeline()
 
+# Classify
+import json
 result = clf("Which laptop should I buy for college?")
+print(json.dumps(result, indent=2))
+```
+
+Or use the one-liner factory method:
+
+```python
+from pipeline import AdmeshIntentPipeline   # after sys.path.insert above
+
+clf = AdmeshIntentPipeline.from_pretrained("admesh/agentic-intent-classifier")
+result = clf("I need a CRM for a 5-person startup")
 ```
 
-Batch and custom thresholds:
+Batch mode and custom thresholds are also supported:
 
 ```python
-# batch
+# Batch
 results = clf([
     "Best running shoes under $100",
-    "How does TCP work?",
+    "How does gradient descent work?",
     "Buy noise-cancelling headphones",
 ])
 
-# custom confidence thresholds
+# Custom confidence thresholds
 result = clf(
-    "Buy headphones",
+    "Buy noise-cancelling headphones",
     threshold_overrides={"intent_type": 0.6, "intent_subtype": 0.35},
 )
 ```
 
+Verify artifacts and run a smoke test from the CLI:
+
+```bash
+cd "<local_dir>"
+python3 training/pipeline_verify.py
+python3 combined_inference.py "Which CRM should I buy for a 3-person startup?"
+```
+
+Pin a specific revision for reproducibility:
+
+```python
+local_dir = snapshot_download(
+    repo_id="admesh/agentic-intent-classifier",
+    repo_type="model",
+    revision="0584798f8efee6beccd778b0afa06782ab5add60",
+)
+```
+
 ---
 
-### 2. HF Inference Endpoints (managed, deploy to AWS / Azure / GCP)
+## Setup (for local training)
 
-1. Go to https://ui.endpoints.huggingface.co
-2. **New Endpoint** → select `admesh/agentic-intent-classifier`
-3. Framework: **PyTorch** — Task: **Text Classification**
-4. Enable **"Load with trust_remote_code"**
-5. Deploy
+```bash
+python3 -m venv .venv
+source .venv/bin/activate
+pip install -r agentic-intent-classifier/requirements.txt
+```
+
+## Inference (local training path)
 
-The endpoint serves the same `pipeline()` interface above via REST:
+Run one query locally:
 
 ```bash
-curl https://<your-endpoint>.endpoints.huggingface.cloud \
-  -H "Authorization: Bearer $HF_TOKEN" \
-  -H "Content-Type: application/json" \
-  -d '{"inputs": "Which laptop should I buy for college?"}'
+cd agentic-intent-classifier
+python3 training/train_iab.py
+python3 training/calibrate_confidence.py --head iab_content
+python3 combined_inference.py "Which CRM should I buy for a 3-person startup?"
 ```
 
----
+Run only the `intent_type` head:
 
-### 3. HF Spaces (Gradio / Streamlit demo)
+```bash
+cd agentic-intent-classifier
+python3 inference_intent_type.py "best shoes under 100"
+```
 
-```python
-# app.py for a Gradio Space
-import gradio as gr
-from transformers import pipeline
+Run the demo API:
 
-clf = pipeline(
-    "admesh-intent",
-    model="admesh/agentic-intent-classifier",
-    trust_remote_code=True,
-)
+```bash
+cd agentic-intent-classifier
+python3 demo_api.py
+```
 
-def classify(text):
-    return clf(text)
+Example request:
 
-gr.Interface(fn=classify, inputs="text", outputs="json").launch()
+```bash
+curl -sS -X POST http://127.0.0.1:8008/classify \
+  -H 'Content-Type: application/json' \
+  -d '{"text":"I cannot log into my account"}'
 ```
 
----
+Infra endpoints:
 
-### 4. Local / notebook via `snapshot_download`
+```bash
+curl -sS http://127.0.0.1:8008/health
+curl -sS http://127.0.0.1:8008/version
+```
 
-```python
-import sys
-from huggingface_hub import snapshot_download
+Train only the IAB classifier head:
 
-local_dir = snapshot_download(
-    repo_id="admesh/agentic-intent-classifier",
-    repo_type="model",
-)
-sys.path.insert(0, local_dir)
+```bash
+cd agentic-intent-classifier
+python3 training/train_iab.py
+python3 training/calibrate_confidence.py --head iab_content
+```
 
-from pipeline import AdmeshIntentPipeline
-clf = AdmeshIntentPipeline()
-result = clf("I need a CRM for a 5-person startup")
+The online `iab_content` path now uses the compact supervised classifier. Retrieval is still available as an optional shadow baseline.
+
+Build the optional retrieval shadow index:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_iab_taxonomy_embeddings.py
 ```
 
-Or the one-liner factory:
+By default the shadow retrieval path uses `Alibaba-NLP/gte-Qwen2-1.5B-instruct`. The retrieval runtime applies the model's query-side instruction format and last-token pooling, matching the Hugging Face usage guidance. If you want to point retrieval at a different embedding model, set `IAB_RETRIEVAL_MODEL_NAME_OVERRIDE` before building the index.
 
-```python
-from pipeline import AdmeshIntentPipeline
-clf = AdmeshIntentPipeline.from_pretrained("admesh/agentic-intent-classifier")
+Open-source users can swap in their own embedding model, but the contract is:
+
+- query embeddings and taxonomy-node embeddings must be produced by the same model and model revision
+- after changing models, you must rebuild `artifacts/iab/taxonomy_embeddings.pt`
+- the repository only tests and supports the default model path out of the box
+- not every Hugging Face embedding model is drop-in compatible with this runtime; some require custom pooling, query instructions, or `trust_remote_code`
+
+Example override:
+
+```bash
+cd agentic-intent-classifier
+export IAB_RETRIEVAL_MODEL_NAME_OVERRIDE=mixedbread-ai/mxbai-embed-large-v1
+python3 training/build_iab_taxonomy_embeddings.py
 ```
 
----
+This writes:
+
+- `artifacts/iab/taxonomy_nodes.json`
+- `artifacts/iab/taxonomy_embeddings.pt`
+
+## Training
+
+### Full local pipeline
+
+```bash
+cd agentic-intent-classifier
+python3 training/run_full_training_pipeline.py
+```
+
+This pipeline now does:
+
+1. build separate full-intent-taxonomy augmentation data
+2. build separate `intent_type` difficulty augmentation + benchmark
+3. train `intent_type`
+4. build subtype corpus
+5. build separate `intent_subtype` difficulty augmentation + benchmark
+6. train `intent_subtype`
+7. build separate `decision_phase` difficulty augmentation + benchmark
+8. train `decision_phase`
+9. train `iab_content`
+10. calibrate all classifier heads, including `iab_content`
+11. run regression/evaluation unless `--skip-full-eval` is used
+
+### Build datasets individually
+
+Separate full-intent augmentation:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_full_intent_taxonomy_dataset.py
+```
+
+Intent-type difficulty augmentation and benchmark:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_intent_type_difficulty_dataset.py
+```
+
+Decision-phase difficulty augmentation and benchmark:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_decision_phase_difficulty_dataset.py
+```
+
+Subtype difficulty augmentation and benchmark:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_subtype_difficulty_dataset.py
+```
+
+Subtype dataset:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_subtype_dataset.py
+```
+
+IAB embedding index:
+
+```bash
+cd agentic-intent-classifier
+python3 training/build_iab_taxonomy_embeddings.py
+```
+
+### Train heads individually
+
+```bash
+cd agentic-intent-classifier
+python3 training/train.py
+python3 training/train_subtype.py
+python3 training/train_decision_phase.py
+```
+
+### Calibration
+
+```bash
+cd agentic-intent-classifier
+python3 training/calibrate_confidence.py --head intent_type
+python3 training/calibrate_confidence.py --head intent_subtype
+python3 training/calibrate_confidence.py --head decision_phase
+```
+
+## Evaluation
+
+Full evaluation:
+
+```bash
+cd agentic-intent-classifier
+python3 evaluation/run_evaluation.py
+```
 
-## Troubleshooting (avoid environment errors)
+Known-failure regression:
 
-### `No module named 'combined_inference'` (or similar)
+```bash
+cd agentic-intent-classifier
+python3 evaluation/run_regression_suite.py
+```
 
-This means the Hub repo root is missing required Python files. Ensure these exist at the **root of the model repo** (same level as `pipeline.py`):
+IAB behavior-lock regression:
+
+```bash
+cd agentic-intent-classifier
+python3 evaluation/run_iab_mapping_suite.py
+```
 
-- `pipeline.py`, `config.json`, `config.py`
-- `combined_inference.py`, `schemas.py`
-- `model_runtime.py`, `multitask_runtime.py`, `multitask_model.py`
-- `inference_intent_type.py`, `inference_subtype.py`, `inference_decision_phase.py`, `inference_iab_classifier.py`
-- `iab_classifier.py`, `iab_taxonomy.py`
+IAB quality-target evaluation:
 
-### `does not appear to have a file named model.safetensors`
+```bash
+cd agentic-intent-classifier
+python3 evaluation/run_iab_quality_suite.py
+```
+
+Threshold sweeps:
+
+```bash
+cd agentic-intent-classifier
+python3 evaluation/sweep_intent_threshold.py
+```
 
-Transformers requires a standard checkpoint at the repo root for `pipeline()` to initialize. This repo includes a **small dummy** `model.safetensors` + tokenizer files at the root for compatibility; the *real* production weights live in:
+Artifacts are written to:
 
-- `multitask_intent_model_output/`
-- `iab_classifier_model_output/`
 - `artifacts/calibration/`
+- `artifacts/evaluation/latest/`
 
----
+## Google Colab
 
-## Example Output
-
-```json
-{
-  "model_output": {
-    "classification": {
-      "iab_content": {
-        "taxonomy": "IAB Content Taxonomy",
-        "taxonomy_version": "3.0",
-        "tier1": {"id": "552", "label": "Style & Fashion"},
-        "tier2": {"id": "579", "label": "Men's Fashion"},
-        "mapping_mode": "exact",
-        "mapping_confidence": 0.73
-      },
-      "intent": {
-        "type": "commercial",
-        "subtype": "product_discovery",
-        "decision_phase": "consideration",
-        "confidence": 0.9549,
-        "commercial_score": 0.656
-      }
-    }
-  },
-  "system_decision": {
-    "policy": {
-      "monetization_eligibility": "allowed_with_caution",
-      "eligibility_reason": "commercial_discovery_signal_present"
-    },
-    "opportunity": {"type": "soft_recommendation", "strength": "medium"}
-  },
-  "meta": {
-    "system_version": "0.6.0-phase4",
-    "calibration_enabled": true,
-    "iab_mapping_is_placeholder": false
-  }
-}
-```
-
-## Reproducible Revision
+Use Colab for the full retraining pass if local memory is limited.
 
-```python
-from huggingface_hub import snapshot_download
-local_dir = snapshot_download(
-    repo_id="admesh/agentic-intent-classifier",
-    repo_type="model",
-    revision="0584798f8efee6beccd778b0afa06782ab5add60",
-)
+Clone once:
+
+```bash
+%cd /content
+!git clone https://github.com/GouniManikumar12/agentic-intent-classifier.git
+%cd /content/agentic-intent-classifier
+```
+
+If the repo is already cloned and you want the latest code, pull manually:
+
+```bash
+!git pull origin main
 ```
 
-## Included Artifacts
+Full pipeline:
+
+```bash
+!python training/run_full_training_pipeline.py
+```
+
+If full evaluation is too heavy for the current Colab runtime:
+
+```bash
+!python training/run_full_training_pipeline.py \
+  --iab-embedding-batch-size 32 \
+  --skip-full-eval
+```
+
+Then run eval separately after training:
+
+```bash
+!python evaluation/run_regression_suite.py
+!python evaluation/run_iab_mapping_suite.py
+!python evaluation/run_iab_quality_suite.py
+!python evaluation/run_evaluation.py
+```
+
+## Current Saved Metrics
+
+Generate fresh metrics with:
+
+```bash
+cd agentic-intent-classifier
+python3 evaluation/run_evaluation.py
+```
+
+Do not treat any checked-in summary as canonical unless it was regenerated after the current code and artifacts were built. The IAB path is now retrieval-based, so older saved reports from the deleted hierarchy stack are not meaningful.
+
+## Latency Note
+
+`combined_inference.py` is a debugging/offline path, not a production latency path.
+
+Current production truth:
+
+- per-request CLI execution is not a sub-50ms architecture
+- production serving should use a long-lived API process with preloaded models
+- if sub-50ms becomes a hard requirement, the serving path will need:
+  - persistent loaded models
+  - runtime optimization
+  - likely fewer model passes or a shared multi-head model
+
+## Current Status
+
+Current repo status:
 
-| Path | Contents |
-|---|---|
-| `multitask_intent_model_output/` | DistilBERT multitask weights + tokenizer |
-| `iab_classifier_model_output/` | IAB content classifier weights + tokenizer |
-| `artifacts/calibration/` | Per-head temperature + threshold JSONs |
-| `pipeline.py` | `AdmeshIntentPipeline` (transformers.Pipeline subclass) |
-| `combined_inference.py` | Core inference logic |
+- full 10-class `intent.type` taxonomy is wired
+- subtype and phase heads are present
+- difficulty benchmarks are wired for `intent_type`, `intent_subtype`, and `decision_phase`
+- full-TSV IAB taxonomy retrieval is wired through tier4
+- separate full-intent augmentation dataset is in place
+- evaluation/runtime memory handling is improved for large IAB splits
 
-## Notes
+The main remaining gap is not basic infrastructure anymore. It is improving real-world robustness, especially for:
 
-- `trust_remote_code=True` is required because this model uses a custom multi-head architecture that does not map to a single standard `AutoModel` checkpoint.
-- `meta.iab_mapping_is_placeholder: true` means IAB artifacts were missing or skipped; train and calibrate IAB for full production accuracy.
-- For long-running servers, instantiate once and reuse — models are cached in memory after the first call.
+- `decision_phase`
+- `intent_subtype`
+- confidence quality on borderline commercial queries
+- real-traffic supervision beyond synthetic data
diff --git a/artifacts/calibration/decision_phase.json b/artifacts/calibration/decision_phase.json
index 06b02a1a1b4b84af5804899ce505f8536fd5c41b..059c90fb3995abd233e041d70ff9430aa692b2e5 100644
--- a/artifacts/calibration/decision_phase.json
+++ b/artifacts/calibration/decision_phase.json
@@ -1,31 +1,31 @@
 {
   "calibrated": true,
   "confidence_threshold": 0.22,
-  "generated_at": "2026-03-25T20:15:26.091588+00:00",
+  "generated_at": "2026-03-25T21:20:02.753657+00:00",
   "head": "decision_phase",
   "metrics": {
-    "calibrated_accuracy": 0.8621,
-    "calibrated_expected_calibration_error": 0.1012,
-    "calibrated_negative_log_likelihood": 0.4275,
-    "mean_calibrated_confidence": 0.8431,
-    "mean_raw_confidence": 0.8274,
-    "raw_accuracy": 0.8621,
-    "raw_expected_calibration_error": 0.0985,
-    "raw_negative_log_likelihood": 0.4338
+    "calibrated_accuracy": 0.8276,
+    "calibrated_expected_calibration_error": 0.0811,
+    "calibrated_negative_log_likelihood": 0.5525,
+    "mean_calibrated_confidence": 0.8739,
+    "mean_raw_confidence": 0.8828,
+    "raw_accuracy": 0.8276,
+    "raw_expected_calibration_error": 0.0757,
+    "raw_negative_log_likelihood": 0.5551
   },
   "minimum_threshold_floor": 0.22,
-  "optimized_temperature_candidate": 0.940009,
+  "optimized_temperature_candidate": 1.032831,
   "selected_threshold_before_floor": {
-    "accepted_accuracy": 0.8621,
+    "accepted_accuracy": 0.8276,
     "coverage": 1.0,
     "threshold": 0.0
   },
   "selection_split": "val",
   "selection_target_precision": 0.75,
-  "temperature": 0.940009,
+  "temperature": 1.032831,
   "temperature_scaling_applied": true,
   "threshold_summary": {
-    "accepted_accuracy": 0.8621,
+    "accepted_accuracy": 0.8276,
     "coverage": 1.0,
     "threshold": 0.22
   }
diff --git a/artifacts/calibration/iab_content.json b/artifacts/calibration/iab_content.json
index ce93446814fa41d584fe9d70db4d48c58b2e4092..cc16ff2e932fe16bcac0d254be8192dbbbb41540 100644
--- a/artifacts/calibration/iab_content.json
+++ b/artifacts/calibration/iab_content.json
@@ -1,32 +1,32 @@
 {
   "calibrated": true,
   "confidence_threshold": 0.12,
-  "generated_at": "2026-03-25T20:39:18.586053+00:00",
+  "generated_at": "2026-03-25T21:21:46.770447+00:00",
   "head": "iab_content",
   "metrics": {
-    "calibrated_accuracy": 0.9485,
-    "calibrated_expected_calibration_error": 0.2692,
-    "calibrated_negative_log_likelihood": 0.5281,
-    "mean_calibrated_confidence": 0.6793,
-    "mean_raw_confidence": 0.1987,
-    "raw_accuracy": 0.9485,
-    "raw_expected_calibration_error": 0.7498,
-    "raw_negative_log_likelihood": 1.7931
+    "calibrated_accuracy": 0.9321,
+    "calibrated_expected_calibration_error": 0.2607,
+    "calibrated_negative_log_likelihood": 0.5642,
+    "mean_calibrated_confidence": 0.6714,
+    "mean_raw_confidence": 0.1481,
+    "raw_accuracy": 0.9321,
+    "raw_expected_calibration_error": 0.7839,
+    "raw_negative_log_likelihood": 2.103
   },
   "minimum_threshold_floor": 0.12,
-  "optimized_temperature_candidate": 0.573651,
+  "optimized_temperature_candidate": 0.502066,
   "selected_threshold_before_floor": {
-    "accepted_accuracy": 0.9485,
+    "accepted_accuracy": 0.9321,
     "coverage": 1.0,
     "threshold": 0.0
   },
   "selection_split": "val",
   "selection_target_precision": 0.7,
-  "temperature": 0.573651,
+  "temperature": 0.502066,
   "temperature_scaling_applied": true,
   "threshold_summary": {
-    "accepted_accuracy": 0.9553,
-    "coverage": 0.9875,
+    "accepted_accuracy": 0.9472,
+    "coverage": 0.975,
     "threshold": 0.12
   }
 }
diff --git a/artifacts/calibration/intent_subtype.json b/artifacts/calibration/intent_subtype.json
index 27f1c5055f81fd986302160619cfeeb6f4b943c3..3e4237120819a25ff36be0234d5e40bc3c183d32 100644
--- a/artifacts/calibration/intent_subtype.json
+++ b/artifacts/calibration/intent_subtype.json
@@ -1,20 +1,20 @@
 {
   "calibrated": true,
   "confidence_threshold": 0.25,
-  "generated_at": "2026-03-25T20:15:16.332284+00:00",
+  "generated_at": "2026-03-25T21:19:52.208134+00:00",
   "head": "intent_subtype",
   "metrics": {
     "calibrated_accuracy": 0.8625,
-    "calibrated_expected_calibration_error": 0.0827,
-    "calibrated_negative_log_likelihood": 0.3943,
-    "mean_calibrated_confidence": 0.8131,
-    "mean_raw_confidence": 0.7338,
+    "calibrated_expected_calibration_error": 0.0584,
+    "calibrated_negative_log_likelihood": 0.4288,
+    "mean_calibrated_confidence": 0.8389,
+    "mean_raw_confidence": 0.7787,
     "raw_accuracy": 0.8625,
-    "raw_expected_calibration_error": 0.152,
-    "raw_negative_log_likelihood": 0.4841
+    "raw_expected_calibration_error": 0.0904,
+    "raw_negative_log_likelihood": 0.4795
   },
   "minimum_threshold_floor": 0.25,
-  "optimized_temperature_candidate": 0.789295,
+  "optimized_temperature_candidate": 0.861152,
   "selected_threshold_before_floor": {
     "accepted_accuracy": 0.8625,
     "coverage": 1.0,
@@ -22,7 +22,7 @@
   },
   "selection_split": "val",
   "selection_target_precision": 0.75,
-  "temperature": 0.789295,
+  "temperature": 0.861152,
   "temperature_scaling_applied": true,
   "threshold_summary": {
     "accepted_accuracy": 0.8625,
diff --git a/artifacts/calibration/intent_type.json b/artifacts/calibration/intent_type.json
index 9601552ffc8cf86244a0cad32057b3f7e76897ac..faa6f28b51b2f87500961305845fedff611b80ca 100644
--- a/artifacts/calibration/intent_type.json
+++ b/artifacts/calibration/intent_type.json
@@ -1,32 +1,32 @@
 {
   "calibrated": true,
   "confidence_threshold": 0.4,
-  "generated_at": "2026-03-25T20:15:05.272668+00:00",
+  "generated_at": "2026-03-25T21:19:41.302013+00:00",
   "head": "intent_type",
   "metrics": {
-    "calibrated_accuracy": 0.8936,
-    "calibrated_expected_calibration_error": 0.0915,
-    "calibrated_negative_log_likelihood": 0.2696,
-    "mean_calibrated_confidence": 0.9148,
-    "mean_raw_confidence": 0.8914,
-    "raw_accuracy": 0.8936,
-    "raw_expected_calibration_error": 0.0842,
-    "raw_negative_log_likelihood": 0.2831
+    "calibrated_accuracy": 0.8723,
+    "calibrated_expected_calibration_error": 0.0798,
+    "calibrated_negative_log_likelihood": 0.2692,
+    "mean_calibrated_confidence": 0.8962,
+    "mean_raw_confidence": 0.8671,
+    "raw_accuracy": 0.8723,
+    "raw_expected_calibration_error": 0.1073,
+    "raw_negative_log_likelihood": 0.2907
   },
   "minimum_threshold_floor": 0.4,
-  "optimized_temperature_candidate": 0.918544,
+  "optimized_temperature_candidate": 0.889496,
   "selected_threshold_before_floor": {
-    "accepted_accuracy": 0.8936,
+    "accepted_accuracy": 0.8723,
     "coverage": 1.0,
     "threshold": 0.0
   },
   "selection_split": "val",
   "selection_target_precision": 0.8,
-  "temperature": 0.918544,
+  "temperature": 0.889496,
   "temperature_scaling_applied": true,
   "threshold_summary": {
-    "accepted_accuracy": 0.8936,
-    "coverage": 1.0,
+    "accepted_accuracy": 0.8913,
+    "coverage": 0.9787,
     "threshold": 0.4
   }
 }
diff --git a/artifacts/evaluation/latest/combined_demo_benchmark.json b/artifacts/evaluation/latest/combined_demo_benchmark.json
index 1b8515d1b9d8fb2e9a0e212a6554af6b18f71f1d..27570895691ea783aa71c9f1a001299a8240f150 100644
--- a/artifacts/evaluation/latest/combined_demo_benchmark.json
+++ b/artifacts/evaluation/latest/combined_demo_benchmark.json
@@ -11,21 +11,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.5429,
+            "mapping_confidence": 0.3078,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
               "id": "596",
               "label": "Technology & Computing"
-            },
-            "tier2": {
-              "id": "599",
-              "label": "Computing"
-            },
-            "tier3": {
-              "id": "602",
-              "label": "Software and Applications"
             }
           },
           "intent": {
@@ -33,31 +25,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.962,
+                "confidence": 0.9548,
                 "confidence_threshold": 0.22,
                 "label": "awareness",
                 "meets_threshold": true,
-                "raw_confidence": 0.9633
+                "raw_confidence": 0.9611
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9805,
+                "confidence": 0.9731,
                 "confidence_threshold": 0.25,
                 "label": "education",
                 "meets_threshold": true,
-                "raw_confidence": 0.9549
+                "raw_confidence": 0.9378
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9817,
+                "confidence": 0.9816,
                 "confidence_threshold": 0.4,
                 "label": "informational",
                 "meets_threshold": true,
-                "raw_confidence": 0.9658
+                "raw_confidence": 0.9644
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.962,
+            "confidence": 0.9548,
             "decision_phase": "awareness",
             "subtype": "education",
             "summary": "Classified as informational intent with subtype education in the awareness phase.",
@@ -102,7 +94,7 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.4784,
+            "mapping_confidence": 0.2281,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
@@ -116,31 +108,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9277,
+                "confidence": 0.9159,
                 "confidence_threshold": 0.22,
                 "label": "awareness",
                 "meets_threshold": true,
-                "raw_confidence": 0.9297
+                "raw_confidence": 0.9256
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9749,
+                "confidence": 0.9671,
                 "confidence_threshold": 0.25,
                 "label": "education",
                 "meets_threshold": true,
-                "raw_confidence": 0.9445
+                "raw_confidence": 0.9273
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9797,
+                "confidence": 0.9771,
                 "confidence_threshold": 0.4,
                 "label": "informational",
                 "meets_threshold": true,
-                "raw_confidence": 0.9626
+                "raw_confidence": 0.957
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.9277,
+            "confidence": 0.9159,
             "decision_phase": "awareness",
             "subtype": "education",
             "summary": "Classified as informational intent with subtype education in the awareness phase.",
@@ -185,17 +177,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.2179,
-            "mapping_mode": "exact",
+            "mapping_confidence": 0.6271,
+            "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
               "id": "483",
               "label": "Sports"
-            },
-            "tier2": {
-              "id": "496",
-              "label": "Equine Sports"
             }
           },
           "intent": {
@@ -203,31 +191,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9444,
+                "confidence": 0.9469,
                 "confidence_threshold": 0.22,
                 "label": "consideration",
                 "meets_threshold": true,
-                "raw_confidence": 0.9461
+                "raw_confidence": 0.954
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.4804,
+                "confidence": 0.4849,
                 "confidence_threshold": 0.25,
                 "label": "comparison",
                 "meets_threshold": true,
-                "raw_confidence": 0.4327
+                "raw_confidence": 0.4322
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.981,
+                "confidence": 0.9863,
                 "confidence_threshold": 0.4,
                 "label": "commercial",
                 "meets_threshold": true,
-                "raw_confidence": 0.9653
+                "raw_confidence": 0.9724
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.4804,
+            "confidence": 0.4849,
             "decision_phase": "consideration",
             "subtype": "comparison",
             "summary": "Classified as commercial intent with subtype comparison in the consideration phase.",
@@ -272,17 +260,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.3122,
-            "mapping_mode": "exact",
+            "mapping_confidence": 0.1648,
+            "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
               "id": "596",
               "label": "Technology & Computing"
-            },
-            "tier2": {
-              "id": "638",
-              "label": "Robotics"
             }
           },
           "intent": {
@@ -290,31 +274,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.8858,
+                "confidence": 0.9303,
                 "confidence_threshold": 0.22,
                 "label": "consideration",
                 "meets_threshold": true,
-                "raw_confidence": 0.8885
+                "raw_confidence": 0.9389
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9538,
+                "confidence": 0.9598,
                 "confidence_threshold": 0.25,
                 "label": "comparison",
                 "meets_threshold": true,
-                "raw_confidence": 0.9083
+                "raw_confidence": 0.9157
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9676,
+                "confidence": 0.9746,
                 "confidence_threshold": 0.4,
                 "label": "commercial",
                 "meets_threshold": true,
-                "raw_confidence": 0.9435
+                "raw_confidence": 0.953
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.8858,
+            "confidence": 0.9303,
             "decision_phase": "consideration",
             "subtype": "comparison",
             "summary": "Classified as commercial intent with subtype comparison in the consideration phase.",
@@ -359,13 +343,21 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.5309,
+            "mapping_confidence": 0.1701,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
-              "id": "596",
-              "label": "Technology & Computing"
+              "id": "52",
+              "label": "Business and Finance"
+            },
+            "tier2": {
+              "id": "53",
+              "label": "Business"
+            },
+            "tier3": {
+              "id": "61",
+              "label": "Startups"
             }
           },
           "intent": {
@@ -373,31 +365,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.6077,
+                "confidence": 0.6389,
                 "confidence_threshold": 0.22,
                 "label": "decision",
                 "meets_threshold": true,
-                "raw_confidence": 0.6097
+                "raw_confidence": 0.6498
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.7801,
+                "confidence": 0.7851,
                 "confidence_threshold": 0.25,
                 "label": "provider_selection",
                 "meets_threshold": true,
-                "raw_confidence": 0.6968
+                "raw_confidence": 0.6921
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9843,
+                "confidence": 0.9784,
                 "confidence_threshold": 0.4,
                 "label": "commercial",
                 "meets_threshold": true,
-                "raw_confidence": 0.9703
+                "raw_confidence": 0.9591
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.6077,
+            "confidence": 0.6389,
             "decision_phase": "decision",
             "subtype": "provider_selection",
             "summary": "Classified as commercial intent with subtype provider_selection in the decision phase.",
@@ -442,17 +434,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.2299,
-            "mapping_mode": "exact",
+            "mapping_confidence": 0.24,
+            "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
-              "id": "v9i3On",
-              "label": "Sensitive Topics"
-            },
-            "tier2": {
-              "id": "XtODT3",
-              "label": "Crime & Harmful Acts to Individuals, Society & Human Right Violations"
+              "id": "483",
+              "label": "Sports"
             }
           },
           "intent": {
@@ -460,31 +448,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9662,
+                "confidence": 0.937,
                 "confidence_threshold": 0.22,
                 "label": "action",
                 "meets_threshold": true,
-                "raw_confidence": 0.9674
+                "raw_confidence": 0.9451
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9473,
+                "confidence": 0.9242,
                 "confidence_threshold": 0.25,
                 "label": "signup",
                 "meets_threshold": true,
-                "raw_confidence": 0.8993
+                "raw_confidence": 0.8636
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9788,
+                "confidence": 0.9809,
                 "confidence_threshold": 0.4,
                 "label": "transactional",
                 "meets_threshold": true,
-                "raw_confidence": 0.9614
+                "raw_confidence": 0.9633
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.9473,
+            "confidence": 0.9242,
             "decision_phase": "action",
             "subtype": "signup",
             "summary": "Classified as transactional intent with subtype signup in the action phase.",
@@ -529,7 +517,7 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.8304,
+            "mapping_confidence": 0.4786,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
@@ -547,31 +535,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9595,
+                "confidence": 0.9436,
                 "confidence_threshold": 0.22,
                 "label": "action",
                 "meets_threshold": true,
-                "raw_confidence": 0.9608
+                "raw_confidence": 0.951
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.8434,
+                "confidence": 0.8891,
                 "confidence_threshold": 0.25,
                 "label": "booking",
                 "meets_threshold": true,
-                "raw_confidence": 0.7616
+                "raw_confidence": 0.8107
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9805,
+                "confidence": 0.9715,
                 "confidence_threshold": 0.4,
                 "label": "transactional",
                 "meets_threshold": true,
-                "raw_confidence": 0.9649
+                "raw_confidence": 0.9481
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.8434,
+            "confidence": 0.8891,
             "decision_phase": "action",
             "subtype": "booking",
             "summary": "Classified as transactional intent with subtype booking in the action phase.",
@@ -616,8 +604,8 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.5261,
-            "mapping_mode": "exact",
+            "mapping_confidence": 0.3826,
+            "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
@@ -630,31 +618,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9573,
+                "confidence": 0.9613,
                 "confidence_threshold": 0.22,
                 "label": "post_purchase",
                 "meets_threshold": true,
-                "raw_confidence": 0.9587
+                "raw_confidence": 0.9669
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.967,
+                "confidence": 0.965,
                 "confidence_threshold": 0.25,
                 "label": "onboarding_setup",
                 "meets_threshold": true,
-                "raw_confidence": 0.9306
+                "raw_confidence": 0.9235
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.5834,
+                "confidence": 0.5393,
                 "confidence_threshold": 0.4,
                 "label": "transactional",
                 "meets_threshold": true,
-                "raw_confidence": 0.5253
+                "raw_confidence": 0.4772
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.5834,
+            "confidence": 0.5393,
             "decision_phase": "post_purchase",
             "subtype": "onboarding_setup",
             "summary": "Classified as transactional intent with subtype onboarding_setup in the post_purchase phase.",
@@ -699,21 +687,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.272,
-            "mapping_mode": "exact",
+            "mapping_confidence": 0.3628,
+            "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
-              "id": "52",
-              "label": "Business and Finance"
-            },
-            "tier2": {
-              "id": "53",
-              "label": "Business"
-            },
-            "tier3": {
-              "id": "72",
-              "label": "Business I.T."
+              "id": "391",
+              "label": "Personal Finance"
             }
           },
           "intent": {
@@ -721,31 +701,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9589,
+                "confidence": 0.9481,
                 "confidence_threshold": 0.22,
                 "label": "support",
                 "meets_threshold": true,
-                "raw_confidence": 0.9603
+                "raw_confidence": 0.9551
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.8859,
+                "confidence": 0.934,
                 "confidence_threshold": 0.25,
                 "label": "account_help",
                 "meets_threshold": true,
-                "raw_confidence": 0.8147
+                "raw_confidence": 0.8749
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9699,
+                "confidence": 0.9542,
                 "confidence_threshold": 0.4,
                 "label": "support",
                 "meets_threshold": true,
-                "raw_confidence": 0.9476
+                "raw_confidence": 0.9232
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.8859,
+            "confidence": 0.934,
             "decision_phase": "support",
             "subtype": "account_help",
             "summary": "Classified as support intent with subtype account_help in the support phase.",
@@ -796,8 +776,8 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.7892,
-            "mapping_mode": "exact",
+            "mapping_confidence": 0.3231,
+            "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
@@ -810,31 +790,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9219,
+                "confidence": 0.8468,
                 "confidence_threshold": 0.22,
                 "label": "awareness",
                 "meets_threshold": true,
-                "raw_confidence": 0.9239
+                "raw_confidence": 0.8606
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9492,
+                "confidence": 0.9639,
                 "confidence_threshold": 0.25,
                 "label": "emotional_reflection",
                 "meets_threshold": true,
-                "raw_confidence": 0.9021
+                "raw_confidence": 0.9211
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9388,
+                "confidence": 0.9627,
                 "confidence_threshold": 0.4,
                 "label": "personal_reflection",
                 "meets_threshold": true,
-                "raw_confidence": 0.9059
+                "raw_confidence": 0.9348
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.9219,
+            "confidence": 0.8468,
             "decision_phase": "awareness",
             "subtype": "emotional_reflection",
             "summary": "Classified as personal_reflection intent with subtype emotional_reflection in the awareness phase.",
@@ -885,7 +865,7 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.2238,
+            "mapping_confidence": 0.3327,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
@@ -899,31 +879,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.8763,
+                "confidence": 0.8651,
                 "confidence_threshold": 0.22,
                 "label": "research",
                 "meets_threshold": true,
-                "raw_confidence": 0.8791
+                "raw_confidence": 0.8781
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9683,
+                "confidence": 0.9652,
                 "confidence_threshold": 0.25,
                 "label": "follow_up",
                 "meets_threshold": true,
-                "raw_confidence": 0.9314
+                "raw_confidence": 0.9229
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9623,
+                "confidence": 0.9746,
                 "confidence_threshold": 0.4,
                 "label": "ambiguous",
                 "meets_threshold": true,
-                "raw_confidence": 0.9367
+                "raw_confidence": 0.9541
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.8763,
+            "confidence": 0.8651,
             "decision_phase": "research",
             "subtype": "follow_up",
             "summary": "Classified as ambiguous intent with subtype follow_up in the research phase.",
@@ -974,17 +954,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.2371,
+            "mapping_confidence": 0.1481,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
               "id": "391",
               "label": "Personal Finance"
-            },
-            "tier2": {
-              "id": "396",
-              "label": "Financial Planning"
             }
           },
           "intent": {
@@ -992,31 +968,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9225,
+                "confidence": 0.9177,
                 "confidence_threshold": 0.22,
                 "label": "research",
                 "meets_threshold": true,
-                "raw_confidence": 0.9246
+                "raw_confidence": 0.9273
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9586,
+                "confidence": 0.9506,
                 "confidence_threshold": 0.25,
                 "label": "follow_up",
                 "meets_threshold": true,
-                "raw_confidence": 0.9146
+                "raw_confidence": 0.8983
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9488,
+                "confidence": 0.9628,
                 "confidence_threshold": 0.4,
                 "label": "ambiguous",
                 "meets_threshold": true,
-                "raw_confidence": 0.9179
+                "raw_confidence": 0.9356
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.9225,
+            "confidence": 0.9177,
             "decision_phase": "research",
             "subtype": "follow_up",
             "summary": "Classified as ambiguous intent with subtype follow_up in the research phase.",
@@ -1067,13 +1043,13 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.2131,
+            "mapping_confidence": 0.0729,
             "mapping_mode": "nearest_equivalent",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
-              "id": "42",
-              "label": "Books and Literature"
+              "id": "123",
+              "label": "Careers"
             }
           },
           "intent": {
@@ -1081,31 +1057,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9861,
+                "confidence": 0.9739,
                 "confidence_threshold": 0.22,
                 "label": "action",
                 "meets_threshold": true,
-                "raw_confidence": 0.9867
+                "raw_confidence": 0.9781
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.7335,
+                "confidence": 0.7259,
                 "confidence_threshold": 0.25,
                 "label": "signup",
                 "meets_threshold": true,
-                "raw_confidence": 0.6454
+                "raw_confidence": 0.6331
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9628,
+                "confidence": 0.9763,
                 "confidence_threshold": 0.4,
                 "label": "transactional",
                 "meets_threshold": true,
-                "raw_confidence": 0.938
+                "raw_confidence": 0.9557
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.7335,
+            "confidence": 0.7259,
             "decision_phase": "action",
             "subtype": "signup",
             "summary": "Classified as transactional intent with subtype signup in the action phase.",
@@ -1150,17 +1126,17 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.3327,
+            "mapping_confidence": 0.1383,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
             "tier1": {
-              "id": "596",
-              "label": "Technology & Computing"
+              "id": "123",
+              "label": "Careers"
             },
             "tier2": {
-              "id": "639",
-              "label": "Virtual Reality"
+              "id": "127",
+              "label": "Job Search"
             }
           },
           "intent": {
@@ -1168,31 +1144,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9295,
+                "confidence": 0.9578,
                 "confidence_threshold": 0.22,
                 "label": "consideration",
                 "meets_threshold": true,
-                "raw_confidence": 0.9315
+                "raw_confidence": 0.9639
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9374,
+                "confidence": 0.9095,
                 "confidence_threshold": 0.25,
                 "label": "comparison",
                 "meets_threshold": true,
-                "raw_confidence": 0.8838
+                "raw_confidence": 0.8429
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9602,
+                "confidence": 0.9747,
                 "confidence_threshold": 0.4,
                 "label": "commercial",
                 "meets_threshold": true,
-                "raw_confidence": 0.9329
+                "raw_confidence": 0.953
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.9295,
+            "confidence": 0.9095,
             "decision_phase": "consideration",
             "subtype": "comparison",
             "summary": "Classified as commercial intent with subtype comparison in the consideration phase.",
@@ -1237,7 +1213,7 @@
       "model_output": {
         "classification": {
           "iab_content": {
-            "mapping_confidence": 0.3227,
+            "mapping_confidence": 0.1608,
             "mapping_mode": "exact",
             "taxonomy": "IAB Content Taxonomy",
             "taxonomy_version": "3.0",
@@ -1251,31 +1227,31 @@
             "component_confidence": {
               "decision_phase": {
                 "calibrated": true,
-                "confidence": 0.9535,
+                "confidence": 0.9436,
                 "confidence_threshold": 0.22,
                 "label": "awareness",
                 "meets_threshold": true,
-                "raw_confidence": 0.955
+                "raw_confidence": 0.951
               },
               "intent_subtype": {
                 "calibrated": true,
-                "confidence": 0.9793,
+                "confidence": 0.9692,
                 "confidence_threshold": 0.25,
                 "label": "education",
                 "meets_threshold": true,
-                "raw_confidence": 0.9527
+                "raw_confidence": 0.931
               },
               "intent_type": {
                 "calibrated": true,
-                "confidence": 0.9769,
+                "confidence": 0.9775,
                 "confidence_threshold": 0.4,
                 "label": "informational",
                 "meets_threshold": true,
-                "raw_confidence": 0.9584
+                "raw_confidence": 0.9578
               },
               "overall_strategy": "min_required_component_confidence"
             },
-            "confidence": 0.9535,
+            "confidence": 0.9436,
             "decision_phase": "awareness",
             "subtype": "education",
             "summary": "Classified as informational intent with subtype education in the awareness phase.",
diff --git a/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_confusion_matrix.csv b/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_confusion_matrix.csv
index d8629f9e7297b88be3145c27c471f3daf9a612e3..7202242ef2e3a9c8e69609744a3de88723b07f62 100644
--- a/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_confusion_matrix.csv
@@ -1,8 +1,8 @@
 ,awareness,research,consideration,decision,action,post_purchase,support
 awareness,14,1,0,0,0,0,0
 research,0,15,0,0,0,0,0
-consideration,0,2,13,0,0,0,0
+consideration,0,1,14,0,0,0,0
 decision,0,1,0,14,0,0,0
-action,0,0,0,1,14,0,0
+action,0,1,0,0,14,0,0
 post_purchase,0,0,0,0,0,15,0
 support,0,0,0,0,0,0,15
diff --git a/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_report.json b/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_report.json
index 6611eedc4620df66249f5d22a22d6faec44d5e48..acc5ea446813bf822b75ceb38fb4248b18d3a87e 100644
--- a/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_report.json
+++ b/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_report.json
@@ -1,7 +1,7 @@
 {
-  "accepted_accuracy": 0.9524,
+  "accepted_accuracy": 0.9619,
   "accepted_coverage": 1.0,
-  "accuracy": 0.9524,
+  "accuracy": 0.9619,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_confusion_matrix.csv",
   "count": 105,
   "dataset_path": "/content/agentic-intent-classifier/data/decision_phase_benchmark.jsonl",
@@ -15,12 +15,12 @@
       "macro_f1": 0.9711
     },
     "hard": {
-      "accepted_accuracy": 0.8857,
+      "accepted_accuracy": 0.9143,
       "accepted_coverage": 1.0,
-      "accuracy": 0.8857,
+      "accuracy": 0.9143,
       "count": 35,
       "fallback_rate": 0.0,
-      "macro_f1": 0.8908
+      "macro_f1": 0.9194
     },
     "medium": {
       "accepted_accuracy": 1.0,
@@ -33,9 +33,9 @@
   },
   "fallback_rate": 0.0,
   "head": "decision_phase",
-  "macro_f1": 0.9536,
+  "macro_f1": 0.9635,
   "per_class_metrics": {
-    "accuracy": 0.9523809523809523,
+    "accuracy": 0.9619047619047619,
     "action": {
       "f1-score": 0.9655172413793104,
       "precision": 1.0,
@@ -49,21 +49,21 @@
       "support": 15.0
     },
     "consideration": {
-      "f1-score": 0.9285714285714286,
+      "f1-score": 0.9655172413793104,
       "precision": 1.0,
-      "recall": 0.8666666666666667,
+      "recall": 0.9333333333333333,
       "support": 15.0
     },
     "decision": {
-      "f1-score": 0.9333333333333333,
-      "precision": 0.9333333333333333,
+      "f1-score": 0.9655172413793104,
+      "precision": 1.0,
       "recall": 0.9333333333333333,
       "support": 15.0
     },
     "macro avg": {
-      "f1-score": 0.9536131694056934,
-      "precision": 0.9604010025062657,
-      "recall": 0.9523809523809524,
+      "f1-score": 0.9634888438133874,
+      "precision": 0.9699248120300752,
+      "recall": 0.9619047619047619,
       "support": 105.0
     },
     "post_purchase": {
@@ -85,9 +85,9 @@
       "support": 15.0
     },
     "weighted avg": {
-      "f1-score": 0.9536131694056934,
-      "precision": 0.9604010025062656,
-      "recall": 0.9523809523809523,
+      "f1-score": 0.9634888438133875,
+      "precision": 0.9699248120300752,
+      "recall": 0.9619047619047619,
       "support": 105.0
     }
   },
diff --git a/artifacts/evaluation/latest/decision_phase_final_wave_cases_confusion_matrix.csv b/artifacts/evaluation/latest/decision_phase_final_wave_cases_confusion_matrix.csv
index 7066b2cd2062314bb54e438130f1d5a1cb7ecb9c..5eec12148ff61e8dc414ad8803de0fd9b41e9e95 100644
--- a/artifacts/evaluation/latest/decision_phase_final_wave_cases_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/decision_phase_final_wave_cases_confusion_matrix.csv
@@ -1,7 +1,7 @@
 ,awareness,research,consideration,decision,action,post_purchase,support
 awareness,5,0,0,0,0,0,0
-research,1,3,0,0,0,0,0
-consideration,0,2,3,0,0,0,0
+research,3,1,0,0,0,0,0
+consideration,0,1,4,0,0,0,0
 decision,0,0,0,5,0,0,0
 action,0,0,0,0,0,0,0
 post_purchase,0,0,0,0,0,4,0
diff --git a/artifacts/evaluation/latest/decision_phase_final_wave_cases_report.json b/artifacts/evaluation/latest/decision_phase_final_wave_cases_report.json
index 3eb90650ac8929e108c96942ba6b2f26bed04c42..323d2f8a894f7fe1444bf5df8afa2f2041773640 100644
--- a/artifacts/evaluation/latest/decision_phase_final_wave_cases_report.json
+++ b/artifacts/evaluation/latest/decision_phase_final_wave_cases_report.json
@@ -1,15 +1,15 @@
 {
-  "accepted_accuracy": 0.8889,
+  "accepted_accuracy": 0.8519,
   "accepted_coverage": 1.0,
-  "accuracy": 0.8889,
+  "accuracy": 0.8519,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_final_wave_cases_confusion_matrix.csv",
   "count": 27,
   "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/final_wave_cases.jsonl",
   "fallback_rate": 0.0,
   "head": "decision_phase",
-  "macro_f1": 0.8876,
+  "macro_f1": 0.8319,
   "per_class_metrics": {
-    "accuracy": 0.8888888888888888,
+    "accuracy": 0.8518518518518519,
     "action": {
       "f1-score": 0.0,
       "precision": 0.0,
@@ -17,15 +17,15 @@
       "support": 0.0
     },
     "awareness": {
-      "f1-score": 0.9090909090909091,
-      "precision": 0.8333333333333334,
+      "f1-score": 0.7692307692307693,
+      "precision": 0.625,
       "recall": 1.0,
       "support": 5.0
     },
     "consideration": {
-      "f1-score": 0.75,
+      "f1-score": 0.8888888888888888,
       "precision": 1.0,
-      "recall": 0.6,
+      "recall": 0.8,
       "support": 5.0
     },
     "decision": {
@@ -35,9 +35,9 @@
       "support": 5.0
     },
     "macro avg": {
-      "f1-score": 0.7608225108225108,
-      "precision": 0.7761904761904762,
-      "recall": 0.7642857142857142,
+      "f1-score": 0.7130647130647131,
+      "precision": 0.7321428571428571,
+      "recall": 0.7214285714285714,
       "support": 27.0
     },
     "post_purchase": {
@@ -47,9 +47,9 @@
       "support": 4.0
     },
     "research": {
-      "f1-score": 0.6666666666666666,
-      "precision": 0.6,
-      "recall": 0.75,
+      "f1-score": 0.3333333333333333,
+      "precision": 0.5,
+      "recall": 0.25,
       "support": 4.0
     },
     "support": {
@@ -59,9 +59,9 @@
       "support": 4.0
     },
     "weighted avg": {
-      "f1-score": 0.8874859708193041,
-      "precision": 0.9098765432098765,
-      "recall": 0.8888888888888888,
+      "f1-score": 0.8379233934789491,
+      "precision": 0.8564814814814815,
+      "recall": 0.8518518518518519,
       "support": 27.0
     }
   },
diff --git a/artifacts/evaluation/latest/decision_phase_test_confusion_matrix.csv b/artifacts/evaluation/latest/decision_phase_test_confusion_matrix.csv
index 7ee810e23bfb8a0cc3747e0962776fc940f083ed..dc7f621094e1b20427da3662992b82da4002a7b5 100644
--- a/artifacts/evaluation/latest/decision_phase_test_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/decision_phase_test_confusion_matrix.csv
@@ -1,8 +1,8 @@
 ,awareness,research,consideration,decision,action,post_purchase,support
 awareness,3,0,0,0,0,0,0
-research,3,2,0,0,0,0,0
+research,4,1,0,0,0,0,0
 consideration,0,2,3,0,0,0,0
-decision,0,0,0,4,0,1,0
+decision,0,0,0,5,0,0,0
 action,0,0,0,0,3,0,0
 post_purchase,0,0,0,0,0,4,0
 support,0,0,0,0,0,1,3
diff --git a/artifacts/evaluation/latest/decision_phase_test_report.json b/artifacts/evaluation/latest/decision_phase_test_report.json
index e87d6de18080785defa8596fe97d628346bbe418..cd8290a56c2f38f42ad074f35a6e108595a77ebd 100644
--- a/artifacts/evaluation/latest/decision_phase_test_report.json
+++ b/artifacts/evaluation/latest/decision_phase_test_report.json
@@ -7,7 +7,7 @@
   "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/test.jsonl",
   "fallback_rate": 0.0,
   "head": "decision_phase",
-  "macro_f1": 0.7724,
+  "macro_f1": 0.7637,
   "per_class_metrics": {
     "accuracy": 0.7586206896551724,
     "action": {
@@ -17,8 +17,8 @@
       "support": 3.0
     },
     "awareness": {
-      "f1-score": 0.6666666666666666,
-      "precision": 0.5,
+      "f1-score": 0.6,
+      "precision": 0.42857142857142855,
       "recall": 1.0,
       "support": 3.0
     },
@@ -29,27 +29,27 @@
       "support": 5.0
     },
     "decision": {
-      "f1-score": 0.8888888888888888,
+      "f1-score": 1.0,
       "precision": 1.0,
-      "recall": 0.8,
+      "recall": 1.0,
       "support": 5.0
     },
     "macro avg": {
-      "f1-score": 0.7724489795918367,
-      "precision": 0.8095238095238095,
+      "f1-score": 0.763718820861678,
+      "precision": 0.7945578231292517,
       "recall": 0.7928571428571428,
       "support": 29.0
     },
     "post_purchase": {
-      "f1-score": 0.8,
-      "precision": 0.6666666666666666,
+      "f1-score": 0.8888888888888888,
+      "precision": 0.8,
       "recall": 1.0,
       "support": 4.0
     },
     "research": {
-      "f1-score": 0.4444444444444444,
-      "precision": 0.5,
-      "recall": 0.4,
+      "f1-score": 0.25,
+      "precision": 0.3333333333333333,
+      "recall": 0.2,
       "support": 5.0
     },
     "support": {
@@ -59,8 +59,8 @@
       "support": 4.0
     },
     "weighted avg": {
-      "f1-score": 0.7601806239737274,
-      "precision": 0.8160919540229885,
+      "f1-score": 0.7511767925561028,
+      "precision": 0.7983579638752052,
       "recall": 0.7586206896551724,
       "support": 29.0
     }
diff --git a/artifacts/evaluation/latest/decision_phase_train_confusion_matrix.csv b/artifacts/evaluation/latest/decision_phase_train_confusion_matrix.csv
index ead354eb5bbc296162fab26b8847d38e9f27e6ab..236c876baa9dbcafc7278f834bce97012e94edf6 100644
--- a/artifacts/evaluation/latest/decision_phase_train_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/decision_phase_train_confusion_matrix.csv
@@ -1,8 +1,8 @@
 ,awareness,research,consideration,decision,action,post_purchase,support
 awareness,16,0,0,0,0,0,0
 research,1,14,0,0,0,0,0
-consideration,0,4,13,0,0,0,0
-decision,0,0,1,15,0,0,0
+consideration,0,2,15,0,0,0,0
+decision,0,0,0,16,0,0,0
 action,0,0,0,0,10,0,0
 post_purchase,0,0,0,0,0,14,0
 support,0,0,0,0,0,0,14
diff --git a/artifacts/evaluation/latest/decision_phase_train_report.json b/artifacts/evaluation/latest/decision_phase_train_report.json
index ca4b8e52b24daa95d504bffa5e1f7ec961247fa9..7677c764664c9c3998fd1bc3e899e0597d3130de 100644
--- a/artifacts/evaluation/latest/decision_phase_train_report.json
+++ b/artifacts/evaluation/latest/decision_phase_train_report.json
@@ -1,15 +1,15 @@
 {
-  "accepted_accuracy": 0.9412,
+  "accepted_accuracy": 0.9706,
   "accepted_coverage": 1.0,
-  "accuracy": 0.9412,
+  "accuracy": 0.9706,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_train_confusion_matrix.csv",
   "count": 102,
   "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/train.jsonl",
   "fallback_rate": 0.0,
   "head": "decision_phase",
-  "macro_f1": 0.9464,
+  "macro_f1": 0.9729,
   "per_class_metrics": {
-    "accuracy": 0.9411764705882353,
+    "accuracy": 0.9705882352941176,
     "action": {
       "f1-score": 1.0,
       "precision": 1.0,
@@ -23,21 +23,21 @@
       "support": 16.0
     },
     "consideration": {
-      "f1-score": 0.8387096774193549,
-      "precision": 0.9285714285714286,
-      "recall": 0.7647058823529411,
+      "f1-score": 0.9375,
+      "precision": 1.0,
+      "recall": 0.8823529411764706,
       "support": 17.0
     },
     "decision": {
-      "f1-score": 0.967741935483871,
+      "f1-score": 1.0,
       "precision": 1.0,
-      "recall": 0.9375,
+      "recall": 1.0,
       "support": 16.0
     },
     "macro avg": {
-      "f1-score": 0.9463762044407206,
-      "precision": 0.9496465252767774,
-      "recall": 0.9479341736694679,
+      "f1-score": 0.9729175394497975,
+      "precision": 0.9737394957983193,
+      "recall": 0.9736694677871148,
       "support": 102.0
     },
     "post_purchase": {
@@ -47,8 +47,8 @@
       "support": 14.0
     },
     "research": {
-      "f1-score": 0.8484848484848485,
-      "precision": 0.7777777777777778,
+      "f1-score": 0.9032258064516129,
+      "precision": 0.875,
       "recall": 0.9333333333333333,
       "support": 15.0
     },
@@ -59,9 +59,9 @@
       "support": 14.0
     },
     "weighted avg": {
-      "f1-score": 0.9410231345715216,
-      "precision": 0.946188279233262,
-      "recall": 0.9411764705882353,
+      "f1-score": 0.9705984177639775,
+      "precision": 0.9723904267589389,
+      "recall": 0.9705882352941176,
       "support": 102.0
     }
   },
diff --git a/artifacts/evaluation/latest/decision_phase_val_confusion_matrix.csv b/artifacts/evaluation/latest/decision_phase_val_confusion_matrix.csv
index 36b4bb778c2261e6bdf697a55cfcf935669db80a..cc0caef96905f8e129271e3425b69d3b226cffa3 100644
--- a/artifacts/evaluation/latest/decision_phase_val_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/decision_phase_val_confusion_matrix.csv
@@ -1,8 +1,8 @@
 ,awareness,research,consideration,decision,action,post_purchase,support
 awareness,5,0,0,0,0,0,0
-research,1,3,0,0,0,0,0
+research,2,2,0,0,0,0,0
 consideration,0,0,5,0,0,0,0
 decision,0,0,1,3,0,0,0
 action,0,0,0,0,3,0,0
 post_purchase,0,1,0,0,0,3,0
-support,0,0,0,0,1,0,3
+support,0,0,0,0,0,1,3
diff --git a/artifacts/evaluation/latest/decision_phase_val_report.json b/artifacts/evaluation/latest/decision_phase_val_report.json
index a22554f814b4084ed90614f08a4ad33fbec443be..ca55fe08985e1ca262142d1969b0e8e3e1163a4d 100644
--- a/artifacts/evaluation/latest/decision_phase_val_report.json
+++ b/artifacts/evaluation/latest/decision_phase_val_report.json
@@ -1,24 +1,24 @@
 {
-  "accepted_accuracy": 0.8621,
+  "accepted_accuracy": 0.8276,
   "accepted_coverage": 1.0,
-  "accuracy": 0.8621,
+  "accuracy": 0.8276,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_val_confusion_matrix.csv",
   "count": 29,
   "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/val.jsonl",
   "fallback_rate": 0.0,
   "head": "decision_phase",
-  "macro_f1": 0.8567,
+  "macro_f1": 0.8254,
   "per_class_metrics": {
-    "accuracy": 0.8620689655172413,
+    "accuracy": 0.8275862068965517,
     "action": {
-      "f1-score": 0.8571428571428571,
-      "precision": 0.75,
+      "f1-score": 1.0,
+      "precision": 1.0,
       "recall": 1.0,
       "support": 3.0
     },
     "awareness": {
-      "f1-score": 0.9090909090909091,
-      "precision": 0.8333333333333334,
+      "f1-score": 0.8333333333333334,
+      "precision": 0.7142857142857143,
       "recall": 1.0,
       "support": 5.0
     },
@@ -35,21 +35,21 @@
       "support": 4.0
     },
     "macro avg": {
-      "f1-score": 0.8566790352504637,
-      "precision": 0.880952380952381,
-      "recall": 0.8571428571428571,
+      "f1-score": 0.8254483611626469,
+      "precision": 0.8520408163265306,
+      "recall": 0.8214285714285714,
       "support": 29.0
     },
     "post_purchase": {
-      "f1-score": 0.8571428571428571,
-      "precision": 1.0,
+      "f1-score": 0.75,
+      "precision": 0.75,
       "recall": 0.75,
       "support": 4.0
     },
     "research": {
-      "f1-score": 0.75,
-      "precision": 0.75,
-      "recall": 0.75,
+      "f1-score": 0.5714285714285714,
+      "precision": 0.6666666666666666,
+      "recall": 0.5,
       "support": 4.0
     },
     "support": {
@@ -59,9 +59,9 @@
       "support": 4.0
     },
     "weighted avg": {
-      "f1-score": 0.8602776533811015,
-      "precision": 0.8821839080459771,
-      "recall": 0.8620689655172413,
+      "f1-score": 0.822585460516495,
+      "precision": 0.8415435139573071,
+      "recall": 0.8275862068965517,
       "support": 29.0
     }
   },
diff --git a/artifacts/evaluation/latest/iab_behavior_lock_regression.json b/artifacts/evaluation/latest/iab_behavior_lock_regression.json
index 53e99132e43ed85484e31771835d74752e087a91..8998938437c0d692651a727f43f935f4c94cecf7 100644
--- a/artifacts/evaluation/latest/iab_behavior_lock_regression.json
+++ b/artifacts/evaluation/latest/iab_behavior_lock_regression.json
@@ -13,7 +13,7 @@
   "results": [
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Automotive",
         "model_output.classification.iab_content.tier2.label": null
       },
@@ -24,11 +24,6 @@
       },
       "id": "car-buying-maps-to-automotive-buying",
       "mismatches": [
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
         {
           "actual": null,
           "expected": "Auto Type",
@@ -106,8 +101,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -121,6 +116,16 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Software and Applications",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "CRM education should resolve to the closest business/sales path, not generic software.",
@@ -130,9 +135,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Robotics",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -144,12 +149,7 @@
       "id": "crm-comparison-maps-to-sales",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Robotics",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -166,9 +166,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Careers",
-        "model_output.classification.iab_content.tier2.label": "Job Search",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -185,12 +185,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Job Search",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -208,7 +203,7 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-        "model_output.classification.iab_content.tier1.label": "Science"
+        "model_output.classification.iab_content.tier1.label": "Real Estate"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -217,7 +212,7 @@
       "id": "ml-explanation-maps-to-ai",
       "mismatches": [
         {
-          "actual": "Science",
+          "actual": "Real Estate",
           "expected": "Technology & Computing",
           "path": "model_output.classification.iab_content.tier1.label"
         }
@@ -229,10 +224,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Personal Finance",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -243,12 +238,17 @@
       "id": "support-credential-help-maps-to-business-it",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Personal Finance",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Information and Network Security",
+          "actual": null,
           "expected": "Internet",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -284,9 +284,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-        "model_output.classification.iab_content.tier2.label": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Sports",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -298,17 +298,12 @@
       "id": "trial-signup-maps-to-software",
       "mismatches": [
         {
-          "actual": "Sensitive Topics",
+          "actual": "Sports",
           "expected": "Hobbies & Interests",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+          "actual": null,
           "expected": "Content Production",
           "path": "model_output.classification.iab_content.tier2.label"
         },
diff --git a/artifacts/evaluation/latest/iab_content_cross_vertical_benchmark_report.json b/artifacts/evaluation/latest/iab_content_cross_vertical_benchmark_report.json
index 0afa8ca3e8e1857c41d097763a88551bc37cac7c..cd86f797b235c0670b111933adb49e888554cba0 100644
--- a/artifacts/evaluation/latest/iab_content_cross_vertical_benchmark_report.json
+++ b/artifacts/evaluation/latest/iab_content_cross_vertical_benchmark_report.json
@@ -1,90 +1,90 @@
 {
-  "accepted_accuracy": 0.427,
-  "accepted_coverage": 0.9889,
-  "accuracy": 0.4222,
+  "accepted_accuracy": 0.3108,
+  "accepted_coverage": 0.8222,
+  "accuracy": 0.2556,
   "count": 90,
   "dataset_path": "/content/agentic-intent-classifier/data/iab_cross_vertical_benchmark.jsonl",
   "difficulty_breakdown": {
     "easy": {
-      "accepted_accuracy": 0.4138,
-      "accepted_coverage": 0.9667,
-      "accuracy": 0.4,
+      "accepted_accuracy": 0.3636,
+      "accepted_coverage": 0.7333,
+      "accuracy": 0.2667,
       "count": 30,
-      "fallback_rate": 0.0333,
-      "macro_f1": 0.2727
+      "fallback_rate": 0.2667,
+      "macro_f1": 0.1778
     },
     "hard": {
-      "accepted_accuracy": 0.4667,
-      "accepted_coverage": 1.0,
-      "accuracy": 0.4667,
+      "accepted_accuracy": 0.3077,
+      "accepted_coverage": 0.8667,
+      "accuracy": 0.2667,
       "count": 30,
-      "fallback_rate": 0.0,
-      "macro_f1": 0.3106
+      "fallback_rate": 0.1333,
+      "macro_f1": 0.1562
     },
     "medium": {
-      "accepted_accuracy": 0.4,
-      "accepted_coverage": 1.0,
-      "accuracy": 0.4,
+      "accepted_accuracy": 0.2692,
+      "accepted_coverage": 0.8667,
+      "accuracy": 0.2333,
       "count": 30,
-      "fallback_rate": 0.0,
-      "macro_f1": 0.2667
+      "fallback_rate": 0.1333,
+      "macro_f1": 0.1591
     }
   },
-  "fallback_rate": 0.0111,
+  "fallback_rate": 0.1778,
   "head": "iab_content",
-  "macro_f1": 0.227,
+  "macro_f1": 0.1228,
   "primary_source": "supervised_classifier",
   "suite": "cross_vertical_benchmark",
   "tier_metrics": {
-    "average_prediction_depth": 2.4,
+    "average_prediction_depth": 1.9222,
     "error_buckets": {
-      "exact_match": 38,
-      "parent_safe_stop": 1,
-      "right_tier1_wrong_tier2": 14,
+      "exact_match": 23,
+      "parent_safe_stop": 3,
+      "right_tier1_wrong_tier2": 23,
       "wrong_deep_leaf": 8,
-      "wrong_tier1": 29
+      "wrong_tier1": 33
     },
-    "exact_path_accuracy": 0.4222,
-    "parent_safe_accuracy": 0.4444,
-    "tier1_accuracy": 0.6778,
-    "tier2_accuracy": 0.4881,
-    "tier3_accuracy": 0.5238,
-    "tier4_accuracy": 0.5
+    "exact_path_accuracy": 0.2556,
+    "parent_safe_accuracy": 0.4222,
+    "tier1_accuracy": 0.6333,
+    "tier2_accuracy": 0.3571,
+    "tier3_accuracy": 0.2381,
+    "tier4_accuracy": 0.0
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.4,
+      "average_prediction_depth": 1.9222,
       "error_buckets": {
-        "exact_match": 37,
-        "parent_safe_stop": 1,
-        "right_tier1_wrong_tier2": 14,
-        "wrong_deep_leaf": 9,
-        "wrong_tier1": 29
+        "exact_match": 23,
+        "parent_safe_stop": 3,
+        "right_tier1_wrong_tier2": 23,
+        "wrong_deep_leaf": 8,
+        "wrong_tier1": 33
       },
-      "exact_path_accuracy": 0.4111,
-      "parent_safe_accuracy": 0.4333,
-      "tier1_accuracy": 0.6778,
-      "tier2_accuracy": 0.4881,
-      "tier3_accuracy": 0.4762,
-      "tier4_accuracy": 0.5
+      "exact_path_accuracy": 0.2556,
+      "parent_safe_accuracy": 0.4222,
+      "tier1_accuracy": 0.6333,
+      "tier2_accuracy": 0.3571,
+      "tier3_accuracy": 0.2381,
+      "tier4_accuracy": 0.0
     },
     "combined_path": {
-      "average_prediction_depth": 2.4,
+      "average_prediction_depth": 1.9222,
       "error_buckets": {
-        "exact_match": 37,
-        "parent_safe_stop": 1,
-        "right_tier1_wrong_tier2": 14,
-        "wrong_deep_leaf": 9,
-        "wrong_tier1": 29
+        "exact_match": 23,
+        "parent_safe_stop": 3,
+        "right_tier1_wrong_tier2": 23,
+        "wrong_deep_leaf": 8,
+        "wrong_tier1": 33
       },
-      "exact_path_accuracy": 0.4111,
-      "fallback_overuse_count": 25,
-      "fallback_rate": 0.2778,
-      "parent_safe_accuracy": 0.4333,
-      "tier1_accuracy": 0.6778,
-      "tier2_accuracy": 0.4881,
-      "tier3_accuracy": 0.4762,
-      "tier4_accuracy": 0.5
+      "exact_path_accuracy": 0.2556,
+      "fallback_overuse_count": 19,
+      "fallback_rate": 0.2111,
+      "parent_safe_accuracy": 0.4222,
+      "tier1_accuracy": 0.6333,
+      "tier2_accuracy": 0.3571,
+      "tier3_accuracy": 0.2381,
+      "tier4_accuracy": 0.0
     },
     "disagreements": {
       "classifier_vs_combined": 0
diff --git a/artifacts/evaluation/latest/iab_content_difficulty_benchmark_report.json b/artifacts/evaluation/latest/iab_content_difficulty_benchmark_report.json
index 4e5d6b9465b37fb5007a3b6237115dba9b3e1eb9..1d8534bf2659e4204b8e20a7115ad0c54bfc9906 100644
--- a/artifacts/evaluation/latest/iab_content_difficulty_benchmark_report.json
+++ b/artifacts/evaluation/latest/iab_content_difficulty_benchmark_report.json
@@ -1,90 +1,90 @@
 {
-  "accepted_accuracy": 0.4231,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.4231,
+  "accepted_accuracy": 0.32,
+  "accepted_coverage": 0.8013,
+  "accuracy": 0.2564,
   "count": 156,
   "dataset_path": "/content/agentic-intent-classifier/data/iab_benchmark.jsonl",
   "difficulty_breakdown": {
     "easy": {
-      "accepted_accuracy": 0.4615,
-      "accepted_coverage": 1.0,
-      "accuracy": 0.4615,
+      "accepted_accuracy": 0.35,
+      "accepted_coverage": 0.7692,
+      "accuracy": 0.2692,
       "count": 52,
-      "fallback_rate": 0.0,
-      "macro_f1": 0.2359
+      "fallback_rate": 0.2308,
+      "macro_f1": 0.153
     },
     "hard": {
-      "accepted_accuracy": 0.3654,
-      "accepted_coverage": 1.0,
-      "accuracy": 0.3654,
+      "accepted_accuracy": 0.275,
+      "accepted_coverage": 0.7692,
+      "accuracy": 0.2115,
       "count": 52,
-      "fallback_rate": 0.0,
-      "macro_f1": 0.1892
+      "fallback_rate": 0.2308,
+      "macro_f1": 0.1108
     },
     "medium": {
-      "accepted_accuracy": 0.4423,
-      "accepted_coverage": 1.0,
-      "accuracy": 0.4423,
+      "accepted_accuracy": 0.3333,
+      "accepted_coverage": 0.8654,
+      "accuracy": 0.2885,
       "count": 52,
-      "fallback_rate": 0.0,
-      "macro_f1": 0.2338
+      "fallback_rate": 0.1346,
+      "macro_f1": 0.1491
     }
   },
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.1987,
   "head": "iab_content",
-  "macro_f1": 0.1524,
+  "macro_f1": 0.105,
   "primary_source": "supervised_classifier",
   "suite": "difficulty_benchmark",
   "tier_metrics": {
-    "average_prediction_depth": 2.4103,
+    "average_prediction_depth": 1.7564,
     "error_buckets": {
-      "exact_match": 66,
-      "parent_safe_stop": 1,
-      "right_tier1_wrong_tier2": 42,
-      "wrong_deep_leaf": 8,
-      "wrong_tier1": 39
+      "exact_match": 40,
+      "parent_safe_stop": 11,
+      "right_tier1_wrong_tier2": 58,
+      "wrong_deep_leaf": 1,
+      "wrong_tier1": 46
     },
-    "exact_path_accuracy": 0.4231,
-    "parent_safe_accuracy": 0.5385,
-    "tier1_accuracy": 0.75,
-    "tier2_accuracy": 0.4808,
-    "tier3_accuracy": 0.5093,
-    "tier4_accuracy": 0.4583
+    "exact_path_accuracy": 0.2564,
+    "parent_safe_accuracy": 0.6218,
+    "tier1_accuracy": 0.7051,
+    "tier2_accuracy": 0.3333,
+    "tier3_accuracy": 0.2315,
+    "tier4_accuracy": 0.0
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.4103,
+      "average_prediction_depth": 1.7564,
       "error_buckets": {
-        "exact_match": 59,
-        "parent_safe_stop": 1,
-        "right_tier1_wrong_tier2": 42,
-        "wrong_deep_leaf": 15,
-        "wrong_tier1": 39
+        "exact_match": 40,
+        "parent_safe_stop": 11,
+        "right_tier1_wrong_tier2": 58,
+        "wrong_deep_leaf": 1,
+        "wrong_tier1": 46
       },
-      "exact_path_accuracy": 0.3782,
-      "parent_safe_accuracy": 0.4936,
-      "tier1_accuracy": 0.75,
-      "tier2_accuracy": 0.4808,
-      "tier3_accuracy": 0.4259,
-      "tier4_accuracy": 0.1667
+      "exact_path_accuracy": 0.2564,
+      "parent_safe_accuracy": 0.6218,
+      "tier1_accuracy": 0.7051,
+      "tier2_accuracy": 0.3333,
+      "tier3_accuracy": 0.2315,
+      "tier4_accuracy": 0.0
     },
     "combined_path": {
-      "average_prediction_depth": 2.4103,
+      "average_prediction_depth": 1.7564,
       "error_buckets": {
-        "exact_match": 59,
-        "parent_safe_stop": 1,
-        "right_tier1_wrong_tier2": 42,
-        "wrong_deep_leaf": 15,
-        "wrong_tier1": 39
+        "exact_match": 40,
+        "parent_safe_stop": 11,
+        "right_tier1_wrong_tier2": 58,
+        "wrong_deep_leaf": 1,
+        "wrong_tier1": 46
       },
-      "exact_path_accuracy": 0.3782,
-      "fallback_overuse_count": 15,
-      "fallback_rate": 0.0962,
-      "parent_safe_accuracy": 0.4936,
-      "tier1_accuracy": 0.75,
-      "tier2_accuracy": 0.4808,
-      "tier3_accuracy": 0.4259,
-      "tier4_accuracy": 0.1667
+      "exact_path_accuracy": 0.2564,
+      "fallback_overuse_count": 13,
+      "fallback_rate": 0.0833,
+      "parent_safe_accuracy": 0.6218,
+      "tier1_accuracy": 0.7051,
+      "tier2_accuracy": 0.3333,
+      "tier3_accuracy": 0.2315,
+      "tier4_accuracy": 0.0
     },
     "disagreements": {
       "classifier_vs_combined": 0
diff --git a/artifacts/evaluation/latest/iab_content_extended_cases_report.json b/artifacts/evaluation/latest/iab_content_extended_cases_report.json
index e1ac1e327074bcccf079f5bb70a957a2aa80065f..171c7f93e8f9ef6932c330103c5aebd6251a691b 100644
--- a/artifacts/evaluation/latest/iab_content_extended_cases_report.json
+++ b/artifacts/evaluation/latest/iab_content_extended_cases_report.json
@@ -1,58 +1,58 @@
 {
-  "accepted_accuracy": 0.5,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.5,
+  "accepted_accuracy": 0.6,
+  "accepted_coverage": 0.625,
+  "accuracy": 0.375,
   "count": 8,
   "dataset_path": "/content/agentic-intent-classifier/data/iab/extended_cases.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.375,
   "head": "iab_content",
-  "macro_f1": 0.3333,
+  "macro_f1": 0.2308,
   "primary_source": "supervised_classifier",
   "suite": "extended_cases",
   "tier_metrics": {
-    "average_prediction_depth": 2.125,
+    "average_prediction_depth": 1.75,
     "error_buckets": {
-      "exact_match": 4,
-      "right_tier1_wrong_tier2": 2,
+      "exact_match": 3,
+      "right_tier1_wrong_tier2": 1,
       "wrong_deep_leaf": 1,
-      "wrong_tier1": 1
+      "wrong_tier1": 3
     },
-    "exact_path_accuracy": 0.5,
-    "parent_safe_accuracy": 0.5,
-    "tier1_accuracy": 0.875,
+    "exact_path_accuracy": 0.375,
+    "parent_safe_accuracy": 0.375,
+    "tier1_accuracy": 0.625,
     "tier2_accuracy": 0.5714,
     "tier3_accuracy": 0.0,
     "tier4_accuracy": 0.0
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.125,
+      "average_prediction_depth": 1.75,
       "error_buckets": {
-        "exact_match": 4,
-        "right_tier1_wrong_tier2": 2,
+        "exact_match": 3,
+        "right_tier1_wrong_tier2": 1,
         "wrong_deep_leaf": 1,
-        "wrong_tier1": 1
+        "wrong_tier1": 3
       },
-      "exact_path_accuracy": 0.5,
-      "parent_safe_accuracy": 0.5,
-      "tier1_accuracy": 0.875,
+      "exact_path_accuracy": 0.375,
+      "parent_safe_accuracy": 0.375,
+      "tier1_accuracy": 0.625,
       "tier2_accuracy": 0.5714,
       "tier3_accuracy": 0.0,
       "tier4_accuracy": 0.0
     },
     "combined_path": {
-      "average_prediction_depth": 2.125,
+      "average_prediction_depth": 1.75,
       "error_buckets": {
-        "exact_match": 4,
-        "right_tier1_wrong_tier2": 2,
+        "exact_match": 3,
+        "right_tier1_wrong_tier2": 1,
         "wrong_deep_leaf": 1,
-        "wrong_tier1": 1
+        "wrong_tier1": 3
       },
-      "exact_path_accuracy": 0.5,
+      "exact_path_accuracy": 0.375,
       "fallback_overuse_count": 2,
       "fallback_rate": 0.25,
-      "parent_safe_accuracy": 0.5,
-      "tier1_accuracy": 0.875,
+      "parent_safe_accuracy": 0.375,
+      "tier1_accuracy": 0.625,
       "tier2_accuracy": 0.5714,
       "tier3_accuracy": 0.0,
       "tier4_accuracy": 0.0
diff --git a/artifacts/evaluation/latest/iab_content_hard_cases_report.json b/artifacts/evaluation/latest/iab_content_hard_cases_report.json
index fa18dce2744f7b21ccc3bf1bede556a9d8f93a8d..7a85669e782b4334395fda563e76d7fd4453360a 100644
--- a/artifacts/evaluation/latest/iab_content_hard_cases_report.json
+++ b/artifacts/evaluation/latest/iab_content_hard_cases_report.json
@@ -1,16 +1,16 @@
 {
-  "accepted_accuracy": 0.4286,
-  "accepted_coverage": 0.875,
+  "accepted_accuracy": 0.6,
+  "accepted_coverage": 0.625,
   "accuracy": 0.375,
   "count": 8,
   "dataset_path": "/content/agentic-intent-classifier/data/iab/hard_cases.jsonl",
-  "fallback_rate": 0.125,
+  "fallback_rate": 0.375,
   "head": "iab_content",
   "macro_f1": 0.2308,
   "primary_source": "supervised_classifier",
   "suite": "hard_cases",
   "tier_metrics": {
-    "average_prediction_depth": 2.25,
+    "average_prediction_depth": 1.75,
     "error_buckets": {
       "exact_match": 3,
       "right_tier1_wrong_tier2": 1,
@@ -25,7 +25,7 @@
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.25,
+      "average_prediction_depth": 1.75,
       "error_buckets": {
         "exact_match": 3,
         "right_tier1_wrong_tier2": 1,
@@ -39,7 +39,7 @@
       "tier4_accuracy": 0.0
     },
     "combined_path": {
-      "average_prediction_depth": 2.25,
+      "average_prediction_depth": 1.75,
       "error_buckets": {
         "exact_match": 3,
         "right_tier1_wrong_tier2": 1,
diff --git a/artifacts/evaluation/latest/iab_content_test_report.json b/artifacts/evaluation/latest/iab_content_test_report.json
index 27c4e2a3dadfb29307c90dd87a8b951326f852c6..5571768b6fdfe56eef349f0fd54706d3fbb81041 100644
--- a/artifacts/evaluation/latest/iab_content_test_report.json
+++ b/artifacts/evaluation/latest/iab_content_test_report.json
@@ -1,46 +1,46 @@
 {
-  "accepted_accuracy": 0.943,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.943,
+  "accepted_accuracy": 0.9278,
+  "accepted_coverage": 0.996,
+  "accuracy": 0.9247,
   "count": 3282,
   "dataset_path": "/content/agentic-intent-classifier/data/iab/test.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.004,
   "head": "iab_content",
-  "macro_f1": 0.911,
+  "macro_f1": 0.8814,
   "primary_source": "supervised_classifier",
   "suite": "test",
   "tier_metrics": {
-    "average_prediction_depth": 2.213,
+    "average_prediction_depth": 2.1706,
     "error_buckets": {
-      "exact_match": 3095,
-      "parent_safe_stop": 45,
-      "right_tier1_wrong_tier2": 41,
-      "wrong_deep_leaf": 72,
-      "wrong_tier1": 29
+      "exact_match": 3035,
+      "parent_safe_stop": 87,
+      "right_tier1_wrong_tier2": 56,
+      "wrong_deep_leaf": 69,
+      "wrong_tier1": 35
     },
-    "exact_path_accuracy": 0.943,
-    "parent_safe_accuracy": 0.958,
-    "tier1_accuracy": 0.9912,
-    "tier2_accuracy": 0.9776,
-    "tier3_accuracy": 0.9078,
-    "tier4_accuracy": 0.7
+    "exact_path_accuracy": 0.9247,
+    "parent_safe_accuracy": 0.961,
+    "tier1_accuracy": 0.9893,
+    "tier2_accuracy": 0.9707,
+    "tier3_accuracy": 0.8487,
+    "tier4_accuracy": 0.5714
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.213,
+      "average_prediction_depth": 2.1706,
       "error_buckets": {
-        "exact_match": 3052,
-        "parent_safe_stop": 44,
-        "right_tier1_wrong_tier2": 53,
-        "wrong_deep_leaf": 104,
-        "wrong_tier1": 29
+        "exact_match": 3004,
+        "parent_safe_stop": 84,
+        "right_tier1_wrong_tier2": 68,
+        "wrong_deep_leaf": 91,
+        "wrong_tier1": 35
       },
-      "exact_path_accuracy": 0.9299,
-      "parent_safe_accuracy": 0.9445,
-      "tier1_accuracy": 0.9912,
-      "tier2_accuracy": 0.9734,
-      "tier3_accuracy": 0.8725,
-      "tier4_accuracy": 0.5
+      "exact_path_accuracy": 0.9153,
+      "parent_safe_accuracy": 0.9506,
+      "tier1_accuracy": 0.9893,
+      "tier2_accuracy": 0.9665,
+      "tier3_accuracy": 0.8259,
+      "tier4_accuracy": 0.4429
     },
     "combined_path": {
       "count": 3282,
diff --git a/artifacts/evaluation/latest/iab_content_train_report.json b/artifacts/evaluation/latest/iab_content_train_report.json
index c71b063661998121cf55f61491790f628b8be659..ddfc220c4eb4ef19336c5788c65a8d5fd66d7e40 100644
--- a/artifacts/evaluation/latest/iab_content_train_report.json
+++ b/artifacts/evaluation/latest/iab_content_train_report.json
@@ -1,46 +1,46 @@
 {
-  "accepted_accuracy": 0.9459,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.9459,
+  "accepted_accuracy": 0.9314,
+  "accepted_coverage": 0.9972,
+  "accuracy": 0.9295,
   "count": 13211,
   "dataset_path": "/content/agentic-intent-classifier/data/iab/train.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.0028,
   "head": "iab_content",
-  "macro_f1": 0.9194,
+  "macro_f1": 0.8927,
   "primary_source": "supervised_classifier",
   "suite": "train",
   "tier_metrics": {
-    "average_prediction_depth": 2.2105,
+    "average_prediction_depth": 2.1683,
     "error_buckets": {
-      "exact_match": 12496,
-      "parent_safe_stop": 162,
-      "right_tier1_wrong_tier2": 144,
-      "wrong_deep_leaf": 284,
-      "wrong_tier1": 125
+      "exact_match": 12280,
+      "parent_safe_stop": 312,
+      "right_tier1_wrong_tier2": 215,
+      "wrong_deep_leaf": 288,
+      "wrong_tier1": 116
     },
-    "exact_path_accuracy": 0.9459,
-    "parent_safe_accuracy": 0.9585,
-    "tier1_accuracy": 0.9905,
-    "tier2_accuracy": 0.9805,
-    "tier3_accuracy": 0.9135,
-    "tier4_accuracy": 0.7268
+    "exact_path_accuracy": 0.9295,
+    "parent_safe_accuracy": 0.9618,
+    "tier1_accuracy": 0.9912,
+    "tier2_accuracy": 0.9737,
+    "tier3_accuracy": 0.8557,
+    "tier4_accuracy": 0.6107
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.2105,
+      "average_prediction_depth": 2.1683,
       "error_buckets": {
-        "exact_match": 12323,
-        "parent_safe_stop": 157,
-        "right_tier1_wrong_tier2": 192,
-        "wrong_deep_leaf": 414,
-        "wrong_tier1": 125
+        "exact_match": 12145,
+        "parent_safe_stop": 300,
+        "right_tier1_wrong_tier2": 263,
+        "wrong_deep_leaf": 387,
+        "wrong_tier1": 116
       },
-      "exact_path_accuracy": 0.9328,
-      "parent_safe_accuracy": 0.945,
-      "tier1_accuracy": 0.9905,
-      "tier2_accuracy": 0.9764,
-      "tier3_accuracy": 0.8777,
-      "tier4_accuracy": 0.525
+      "exact_path_accuracy": 0.9193,
+      "parent_safe_accuracy": 0.9507,
+      "tier1_accuracy": 0.9912,
+      "tier2_accuracy": 0.9695,
+      "tier3_accuracy": 0.8301,
+      "tier4_accuracy": 0.475
     },
     "combined_path": {
       "count": 13211,
diff --git a/artifacts/evaluation/latest/iab_content_val_report.json b/artifacts/evaluation/latest/iab_content_val_report.json
index e6180d848166b6ebaf680a7c1d583dd7004e9aaf..07f7b7c308a572fcacf4bfda0c00cdefd92f7d3a 100644
--- a/artifacts/evaluation/latest/iab_content_val_report.json
+++ b/artifacts/evaluation/latest/iab_content_val_report.json
@@ -1,46 +1,46 @@
 {
-  "accepted_accuracy": 0.9442,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.9442,
+  "accepted_accuracy": 0.9273,
+  "accepted_coverage": 0.9973,
+  "accuracy": 0.9254,
   "count": 3282,
   "dataset_path": "/content/agentic-intent-classifier/data/iab/val.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.0027,
   "head": "iab_content",
-  "macro_f1": 0.9166,
+  "macro_f1": 0.8864,
   "primary_source": "supervised_classifier",
   "suite": "val",
   "tier_metrics": {
-    "average_prediction_depth": 2.2151,
+    "average_prediction_depth": 2.1709,
     "error_buckets": {
-      "exact_match": 3099,
-      "parent_safe_stop": 35,
-      "right_tier1_wrong_tier2": 45,
-      "wrong_deep_leaf": 72,
-      "wrong_tier1": 31
+      "exact_match": 3037,
+      "parent_safe_stop": 80,
+      "right_tier1_wrong_tier2": 55,
+      "wrong_deep_leaf": 74,
+      "wrong_tier1": 36
     },
-    "exact_path_accuracy": 0.9442,
-    "parent_safe_accuracy": 0.9576,
-    "tier1_accuracy": 0.9906,
-    "tier2_accuracy": 0.9769,
-    "tier3_accuracy": 0.9088,
-    "tier4_accuracy": 0.7286
+    "exact_path_accuracy": 0.9254,
+    "parent_safe_accuracy": 0.9613,
+    "tier1_accuracy": 0.989,
+    "tier2_accuracy": 0.9713,
+    "tier3_accuracy": 0.8549,
+    "tier4_accuracy": 0.6071
   },
   "view_metrics": {
     "classifier": {
-      "average_prediction_depth": 2.2151,
+      "average_prediction_depth": 2.1709,
       "error_buckets": {
-        "exact_match": 3056,
-        "parent_safe_stop": 34,
-        "right_tier1_wrong_tier2": 57,
-        "wrong_deep_leaf": 104,
-        "wrong_tier1": 31
+        "exact_match": 3002,
+        "parent_safe_stop": 78,
+        "right_tier1_wrong_tier2": 67,
+        "wrong_deep_leaf": 99,
+        "wrong_tier1": 36
       },
-      "exact_path_accuracy": 0.9311,
-      "parent_safe_accuracy": 0.9442,
-      "tier1_accuracy": 0.9906,
-      "tier2_accuracy": 0.9727,
-      "tier3_accuracy": 0.8736,
-      "tier4_accuracy": 0.5286
+      "exact_path_accuracy": 0.9147,
+      "parent_safe_accuracy": 0.95,
+      "tier1_accuracy": 0.989,
+      "tier2_accuracy": 0.9672,
+      "tier3_accuracy": 0.829,
+      "tier4_accuracy": 0.4643
     },
     "combined_path": {
       "count": 3282,
diff --git a/artifacts/evaluation/latest/iab_cross_vertical_behavior_lock_regression.json b/artifacts/evaluation/latest/iab_cross_vertical_behavior_lock_regression.json
index b50acd3aa09bda431eba62ef3ebf6ccac57ec156..6de4c767990c75ed447c3089aee7236c8fc47b5a 100644
--- a/artifacts/evaluation/latest/iab_cross_vertical_behavior_lock_regression.json
+++ b/artifacts/evaluation/latest/iab_cross_vertical_behavior_lock_regression.json
@@ -1,21 +1,21 @@
 {
   "by_status": {
     "must_fix": {
-      "failed": 90,
-      "passed": 0,
+      "failed": 88,
+      "passed": 2,
       "total": 90
     }
   },
   "cases_path": "/content/agentic-intent-classifier/examples/iab_cross_vertical_behavior_lock_cases.json",
   "count": 90,
-  "failed": 90,
-  "passed": 0,
+  "failed": 88,
+  "passed": 2,
   "results": [
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Personal Finance",
-        "model_output.classification.iab_content.tier2.label": "Insurance"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Travel",
+        "model_output.classification.iab_content.tier2.label": "Travel Type"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -23,31 +23,15 @@
         "model_output.classification.iab_content.tier2.label": "Travel Type"
       },
       "id": "auto-buying-easy",
-      "mismatches": [
-        {
-          "actual": "Personal Finance",
-          "expected": "Travel",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Insurance",
-          "expected": "Travel Type",
-          "path": "model_output.classification.iab_content.tier2.label"
-        }
-      ],
+      "mismatches": [],
       "notes": "Cross-vertical easy IAB mapping case for Automotive > Auto Buying and Selling.",
-      "pass": false,
+      "pass": true,
       "status": "must_fix",
       "text": "Which car should I buy for commuting?"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Automotive",
         "model_output.classification.iab_content.tier2.label": "Auto Body Styles"
       },
@@ -57,23 +41,17 @@
         "model_output.classification.iab_content.tier2.label": "Auto Body Styles"
       },
       "id": "auto-buying-medium",
-      "mismatches": [
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        }
-      ],
+      "mismatches": [],
       "notes": "Cross-vertical medium IAB mapping case for Automotive > Auto Buying and Selling.",
-      "pass": false,
+      "pass": true,
       "status": "must_fix",
       "text": "Best used SUV for a family of four"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Automotive",
-        "model_output.classification.iab_content.tier2.label": "Car Culture"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -83,12 +61,7 @@
       "id": "auto-buying-hard",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Car Culture",
+          "actual": null,
           "expected": "Auto Type",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -102,8 +75,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -117,6 +90,16 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Software and Applications",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical easy IAB mapping case for Business and Finance > Business > Sales.",
@@ -126,9 +109,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Robotics",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -140,12 +123,7 @@
       "id": "sales-crm-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Robotics",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -165,7 +143,7 @@
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Business and Finance",
         "model_output.classification.iab_content.tier2.label": "Business",
-        "model_output.classification.iab_content.tier3.label": "Sales"
+        "model_output.classification.iab_content.tier3.label": "Startups"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -179,6 +157,11 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": "Startups",
+          "expected": "Sales",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical hard IAB mapping case for Business and Finance > Business > Sales.",
@@ -188,9 +171,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Careers",
-        "model_output.classification.iab_content.tier2.label": "Job Search",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -202,17 +185,7 @@
       "id": "marketing-tools-easy",
       "mismatches": [
         {
-          "actual": "Careers",
-          "expected": "Technology & Computing",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Job Search",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -229,9 +202,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-        "model_output.classification.iab_content.tier2.label": "Terrorism"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -241,17 +214,12 @@
       "id": "marketing-tools-medium",
       "mismatches": [
         {
-          "actual": "Sensitive Topics",
+          "actual": "Technology & Computing",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Terrorism",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -263,9 +231,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Personal Finance",
-        "model_output.classification.iab_content.tier2.label": "Home Utilities",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -277,17 +245,12 @@
       "id": "marketing-tools-hard",
       "mismatches": [
         {
-          "actual": "Personal Finance",
+          "actual": "Careers",
           "expected": "Technology & Computing",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Home Utilities",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -304,10 +267,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -318,12 +281,17 @@
       "id": "business-it-easy",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Careers",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Information and Network Security",
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
           "expected": "Internet",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -335,9 +303,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Careers",
-        "model_output.classification.iab_content.tier2.label": "Job Search"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Personal Finance",
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -347,9 +315,14 @@
       "id": "business-it-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Personal Finance",
+          "expected": "Careers",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": null,
+          "expected": "Job Search",
+          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Business and Finance > Business > Business I.T..",
@@ -415,8 +388,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Attractions",
-        "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+        "model_output.classification.iab_content.tier1.label": "Food & Drink",
+        "model_output.classification.iab_content.tier2.label": "Dining Out"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -425,20 +398,10 @@
       },
       "id": "dining-out-medium",
       "mismatches": [
-        {
-          "actual": "Attractions",
-          "expected": "Food & Drink",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
         {
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Bars & Restaurants",
-          "expected": "Dining Out",
-          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Food & Drink > Dining Out.",
@@ -449,8 +412,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Attractions",
-        "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+        "model_output.classification.iab_content.tier1.label": "Food & Drink",
+        "model_output.classification.iab_content.tier2.label": "Dining Out"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -459,20 +422,10 @@
       },
       "id": "dining-out-hard",
       "mismatches": [
-        {
-          "actual": "Attractions",
-          "expected": "Food & Drink",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
         {
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Bars & Restaurants",
-          "expected": "Dining Out",
-          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical hard IAB mapping case for Food & Drink > Dining Out.",
@@ -549,7 +502,7 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-        "model_output.classification.iab_content.tier1.label": "Science"
+        "model_output.classification.iab_content.tier1.label": "Real Estate"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -558,7 +511,7 @@
       "id": "artificial-intelligence-easy",
       "mismatches": [
         {
-          "actual": "Science",
+          "actual": "Real Estate",
           "expected": "Technology & Computing",
           "path": "model_output.classification.iab_content.tier1.label"
         }
@@ -570,9 +523,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Education",
-        "model_output.classification.iab_content.tier2.label": "Language Learning"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -582,9 +535,9 @@
       "id": "artificial-intelligence-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Language Learning",
+          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Artificial Intelligence.",
@@ -619,8 +572,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing"
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": "Job Search"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -630,7 +583,7 @@
       "id": "software-apps-easy",
       "mismatches": [
         {
-          "actual": "Technology & Computing",
+          "actual": "Careers",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
@@ -640,7 +593,7 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Computing",
+          "actual": "Job Search",
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -654,8 +607,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null,
         "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
@@ -673,7 +626,12 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Software and Applications",
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
           "expected": "Internet",
           "path": "model_output.classification.iab_content.tier3.label"
         },
@@ -691,8 +649,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": "Job Search",
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -703,13 +661,18 @@
       },
       "id": "software-apps-hard",
       "mismatches": [
+        {
+          "actual": "Careers",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
         {
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Virtual Reality",
+          "actual": "Job Search",
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -774,10 +737,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Information and Network Security",
+        "model_output.classification.iab_content.tier3.label": null,
         "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
@@ -790,12 +753,7 @@
       "id": "communication-software-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Information and Network Security",
+          "actual": null,
           "expected": "Software and Applications",
           "path": "model_output.classification.iab_content.tier3.label"
         },
@@ -812,9 +770,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null,
         "model_output.classification.iab_content.tier4.label": null
       },
@@ -828,12 +786,12 @@
       "id": "communication-software-hard",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Careers",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Virtual Reality",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -858,8 +816,8 @@
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Internet",
-        "model_output.classification.iab_content.tier4.label": "Web Hosting"
+        "model_output.classification.iab_content.tier3.label": "Data Storage and Warehousing",
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -874,6 +832,16 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": "Data Storage and Warehousing",
+          "expected": "Internet",
+          "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": null,
+          "expected": "Web Hosting",
+          "path": "model_output.classification.iab_content.tier4.label"
         }
       ],
       "notes": "Cross-vertical easy IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
@@ -883,11 +851,11 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Internet",
-        "model_output.classification.iab_content.tier4.label": "Web Hosting"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null,
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -899,9 +867,19 @@
       "id": "web-hosting-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Internet",
+          "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": null,
+          "expected": "Web Hosting",
+          "path": "model_output.classification.iab_content.tier4.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
@@ -911,11 +889,11 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Internet",
-        "model_output.classification.iab_content.tier4.label": "Web Hosting"
+        "model_output.classification.iab_content.tier3.label": null,
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -927,9 +905,14 @@
       "id": "web-hosting-hard",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Internet",
+          "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": null,
+          "expected": "Web Hosting",
+          "path": "model_output.classification.iab_content.tier4.label"
         }
       ],
       "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
@@ -993,8 +976,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Laptops"
+        "model_output.classification.iab_content.tier2.label": "Consumer Electronics",
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1010,12 +993,7 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Computing",
-          "expected": "Consumer Electronics",
-          "path": "model_output.classification.iab_content.tier2.label"
-        },
-        {
-          "actual": "Laptops",
+          "actual": null,
           "expected": "Smartphones",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1027,11 +1005,11 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications",
-        "model_output.classification.iab_content.tier4.label": "Computer Animation"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null,
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1043,12 +1021,17 @@
       "id": "desktops-easy",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Software and Applications",
+          "path": "model_output.classification.iab_content.tier3.label"
         },
         {
-          "actual": "Computer Animation",
+          "actual": null,
           "expected": "Photo Editing Software",
           "path": "model_output.classification.iab_content.tier4.label"
         }
@@ -1060,10 +1043,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Desktops"
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1074,9 +1057,9 @@
       "id": "desktops-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Desktops",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Desktops.",
@@ -1086,10 +1069,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Desktops"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1100,9 +1083,14 @@
       "id": "desktops-hard",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Desktops",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Desktops.",
@@ -1188,10 +1176,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Men's Fashion",
-        "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Shopping",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1202,17 +1190,17 @@
       "id": "style-fashion-parent-easy",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Shopping",
+          "expected": "Style & Fashion",
+          "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Men's Fashion",
+          "actual": null,
           "expected": "Women's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Men's Shoes and Footwear",
+          "actual": null,
           "expected": "Women's Shoes and Footwear",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1286,9 +1274,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sports",
-        "model_output.classification.iab_content.tier2.label": "Bodybuilding",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1300,17 +1288,7 @@
       "id": "womens-shoes-easy",
       "mismatches": [
         {
-          "actual": "Sports",
-          "expected": "Style & Fashion",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Bodybuilding",
+          "actual": null,
           "expected": "Women's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -1327,9 +1305,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sports",
-        "model_output.classification.iab_content.tier2.label": "Walking",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1341,17 +1319,7 @@
       "id": "womens-shoes-medium",
       "mismatches": [
         {
-          "actual": "Sports",
-          "expected": "Style & Fashion",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Walking",
+          "actual": null,
           "expected": "Women's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -1396,8 +1364,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-        "model_output.classification.iab_content.tier3.label": null
+        "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+        "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1413,12 +1381,12 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Children's Clothing",
+          "actual": "Women's Fashion",
           "expected": "Men's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": null,
+          "actual": "Women's Shoes and Footwear",
           "expected": "Men's Clothing",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1432,8 +1400,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Men's Fashion",
-        "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
+        "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+        "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1449,7 +1417,12 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Men's Shoes and Footwear",
+          "actual": "Women's Fashion",
+          "expected": "Men's Fashion",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": "Women's Shoes and Footwear",
           "expected": "Men's Clothing",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1463,8 +1436,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-        "model_output.classification.iab_content.tier3.label": null
+        "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+        "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1480,12 +1453,12 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Children's Clothing",
+          "actual": "Women's Fashion",
           "expected": "Men's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": null,
+          "actual": "Women's Shoes and Footwear",
           "expected": "Men's Shoes and Footwear",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1531,10 +1504,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Travel",
-        "model_output.classification.iab_content.tier2.label": "Travel Type",
-        "model_output.classification.iab_content.tier3.label": "Hotels and Motels"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1545,9 +1518,14 @@
       "id": "hotels-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": null,
+          "expected": "Travel Type",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Hotels and Motels",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Travel > Travel Type > Hotels and Motels.",
@@ -1559,7 +1537,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Travel",
-        "model_output.classification.iab_content.tier2.label": "Travel Type"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1572,6 +1550,11 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Travel Type",
+          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical hard IAB mapping case for Travel > Travel Type > Hotels and Motels.",
@@ -1672,8 +1655,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Healthy Living",
-        "model_output.classification.iab_content.tier2.label": "Fitness and Exercise",
-        "model_output.classification.iab_content.tier3.label": "Running and Jogging"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1694,12 +1677,12 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Fitness and Exercise",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Running and Jogging",
+          "actual": null,
           "expected": "Green Solutions",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1713,7 +1696,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Sports",
-        "model_output.classification.iab_content.tier2.label": "Walking",
+        "model_output.classification.iab_content.tier2.label": "Bodybuilding",
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1735,7 +1718,7 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Walking",
+          "actual": "Bodybuilding",
           "expected": "Fitness and Exercise",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -1897,9 +1880,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Travel",
-        "model_output.classification.iab_content.tier2.label": "Travel Type"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1914,12 +1897,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Travel Type",
+          "actual": null,
           "expected": "Fiction",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -1932,7 +1910,7 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Books and Literature"
+        "model_output.classification.iab_content.tier1.label": "Genres"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1940,6 +1918,11 @@
       },
       "id": "fiction-hard",
       "mismatches": [
+        {
+          "actual": "Genres",
+          "expected": "Books and Literature",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
         {
           "actual": "exact",
           "expected": "nearest_equivalent",
@@ -1955,7 +1938,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Home & Garden",
-        "model_output.classification.iab_content.tier2.label": "Interior Decorating"
+        "model_output.classification.iab_content.tier2.label": "Remodeling & Construction"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1968,11 +1951,6 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Interior Decorating",
-          "expected": "Remodeling & Construction",
-          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical easy IAB mapping case for Home & Garden > Home Improvement.",
@@ -1983,9 +1961,9 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Home & Garden",
-        "model_output.classification.iab_content.tier2.label": "Interior Decorating",
-        "model_output.classification.iab_content.tier3.label": null
+        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+        "model_output.classification.iab_content.tier2.label": "Personal Care",
+        "model_output.classification.iab_content.tier3.label": "Bath and Shower"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1995,25 +1973,10 @@
       },
       "id": "home-improvement-medium",
       "mismatches": [
-        {
-          "actual": "Home & Garden",
-          "expected": "Style & Fashion",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
         {
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Interior Decorating",
-          "expected": "Personal Care",
-          "path": "model_output.classification.iab_content.tier2.label"
-        },
-        {
-          "actual": null,
-          "expected": "Bath and Shower",
-          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical medium IAB mapping case for Home & Garden > Home Improvement.",
@@ -2057,9 +2020,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Augmented Reality"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2069,17 +2032,12 @@
       "id": "online-education-easy",
       "mismatches": [
         {
-          "actual": "Technology & Computing",
+          "actual": "Careers",
           "expected": "Education",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Augmented Reality",
+          "actual": null,
           "expected": "Language Learning",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -2218,10 +2176,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Medical Health",
-        "model_output.classification.iab_content.tier2.label": "Diseases and Conditions",
-        "model_output.classification.iab_content.tier3.label": "Allergies"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Food & Drink",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2232,9 +2190,19 @@
       "id": "medical-health-easy",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Food & Drink",
+          "expected": "Medical Health",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": null,
+          "expected": "Diseases and Conditions",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Allergies",
+          "path": "model_output.classification.iab_content.tier3.label"
         }
       ],
       "notes": "Cross-vertical easy IAB mapping case for Medical Health.",
@@ -2244,10 +2212,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Medical Health",
         "model_output.classification.iab_content.tier2.label": "Diseases and Conditions",
-        "model_output.classification.iab_content.tier3.label": "Bone and Joint Conditions",
+        "model_output.classification.iab_content.tier3.label": null,
         "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
@@ -2260,12 +2228,7 @@
       "id": "medical-health-medium",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Bone and Joint Conditions",
+          "actual": null,
           "expected": "Injuries",
           "path": "model_output.classification.iab_content.tier3.label"
         },
@@ -2284,7 +2247,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Medical Health",
-        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier2.label": "Surgery",
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -2306,7 +2269,7 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": null,
+          "actual": "Surgery",
           "expected": "Wellness",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -2407,9 +2370,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Holidays",
-        "model_output.classification.iab_content.tier2.label": "National & Civic Holidays"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2424,12 +2387,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "National & Civic Holidays",
+          "actual": null,
           "expected": "Food Movements",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -2530,9 +2488,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Genres",
-        "model_output.classification.iab_content.tier2.label": "Family/Children"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Shopping",
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2542,17 +2500,12 @@
       "id": "parenting-medium",
       "mismatches": [
         {
-          "actual": "Genres",
+          "actual": "Shopping",
           "expected": "Family and Relationships",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Family/Children",
+          "actual": null,
           "expected": "Parenting",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -2567,7 +2520,7 @@
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Family and Relationships",
         "model_output.classification.iab_content.tier2.label": "Parenting",
-        "model_output.classification.iab_content.tier3.label": null
+        "model_output.classification.iab_content.tier3.label": "Parenting Babies and Toddlers"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2583,7 +2536,7 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": null,
+          "actual": "Parenting Babies and Toddlers",
           "expected": "Special Needs Kids",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -2665,7 +2618,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Entertainment",
-        "model_output.classification.iab_content.tier2.label": "Movies"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2678,6 +2631,11 @@
           "actual": "exact",
           "expected": "nearest_equivalent",
           "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Movies",
+          "path": "model_output.classification.iab_content.tier2.label"
         }
       ],
       "notes": "Cross-vertical easy IAB mapping case for Entertainment > Movies.",
@@ -2688,8 +2646,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Entertainment",
-        "model_output.classification.iab_content.tier2.label": "Movies",
+        "model_output.classification.iab_content.tier1.label": "Genres",
+        "model_output.classification.iab_content.tier2.label": "Horror",
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -2701,7 +2659,7 @@
       "id": "movies-medium",
       "mismatches": [
         {
-          "actual": "Entertainment",
+          "actual": "Genres",
           "expected": "Video Gaming",
           "path": "model_output.classification.iab_content.tier1.label"
         },
@@ -2711,7 +2669,7 @@
           "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Movies",
+          "actual": "Horror",
           "expected": "Video Game Genres",
           "path": "model_output.classification.iab_content.tier2.label"
         },
diff --git a/artifacts/evaluation/latest/iab_cross_vertical_quality_target_eval.json b/artifacts/evaluation/latest/iab_cross_vertical_quality_target_eval.json
index b269b2367fdf2ce6764627276bb75045080a0aeb..675bcce768bf0edd39d0d34e05d363b99d3a9492 100644
--- a/artifacts/evaluation/latest/iab_cross_vertical_quality_target_eval.json
+++ b/artifacts/evaluation/latest/iab_cross_vertical_quality_target_eval.json
@@ -1,21 +1,21 @@
 {
   "by_status": {
     "must_fix": {
-      "failed": 49,
-      "passed": 41,
+      "failed": 64,
+      "passed": 26,
       "total": 90
     }
   },
   "cases_path": "/content/agentic-intent-classifier/examples/iab_cross_vertical_mapping_cases.json",
   "count": 90,
-  "failed": 49,
-  "passed": 41,
+  "failed": 64,
+  "passed": 26,
   "results": [
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Personal Finance",
-        "model_output.classification.iab_content.tier2.label": "Insurance"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Travel",
+        "model_output.classification.iab_content.tier2.label": "Travel Type"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -25,17 +25,12 @@
       "id": "auto-buying-easy",
       "mismatches": [
         {
-          "actual": "Personal Finance",
+          "actual": "Travel",
           "expected": "Automotive",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Insurance",
+          "actual": "Travel Type",
           "expected": "Auto Buying and Selling",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -47,7 +42,7 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Automotive",
         "model_output.classification.iab_content.tier2.label": "Auto Body Styles"
       },
@@ -58,11 +53,6 @@
       },
       "id": "auto-buying-medium",
       "mismatches": [
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
         {
           "actual": "Auto Body Styles",
           "expected": "Auto Buying and Selling",
@@ -76,9 +66,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Automotive",
-        "model_output.classification.iab_content.tier2.label": "Car Culture"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -88,12 +78,7 @@
       "id": "auto-buying-hard",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Car Culture",
+          "actual": null,
           "expected": "Auto Buying and Selling",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -107,8 +92,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -124,12 +109,12 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Computing",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Software and Applications",
+          "actual": null,
           "expected": "Sales",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -141,9 +126,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Robotics",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -160,7 +145,12 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Robotics",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -180,7 +170,7 @@
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Business and Finance",
         "model_output.classification.iab_content.tier2.label": "Business",
-        "model_output.classification.iab_content.tier3.label": "Sales"
+        "model_output.classification.iab_content.tier3.label": "Startups"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -189,17 +179,23 @@
         "model_output.classification.iab_content.tier3.label": "Sales"
       },
       "id": "sales-crm-hard",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Startups",
+          "expected": "Sales",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical hard IAB mapping case for Business and Finance > Business > Sales.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Need software to manage leads and pipeline for a startup sales team"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Careers",
-        "model_output.classification.iab_content.tier2.label": "Job Search",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -211,12 +207,17 @@
       "id": "marketing-tools-easy",
       "mismatches": [
         {
-          "actual": "Careers",
+          "actual": "Technology & Computing",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Job Search",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -233,9 +234,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-        "model_output.classification.iab_content.tier2.label": "Terrorism",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -247,12 +248,17 @@
       "id": "marketing-tools-medium",
       "mismatches": [
         {
-          "actual": "Sensitive Topics",
+          "actual": "Technology & Computing",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Terrorism",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -269,9 +275,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Personal Finance",
-        "model_output.classification.iab_content.tier2.label": "Home Utilities",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -283,12 +289,17 @@
       "id": "marketing-tools-hard",
       "mismatches": [
         {
-          "actual": "Personal Finance",
+          "actual": "Careers",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Home Utilities",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -305,10 +316,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -319,17 +330,22 @@
       "id": "business-it-easy",
       "mismatches": [
         {
-          "actual": "Technology & Computing",
+          "actual": "Careers",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Computing",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Information and Network Security",
+          "actual": null,
           "expected": "Business I.T.",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -341,9 +357,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Careers",
-        "model_output.classification.iab_content.tier2.label": "Job Search",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Personal Finance",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -355,12 +371,17 @@
       "id": "business-it-medium",
       "mismatches": [
         {
-          "actual": "Careers",
+          "actual": "Personal Finance",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Job Search",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -432,8 +453,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Attractions",
-        "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+        "model_output.classification.iab_content.tier1.label": "Food & Drink",
+        "model_output.classification.iab_content.tier2.label": "Dining Out"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -441,28 +462,17 @@
         "model_output.classification.iab_content.tier2.label": "Dining Out"
       },
       "id": "dining-out-medium",
-      "mismatches": [
-        {
-          "actual": "Attractions",
-          "expected": "Food & Drink",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
-        {
-          "actual": "Bars & Restaurants",
-          "expected": "Dining Out",
-          "path": "model_output.classification.iab_content.tier2.label"
-        }
-      ],
+      "mismatches": [],
       "notes": "Cross-vertical medium IAB mapping case for Food & Drink > Dining Out.",
-      "pass": false,
+      "pass": true,
       "status": "must_fix",
       "text": "Good restaurants for a client dinner downtown"
     },
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Attractions",
-        "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+        "model_output.classification.iab_content.tier1.label": "Food & Drink",
+        "model_output.classification.iab_content.tier2.label": "Dining Out"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -470,20 +480,9 @@
         "model_output.classification.iab_content.tier2.label": "Dining Out"
       },
       "id": "dining-out-hard",
-      "mismatches": [
-        {
-          "actual": "Attractions",
-          "expected": "Food & Drink",
-          "path": "model_output.classification.iab_content.tier1.label"
-        },
-        {
-          "actual": "Bars & Restaurants",
-          "expected": "Dining Out",
-          "path": "model_output.classification.iab_content.tier2.label"
-        }
-      ],
+      "mismatches": [],
       "notes": "Cross-vertical hard IAB mapping case for Food & Drink > Dining Out.",
-      "pass": false,
+      "pass": true,
       "status": "must_fix",
       "text": "Need a place to eat tonight where I can make a reservation online"
     },
@@ -550,7 +549,7 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-        "model_output.classification.iab_content.tier1.label": "Science",
+        "model_output.classification.iab_content.tier1.label": "Real Estate",
         "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
@@ -561,7 +560,7 @@
       "id": "artificial-intelligence-easy",
       "mismatches": [
         {
-          "actual": "Science",
+          "actual": "Real Estate",
           "expected": "Technology & Computing",
           "path": "model_output.classification.iab_content.tier1.label"
         },
@@ -583,9 +582,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Education",
-        "model_output.classification.iab_content.tier2.label": "Language Learning"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -600,7 +599,12 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Language Learning",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Artificial Intelligence",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -642,9 +646,9 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": "Job Search",
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -653,9 +657,25 @@
         "model_output.classification.iab_content.tier3.label": "Software and Applications"
       },
       "id": "software-apps-easy",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Careers",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "Job Search",
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Software and Applications",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical easy IAB mapping case for Technology & Computing > Computing > Computer Software and Applications.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Best workflow software for a small operations team"
     },
@@ -663,8 +683,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -673,17 +693,28 @@
         "model_output.classification.iab_content.tier3.label": "Software and Applications"
       },
       "id": "software-apps-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Software and Applications",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Computer Software and Applications.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Need project management software for a distributed team"
     },
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": "Job Search",
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -695,7 +726,12 @@
       "id": "software-apps-hard",
       "mismatches": [
         {
-          "actual": "Virtual Reality",
+          "actual": "Careers",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "Job Search",
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -755,10 +791,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Information and Network Security",
+        "model_output.classification.iab_content.tier3.label": null,
         "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
@@ -771,7 +807,12 @@
       "id": "communication-software-medium",
       "mismatches": [
         {
-          "actual": "Information and Network Security",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Software and Applications",
           "path": "model_output.classification.iab_content.tier3.label"
         },
@@ -788,9 +829,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null,
         "model_output.classification.iab_content.tier4.label": null
       },
@@ -804,7 +845,17 @@
       "id": "communication-software-hard",
       "mismatches": [
         {
-          "actual": "Virtual Reality",
+          "actual": "Careers",
+          "expected": "Technology & Computing",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -829,8 +880,8 @@
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Internet",
-        "model_output.classification.iab_content.tier4.label": "Web Hosting"
+        "model_output.classification.iab_content.tier3.label": "Data Storage and Warehousing",
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -840,19 +891,30 @@
         "model_output.classification.iab_content.tier4.label": "Web Hosting"
       },
       "id": "web-hosting-easy",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Data Storage and Warehousing",
+          "expected": "Internet",
+          "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": null,
+          "expected": "Web Hosting",
+          "path": "model_output.classification.iab_content.tier4.label"
+        }
+      ],
       "notes": "Cross-vertical easy IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Vercel vs Netlify for website hosting"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Internet",
-        "model_output.classification.iab_content.tier4.label": "Web Hosting"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null,
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -862,19 +924,40 @@
         "model_output.classification.iab_content.tier4.label": "Web Hosting"
       },
       "id": "web-hosting-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Internet",
+          "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": null,
+          "expected": "Web Hosting",
+          "path": "model_output.classification.iab_content.tier4.label"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Best hosting platform for a startup website"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Internet",
-        "model_output.classification.iab_content.tier4.label": "Web Hosting"
+        "model_output.classification.iab_content.tier3.label": null,
+        "model_output.classification.iab_content.tier4.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -884,9 +967,25 @@
         "model_output.classification.iab_content.tier4.label": "Web Hosting"
       },
       "id": "web-hosting-hard",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Internet",
+          "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": null,
+          "expected": "Web Hosting",
+          "path": "model_output.classification.iab_content.tier4.label"
+        }
+      ],
       "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Need a managed hosting provider to deploy and run our marketing site"
     },
@@ -934,8 +1033,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Laptops"
+        "model_output.classification.iab_content.tier2.label": "Consumer Electronics",
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -944,18 +1043,29 @@
         "model_output.classification.iab_content.tier3.label": "Laptops"
       },
       "id": "laptops-hard",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Consumer Electronics",
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Laptops",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Laptops.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Need a portable computer with good battery life for everyday work"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -966,7 +1076,17 @@
       "id": "desktops-easy",
       "mismatches": [
         {
-          "actual": "Software and Applications",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
           "expected": "Desktops",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -978,10 +1098,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
         "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Desktops"
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -990,18 +1110,29 @@
         "model_output.classification.iab_content.tier3.label": "Desktops"
       },
       "id": "desktops-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Desktops",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Desktops.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Which desktop computer should I buy for a home office?"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Desktops"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1010,9 +1141,25 @@
         "model_output.classification.iab_content.tier3.label": "Desktops"
       },
       "id": "desktops-hard",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Computing",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Desktops",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Desktops.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Need a desktop PC with strong performance for creative work"
     },
@@ -1078,8 +1225,8 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Style & Fashion"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Shopping"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1088,9 +1235,9 @@
       "id": "style-fashion-parent-easy",
       "mismatches": [
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
+          "actual": "Shopping",
+          "expected": "Style & Fashion",
+          "path": "model_output.classification.iab_content.tier1.label"
         }
       ],
       "notes": "Cross-vertical easy IAB mapping case for Style & Fashion.",
@@ -1144,9 +1291,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sports",
-        "model_output.classification.iab_content.tier2.label": "Bodybuilding",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1158,12 +1305,12 @@
       "id": "womens-shoes-easy",
       "mismatches": [
         {
-          "actual": "Sports",
-          "expected": "Style & Fashion",
-          "path": "model_output.classification.iab_content.tier1.label"
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Bodybuilding",
+          "actual": null,
           "expected": "Women's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -1180,9 +1327,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sports",
-        "model_output.classification.iab_content.tier2.label": "Walking",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1194,12 +1341,12 @@
       "id": "womens-shoes-medium",
       "mismatches": [
         {
-          "actual": "Sports",
-          "expected": "Style & Fashion",
-          "path": "model_output.classification.iab_content.tier1.label"
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
         },
         {
-          "actual": "Walking",
+          "actual": null,
           "expected": "Women's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -1238,8 +1385,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-        "model_output.classification.iab_content.tier3.label": null
+        "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+        "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1250,12 +1397,12 @@
       "id": "mens-shoes-easy",
       "mismatches": [
         {
-          "actual": "Children's Clothing",
+          "actual": "Women's Fashion",
           "expected": "Men's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": null,
+          "actual": "Women's Shoes and Footwear",
           "expected": "Men's Shoes and Footwear",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1269,8 +1416,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Men's Fashion",
-        "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
+        "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+        "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1279,9 +1426,20 @@
         "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
       },
       "id": "mens-shoes-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Women's Fashion",
+          "expected": "Men's Fashion",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": "Women's Shoes and Footwear",
+          "expected": "Men's Shoes and Footwear",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Style & Fashion > Men's Fashion > Men's Shoes and Footwear.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Good men's dress shoes for office use"
     },
@@ -1289,8 +1447,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-        "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-        "model_output.classification.iab_content.tier3.label": null
+        "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+        "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1301,12 +1459,12 @@
       "id": "mens-shoes-hard",
       "mismatches": [
         {
-          "actual": "Children's Clothing",
+          "actual": "Women's Fashion",
           "expected": "Men's Fashion",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": null,
+          "actual": "Women's Shoes and Footwear",
           "expected": "Men's Shoes and Footwear",
           "path": "model_output.classification.iab_content.tier3.label"
         }
@@ -1338,10 +1496,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Travel",
-        "model_output.classification.iab_content.tier2.label": "Travel Type",
-        "model_output.classification.iab_content.tier3.label": "Hotels and Motels"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1350,9 +1508,25 @@
         "model_output.classification.iab_content.tier3.label": "Hotels and Motels"
       },
       "id": "hotels-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
+          "expected": "Travel Type",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Hotels and Motels",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Travel > Travel Type > Hotels and Motels.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Best hotels near Times Square for a weekend trip"
     },
@@ -1360,7 +1534,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Travel",
-        "model_output.classification.iab_content.tier2.label": "Travel Type",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1371,6 +1545,11 @@
       },
       "id": "hotels-hard",
       "mismatches": [
+        {
+          "actual": null,
+          "expected": "Travel Type",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
         {
           "actual": null,
           "expected": "Hotels and Motels",
@@ -1457,8 +1636,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Healthy Living",
-        "model_output.classification.iab_content.tier2.label": "Fitness and Exercise",
-        "model_output.classification.iab_content.tier3.label": "Running and Jogging"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1467,9 +1646,20 @@
         "model_output.classification.iab_content.tier3.label": "Running and Jogging"
       },
       "id": "running-and-jogging-easy",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": null,
+          "expected": "Fitness and Exercise",
+          "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": null,
+          "expected": "Running and Jogging",
+          "path": "model_output.classification.iab_content.tier3.label"
+        }
+      ],
       "notes": "Cross-vertical easy IAB mapping case for Healthy Living > Fitness and Exercise > Running and Jogging.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Best running plan for a first 10k"
     },
@@ -1477,7 +1667,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Sports",
-        "model_output.classification.iab_content.tier2.label": "Walking",
+        "model_output.classification.iab_content.tier2.label": "Bodybuilding",
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -1494,7 +1684,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Walking",
+          "actual": "Bodybuilding",
           "expected": "Fitness and Exercise",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -1630,9 +1820,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Travel",
-        "model_output.classification.iab_content.tier2.label": "Travel Type"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1647,12 +1837,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
-        },
-        {
-          "actual": "Travel Type",
+          "actual": null,
           "expected": "Fiction",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -1665,8 +1850,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Books and Literature",
-        "model_output.classification.iab_content.tier2.label": "Fiction"
+        "model_output.classification.iab_content.tier1.label": "Genres",
+        "model_output.classification.iab_content.tier2.label": "Romance"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1674,9 +1859,20 @@
         "model_output.classification.iab_content.tier2.label": "Fiction"
       },
       "id": "fiction-hard",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Genres",
+          "expected": "Books and Literature",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "Romance",
+          "expected": "Fiction",
+          "path": "model_output.classification.iab_content.tier2.label"
+        }
+      ],
       "notes": "Cross-vertical hard IAB mapping case for Books and Literature > Fiction.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Looking for a character-driven novel, not comics or poetry"
     },
@@ -1684,7 +1880,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Home & Garden",
-        "model_output.classification.iab_content.tier2.label": "Interior Decorating"
+        "model_output.classification.iab_content.tier2.label": "Remodeling & Construction"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1694,7 +1890,7 @@
       "id": "home-improvement-easy",
       "mismatches": [
         {
-          "actual": "Interior Decorating",
+          "actual": "Remodeling & Construction",
           "expected": "Home Improvement",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -1707,8 +1903,8 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Home & Garden",
-        "model_output.classification.iab_content.tier2.label": "Interior Decorating"
+        "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+        "model_output.classification.iab_content.tier2.label": "Personal Care"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1718,7 +1914,12 @@
       "id": "home-improvement-medium",
       "mismatches": [
         {
-          "actual": "Interior Decorating",
+          "actual": "Style & Fashion",
+          "expected": "Home & Garden",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "Personal Care",
           "expected": "Home Improvement",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -1759,9 +1960,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Augmented Reality"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Careers",
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -1771,12 +1972,17 @@
       "id": "online-education-easy",
       "mismatches": [
         {
-          "actual": "Technology & Computing",
+          "actual": "Careers",
           "expected": "Education",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Augmented Reality",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Online Education",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -1906,23 +2112,34 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Medical Health"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Food & Drink"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Medical Health"
       },
       "id": "medical-health-easy",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Food & Drink",
+          "expected": "Medical Health",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        }
+      ],
       "notes": "Cross-vertical easy IAB mapping case for Medical Health.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "what do these allergy symptoms mean"
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Medical Health"
       },
       "expected": {
@@ -1930,9 +2147,15 @@
         "model_output.classification.iab_content.tier1.label": "Medical Health"
       },
       "id": "medical-health-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Medical Health.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "when should i see a doctor for persistent knee pain"
     },
@@ -2036,9 +2259,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Holidays",
-        "model_output.classification.iab_content.tier2.label": "National & Civic Holidays"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -2053,7 +2276,12 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "National & Civic Holidays",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Financial Planning",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -2131,9 +2359,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Genres",
-        "model_output.classification.iab_content.tier2.label": "Family/Children"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Shopping",
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -2143,12 +2371,17 @@
       "id": "parenting-medium",
       "mismatches": [
         {
-          "actual": "Genres",
+          "actual": "Shopping",
           "expected": "Family and Relationships",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Family/Children",
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
+        },
+        {
+          "actual": null,
           "expected": "Parenting",
           "path": "model_output.classification.iab_content.tier2.label"
         }
@@ -2234,7 +2467,7 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Entertainment",
-        "model_output.classification.iab_content.tier2.label": "Movies"
+        "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -2242,17 +2475,23 @@
         "model_output.classification.iab_content.tier2.label": "Movies"
       },
       "id": "movies-easy",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": null,
+          "expected": "Movies",
+          "path": "model_output.classification.iab_content.tier2.label"
+        }
+      ],
       "notes": "Cross-vertical easy IAB mapping case for Entertainment > Movies.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "What movie should we watch tonight?"
     },
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Entertainment",
-        "model_output.classification.iab_content.tier2.label": "Movies"
+        "model_output.classification.iab_content.tier1.label": "Genres",
+        "model_output.classification.iab_content.tier2.label": "Horror"
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "exact",
@@ -2260,9 +2499,20 @@
         "model_output.classification.iab_content.tier2.label": "Movies"
       },
       "id": "movies-medium",
-      "mismatches": [],
+      "mismatches": [
+        {
+          "actual": "Genres",
+          "expected": "Entertainment",
+          "path": "model_output.classification.iab_content.tier1.label"
+        },
+        {
+          "actual": "Horror",
+          "expected": "Movies",
+          "path": "model_output.classification.iab_content.tier2.label"
+        }
+      ],
       "notes": "Cross-vertical medium IAB mapping case for Entertainment > Movies.",
-      "pass": true,
+      "pass": false,
       "status": "must_fix",
       "text": "Best thriller movies from the last few years"
     },
diff --git a/artifacts/evaluation/latest/iab_quality_target_eval.json b/artifacts/evaluation/latest/iab_quality_target_eval.json
index 78aa1dece0ad471d7121e942adfff34294db59a2..490411ecf947ac8d5a222bf5e42b71d6f025cfd7 100644
--- a/artifacts/evaluation/latest/iab_quality_target_eval.json
+++ b/artifacts/evaluation/latest/iab_quality_target_eval.json
@@ -13,7 +13,7 @@
   "results": [
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Automotive",
         "model_output.classification.iab_content.tier2.label": null
       },
@@ -28,6 +28,11 @@
           "actual": null,
           "expected": "Auto Buying and Selling",
           "path": "model_output.classification.iab_content.tier2.label"
+        },
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
         }
       ],
       "notes": "Vehicle shopping queries should map into the automotive buying branch, not business sales.",
@@ -90,8 +95,8 @@
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "exact",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Software and Applications"
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -107,12 +112,12 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Computing",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Software and Applications",
+          "actual": null,
           "expected": "Sales",
           "path": "model_output.classification.iab_content.tier3.label"
         },
@@ -129,9 +134,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Robotics",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -148,7 +153,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Robotics",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -156,6 +161,11 @@
           "actual": null,
           "expected": "Sales",
           "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
         }
       ],
       "notes": "Direct CRM vendor comparison should map cleanly into the sales domain.",
@@ -165,9 +175,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
         "model_output.classification.iab_content.tier1.label": "Careers",
-        "model_output.classification.iab_content.tier2.label": "Job Search",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -184,7 +194,7 @@
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Job Search",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -192,6 +202,11 @@
           "actual": null,
           "expected": "Marketing and Advertising",
           "path": "model_output.classification.iab_content.tier3.label"
+        },
+        {
+          "actual": "nearest_equivalent",
+          "expected": "exact",
+          "path": "model_output.classification.iab_content.mapping_mode"
         }
       ],
       "notes": "Marketing tool discovery should map to the marketing and advertising branch.",
@@ -202,7 +217,7 @@
     {
       "actual": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-        "model_output.classification.iab_content.tier1.label": "Science",
+        "model_output.classification.iab_content.tier1.label": "Real Estate",
         "model_output.classification.iab_content.tier2.label": null
       },
       "expected": {
@@ -213,7 +228,7 @@
       "id": "ml-explanation-maps-to-ai",
       "mismatches": [
         {
-          "actual": "Science",
+          "actual": "Real Estate",
           "expected": "Technology & Computing",
           "path": "model_output.classification.iab_content.tier1.label"
         },
@@ -235,10 +250,10 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-        "model_output.classification.iab_content.tier2.label": "Computing",
-        "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Personal Finance",
+        "model_output.classification.iab_content.tier2.label": null,
+        "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
         "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -249,24 +264,19 @@
       "id": "support-credential-help-maps-to-business-it",
       "mismatches": [
         {
-          "actual": "Technology & Computing",
+          "actual": "Personal Finance",
           "expected": "Business and Finance",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Computing",
+          "actual": null,
           "expected": "Business",
           "path": "model_output.classification.iab_content.tier2.label"
         },
         {
-          "actual": "Information and Network Security",
+          "actual": null,
           "expected": "Business I.T.",
           "path": "model_output.classification.iab_content.tier3.label"
-        },
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
         }
       ],
       "notes": "Credential and account help should map to business IT rather than generic business.",
@@ -294,9 +304,9 @@
     },
     {
       "actual": {
-        "model_output.classification.iab_content.mapping_mode": "exact",
-        "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-        "model_output.classification.iab_content.tier2.label": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+        "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+        "model_output.classification.iab_content.tier1.label": "Sports",
+        "model_output.classification.iab_content.tier2.label": null,
         "model_output.classification.iab_content.tier3.label": null
       },
       "expected": {
@@ -308,12 +318,12 @@
       "id": "trial-signup-maps-to-software",
       "mismatches": [
         {
-          "actual": "Sensitive Topics",
+          "actual": "Sports",
           "expected": "Technology & Computing",
           "path": "model_output.classification.iab_content.tier1.label"
         },
         {
-          "actual": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+          "actual": null,
           "expected": "Computing",
           "path": "model_output.classification.iab_content.tier2.label"
         },
@@ -321,11 +331,6 @@
           "actual": null,
           "expected": "Software and Applications",
           "path": "model_output.classification.iab_content.tier3.label"
-        },
-        {
-          "actual": "exact",
-          "expected": "nearest_equivalent",
-          "path": "model_output.classification.iab_content.mapping_mode"
         }
       ],
       "notes": "Software action queries should map to the software/application branch.",
diff --git a/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_confusion_matrix.csv b/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_confusion_matrix.csv
index 16ff6b8456f51d6f9c8270dcbe9a79564117e0ba..f770438e5f907d7b0905ee503cb17c3020cca501 100644
--- a/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_confusion_matrix.csv
@@ -1,19 +1,19 @@
 ,education,product_discovery,comparison,evaluation,deal_seeking,provider_selection,signup,purchase,booking,download,contact_sales,task_execution,onboarding_setup,troubleshooting,account_help,billing_help,follow_up,emotional_reflection
 education,15,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-product_discovery,0,14,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-comparison,2,1,11,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0
-evaluation,1,0,0,13,0,0,0,0,0,0,0,0,1,0,0,0,0,0
-deal_seeking,0,1,0,0,13,1,0,0,0,0,0,0,0,0,0,0,0,0
+product_discovery,0,15,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+comparison,2,0,12,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+evaluation,1,0,0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+deal_seeking,0,2,0,0,12,1,0,0,0,0,0,0,0,0,0,0,0,0
 provider_selection,0,0,0,0,0,16,0,0,0,0,0,0,0,0,0,0,0,0
-signup,0,0,0,0,0,0,15,1,0,0,0,0,0,0,0,0,0,0
-purchase,0,0,0,0,0,0,0,13,0,0,0,0,2,0,0,0,0,0
-booking,0,0,0,0,0,0,1,0,13,0,1,0,0,0,0,0,0,0
-download,0,0,0,0,0,0,0,0,0,13,1,1,0,0,0,0,0,0
+signup,0,0,0,0,0,0,14,0,0,0,0,2,0,0,0,0,0,0
+purchase,0,0,0,0,0,0,1,12,0,0,0,0,2,0,0,0,0,0
+booking,0,0,0,0,0,0,3,0,9,0,1,2,0,0,0,0,0,0
+download,0,0,0,0,0,0,0,0,0,14,0,1,0,0,0,0,0,0
 contact_sales,0,0,0,0,0,0,0,0,0,0,15,0,0,0,0,0,0,0
-task_execution,0,0,0,0,0,0,0,0,0,0,0,18,0,0,0,0,0,0
-onboarding_setup,0,0,0,0,0,0,0,0,0,0,0,1,16,0,0,0,0,0
-troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,12,1,2,0,0
-account_help,0,0,0,0,0,0,2,0,0,0,0,1,0,3,8,1,0,0
-billing_help,0,0,0,0,0,0,0,0,0,0,0,0,1,0,1,13,0,0
-follow_up,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,14,0
+task_execution,0,0,0,0,0,0,0,0,0,0,0,17,1,0,0,0,0,0
+onboarding_setup,0,0,0,0,0,0,0,0,0,0,0,0,17,0,0,0,0,0
+troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,11,1,2,0,1
+account_help,0,0,0,0,0,0,0,0,0,0,1,1,0,3,10,0,0,0
+billing_help,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2,13,0,0
+follow_up,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,14,1
 emotional_reflection,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,15
diff --git a/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_report.json b/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_report.json
index 4b70173425fd552225eb16e2466fd8636e06340f..1be4d328caca9d99263f2e7a21bcfc32a15de83f 100644
--- a/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_report.json
+++ b/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_report.json
@@ -1,7 +1,7 @@
 {
-  "accepted_accuracy": 0.9104,
-  "accepted_coverage": 0.9675,
-  "accuracy": 0.8917,
+  "accepted_accuracy": 0.8901,
+  "accepted_coverage": 0.9856,
+  "accuracy": 0.8845,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_confusion_matrix.csv",
   "count": 277,
   "dataset_path": "/content/agentic-intent-classifier/data/subtype_benchmark.jsonl",
@@ -12,52 +12,52 @@
       "accuracy": 0.913,
       "count": 92,
       "fallback_rate": 0.0,
-      "macro_f1": 0.9109
+      "macro_f1": 0.9124
     },
     "hard": {
-      "accepted_accuracy": 0.8554,
-      "accepted_coverage": 0.9121,
-      "accuracy": 0.8132,
+      "accepted_accuracy": 0.8295,
+      "accepted_coverage": 0.967,
+      "accuracy": 0.8242,
       "count": 91,
-      "fallback_rate": 0.0879,
-      "macro_f1": 0.8025
+      "fallback_rate": 0.033,
+      "macro_f1": 0.8183
     },
     "medium": {
-      "accepted_accuracy": 0.957,
+      "accepted_accuracy": 0.9247,
       "accepted_coverage": 0.9894,
-      "accuracy": 0.9468,
+      "accuracy": 0.9149,
       "count": 94,
       "fallback_rate": 0.0106,
-      "macro_f1": 0.9469
+      "macro_f1": 0.9117
     }
   },
-  "fallback_rate": 0.0325,
+  "fallback_rate": 0.0144,
   "head": "intent_subtype",
-  "macro_f1": 0.8886,
+  "macro_f1": 0.8824,
   "per_class_metrics": {
     "account_help": {
-      "f1-score": 0.64,
-      "precision": 0.8,
-      "recall": 0.5333333333333333,
+      "f1-score": 0.7142857142857143,
+      "precision": 0.7692307692307693,
+      "recall": 0.6666666666666666,
       "support": 15.0
     },
-    "accuracy": 0.8916967509025271,
+    "accuracy": 0.8844765342960289,
     "billing_help": {
-      "f1-score": 0.8387096774193549,
-      "precision": 0.8125,
+      "f1-score": 0.8666666666666667,
+      "precision": 0.8666666666666667,
       "recall": 0.8666666666666667,
       "support": 15.0
     },
     "booking": {
-      "f1-score": 0.9285714285714286,
+      "f1-score": 0.75,
       "precision": 1.0,
-      "recall": 0.8666666666666667,
+      "recall": 0.6,
       "support": 15.0
     },
     "comparison": {
-      "f1-score": 0.8148148148148148,
-      "precision": 0.9166666666666666,
-      "recall": 0.7333333333333333,
+      "f1-score": 0.8888888888888888,
+      "precision": 1.0,
+      "recall": 0.8,
       "support": 15.0
     },
     "contact_sales": {
@@ -67,15 +67,15 @@
       "support": 15.0
     },
     "deal_seeking": {
-      "f1-score": 0.896551724137931,
-      "precision": 0.9285714285714286,
-      "recall": 0.8666666666666667,
+      "f1-score": 0.8888888888888888,
+      "precision": 1.0,
+      "recall": 0.8,
       "support": 15.0
     },
     "download": {
-      "f1-score": 0.9285714285714286,
+      "f1-score": 0.9655172413793104,
       "precision": 1.0,
-      "recall": 0.8666666666666667,
+      "recall": 0.9333333333333333,
       "support": 15.0
     },
     "education": {
@@ -85,15 +85,15 @@
       "support": 15.0
     },
     "emotional_reflection": {
-      "f1-score": 1.0,
-      "precision": 1.0,
+      "f1-score": 0.9375,
+      "precision": 0.8823529411764706,
       "recall": 1.0,
       "support": 15.0
     },
     "evaluation": {
-      "f1-score": 0.896551724137931,
-      "precision": 0.9285714285714286,
-      "recall": 0.8666666666666667,
+      "f1-score": 0.9333333333333333,
+      "precision": 0.9333333333333333,
+      "recall": 0.9333333333333333,
       "support": 15.0
     },
     "follow_up": {
@@ -103,21 +103,21 @@
       "support": 15.0
     },
     "macro avg": {
-      "f1-score": 0.8886471209737711,
-      "precision": 0.8965122159975102,
-      "recall": 0.8895561002178651,
+      "f1-score": 0.8824228919733669,
+      "precision": 0.8968567719420234,
+      "recall": 0.8825617283950618,
       "support": 277.0
     },
     "onboarding_setup": {
-      "f1-score": 0.8648648648648649,
-      "precision": 0.8,
-      "recall": 0.9411764705882353,
+      "f1-score": 0.918918918918919,
+      "precision": 0.85,
+      "recall": 1.0,
       "support": 17.0
     },
     "product_discovery": {
-      "f1-score": 0.9032258064516129,
-      "precision": 0.875,
-      "recall": 0.9333333333333333,
+      "f1-score": 0.9375,
+      "precision": 0.8823529411764706,
+      "recall": 1.0,
       "support": 15.0
     },
     "provider_selection": {
@@ -127,33 +127,33 @@
       "support": 16.0
     },
     "purchase": {
-      "f1-score": 0.896551724137931,
-      "precision": 0.9285714285714286,
-      "recall": 0.8666666666666667,
+      "f1-score": 0.8888888888888888,
+      "precision": 1.0,
+      "recall": 0.8,
       "support": 15.0
     },
     "signup": {
-      "f1-score": 0.8823529411764706,
-      "precision": 0.8333333333333334,
-      "recall": 0.9375,
+      "f1-score": 0.8235294117647058,
+      "precision": 0.7777777777777778,
+      "recall": 0.875,
       "support": 16.0
     },
     "task_execution": {
-      "f1-score": 0.9230769230769231,
-      "precision": 0.8571428571428571,
-      "recall": 1.0,
+      "f1-score": 0.8292682926829268,
+      "precision": 0.7391304347826086,
+      "recall": 0.9444444444444444,
       "support": 18.0
     },
     "troubleshooting": {
-      "f1-score": 0.8,
-      "precision": 0.8,
-      "recall": 0.8,
+      "f1-score": 0.7586206896551724,
+      "precision": 0.7857142857142857,
+      "recall": 0.7333333333333333,
       "support": 15.0
     },
     "weighted avg": {
-      "f1-score": 0.8891181699377334,
-      "precision": 0.8953221541324111,
-      "recall": 0.8916967509025271,
+      "f1-score": 0.8822131766431675,
+      "precision": 0.8945403392673653,
+      "recall": 0.8844765342960289,
       "support": 277.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_subtype_extended_cases_confusion_matrix.csv b/artifacts/evaluation/latest/intent_subtype_extended_cases_confusion_matrix.csv
index 32f3cc3b353284a7c9ed993adff29ad28752c3d8..b2db308583b6fd5ba3f9102eb0491bfdf54f87b0 100644
--- a/artifacts/evaluation/latest/intent_subtype_extended_cases_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_subtype_extended_cases_confusion_matrix.csv
@@ -1,9 +1,9 @@
 ,education,product_discovery,comparison,evaluation,deal_seeking,provider_selection,signup,purchase,booking,download,contact_sales,task_execution,onboarding_setup,troubleshooting,account_help,billing_help,follow_up,emotional_reflection
 education,7,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 product_discovery,0,6,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-comparison,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+comparison,0,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0
 evaluation,1,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0
-deal_seeking,0,0,0,0,9,0,0,0,0,0,0,0,0,0,0,0,0,0
+deal_seeking,1,0,0,0,8,0,0,0,0,0,0,0,0,0,0,0,0,0
 provider_selection,0,0,0,0,0,5,0,0,0,0,0,0,0,0,0,0,0,0
 signup,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 purchase,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
@@ -15,5 +15,5 @@ onboarding_setup,0,0,0,0,0,0,0,0,0,0,0,0,4,0,0,0,0,0
 troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0
 account_help,0,0,0,0,0,0,0,0,0,0,0,0,0,1,2,0,0,0
 billing_help,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-follow_up,1,0,0,0,3,0,0,0,0,0,0,1,0,0,0,0,7,0
+follow_up,0,0,0,0,3,0,0,0,0,0,0,0,0,0,1,0,8,0
 emotional_reflection,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
diff --git a/artifacts/evaluation/latest/intent_subtype_extended_cases_report.json b/artifacts/evaluation/latest/intent_subtype_extended_cases_report.json
index 9827fc99e6be6d8aa6c64a1d86dd23b70aa03055..5c726c13a7efc21faf7f05b5418c8d4a1f931d25 100644
--- a/artifacts/evaluation/latest/intent_subtype_extended_cases_report.json
+++ b/artifacts/evaluation/latest/intent_subtype_extended_cases_report.json
@@ -1,21 +1,21 @@
 {
-  "accepted_accuracy": 0.8302,
+  "accepted_accuracy": 0.8113,
   "accepted_coverage": 1.0,
-  "accuracy": 0.8302,
+  "accuracy": 0.8113,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_extended_cases_confusion_matrix.csv",
   "count": 53,
   "dataset_path": "/content/agentic-intent-classifier/data/subtype/extended_cases.jsonl",
   "fallback_rate": 0.0,
   "head": "intent_subtype",
-  "macro_f1": 0.7668,
+  "macro_f1": 0.7517,
   "per_class_metrics": {
     "account_help": {
-      "f1-score": 0.8,
-      "precision": 1.0,
+      "f1-score": 0.6666666666666666,
+      "precision": 0.6666666666666666,
       "recall": 0.6666666666666666,
       "support": 3.0
     },
-    "accuracy": 0.8301886792452831,
+    "accuracy": 0.8113207547169812,
     "billing_help": {
       "f1-score": 0.0,
       "precision": 0.0,
@@ -29,9 +29,9 @@
       "support": 0.0
     },
     "comparison": {
-      "f1-score": 1.0,
+      "f1-score": 0.6666666666666666,
       "precision": 1.0,
-      "recall": 1.0,
+      "recall": 0.5,
       "support": 2.0
     },
     "contact_sales": {
@@ -41,9 +41,9 @@
       "support": 0.0
     },
     "deal_seeking": {
-      "f1-score": 0.8571428571428571,
-      "precision": 0.75,
-      "recall": 1.0,
+      "f1-score": 0.7619047619047619,
+      "precision": 0.6666666666666666,
+      "recall": 0.8888888888888888,
       "support": 9.0
     },
     "download": {
@@ -71,15 +71,15 @@
       "support": 3.0
     },
     "follow_up": {
-      "f1-score": 0.7368421052631579,
+      "f1-score": 0.8,
       "precision": 1.0,
-      "recall": 0.5833333333333334,
+      "recall": 0.6666666666666666,
       "support": 12.0
     },
     "macro avg": {
-      "f1-score": 0.46858256266151,
-      "precision": 0.4565696649029982,
-      "recall": 0.513888888888889,
+      "f1-score": 0.45939292189292186,
+      "precision": 0.4611992945326278,
+      "recall": 0.48456790123456783,
       "support": 53.0
     },
     "onboarding_setup": {
@@ -113,8 +113,8 @@
       "support": 0.0
     },
     "task_execution": {
-      "f1-score": 0.6666666666666666,
-      "precision": 0.5,
+      "f1-score": 1.0,
+      "precision": 1.0,
       "recall": 1.0,
       "support": 1.0
     },
@@ -125,9 +125,9 @@
       "support": 1.0
     },
     "weighted avg": {
-      "f1-score": 0.8018611395225099,
-      "precision": 0.8208295896975142,
-      "recall": 0.8301886792452831,
+      "f1-score": 0.7861520554916781,
+      "precision": 0.7972446840371369,
+      "recall": 0.8113207547169812,
       "support": 53.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_subtype_hard_cases_confusion_matrix.csv b/artifacts/evaluation/latest/intent_subtype_hard_cases_confusion_matrix.csv
index 5e6557f59f22cd5799cd4591e735e99566b12971..15de48438e9e1f282283839804876ebf5b7650e7 100644
--- a/artifacts/evaluation/latest/intent_subtype_hard_cases_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_subtype_hard_cases_confusion_matrix.csv
@@ -2,7 +2,7 @@
 education,29,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 product_discovery,0,8,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 comparison,0,0,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-evaluation,3,2,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+evaluation,2,3,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 deal_seeking,0,0,0,0,2,1,0,0,0,0,0,0,0,0,0,0,0,0
 provider_selection,0,0,0,1,0,9,0,0,0,0,0,0,0,0,0,0,0,0
 signup,0,0,0,0,0,0,3,0,0,0,0,0,0,0,0,0,0,0
diff --git a/artifacts/evaluation/latest/intent_subtype_hard_cases_report.json b/artifacts/evaluation/latest/intent_subtype_hard_cases_report.json
index ae7f0904cfd23d66cb645f7662fa628497025d30..bb2a077d9aa6aa53f91864f064b671568953f33b 100644
--- a/artifacts/evaluation/latest/intent_subtype_hard_cases_report.json
+++ b/artifacts/evaluation/latest/intent_subtype_hard_cases_report.json
@@ -7,7 +7,7 @@
   "dataset_path": "/content/agentic-intent-classifier/data/subtype/hard_cases.jsonl",
   "fallback_rate": 0.0,
   "head": "intent_subtype",
-  "macro_f1": 0.8447,
+  "macro_f1": 0.8426,
   "per_class_metrics": {
     "account_help": {
       "f1-score": 0.8,
@@ -53,8 +53,8 @@
       "support": 0.0
     },
     "education": {
-      "f1-score": 0.9508196721311475,
-      "precision": 0.90625,
+      "f1-score": 0.9666666666666667,
+      "precision": 0.9354838709677419,
       "recall": 1.0,
       "support": 29.0
     },
@@ -77,8 +77,8 @@
       "support": 12.0
     },
     "macro avg": {
-      "f1-score": 0.7038911013311109,
-      "precision": 0.7234953703703704,
+      "f1-score": 0.7021723995980289,
+      "precision": 0.7210790702726187,
       "recall": 0.7212962962962962,
       "support": 94.0
     },
@@ -89,8 +89,8 @@
       "support": 6.0
     },
     "product_discovery": {
-      "f1-score": 0.8888888888888888,
-      "precision": 0.8,
+      "f1-score": 0.8421052631578947,
+      "precision": 0.7272727272727273,
       "recall": 1.0,
       "support": 8.0
     },
@@ -125,8 +125,8 @@
       "support": 3.0
     },
     "weighted avg": {
-      "f1-score": 0.8798004011763282,
-      "precision": 0.8911125886524823,
+      "f1-score": 0.8807077824069889,
+      "precision": 0.8939419937189327,
       "recall": 0.8936170212765957,
       "support": 94.0
     }
diff --git a/artifacts/evaluation/latest/intent_subtype_test_confusion_matrix.csv b/artifacts/evaluation/latest/intent_subtype_test_confusion_matrix.csv
index 64a5f1bbdf4a4a58655b34163d11851128aa6e0e..e118be1e0635b58dab599e9cc24d9077b75eb2db 100644
--- a/artifacts/evaluation/latest/intent_subtype_test_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_subtype_test_confusion_matrix.csv
@@ -1,9 +1,9 @@
 ,education,product_discovery,comparison,evaluation,deal_seeking,provider_selection,signup,purchase,booking,download,contact_sales,task_execution,onboarding_setup,troubleshooting,account_help,billing_help,follow_up,emotional_reflection
 education,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-product_discovery,0,8,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+product_discovery,0,7,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 comparison,0,0,3,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-evaluation,0,0,1,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0
-deal_seeking,0,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0
+evaluation,0,0,0,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0
+deal_seeking,1,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0
 provider_selection,0,0,0,1,0,4,0,0,0,0,0,0,1,0,0,0,0,0
 signup,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0
 purchase,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
@@ -12,8 +12,8 @@ download,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 contact_sales,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 task_execution,0,0,0,0,0,0,0,0,0,0,0,6,0,0,0,0,0,0
 onboarding_setup,0,0,0,0,0,0,0,0,0,0,0,0,4,0,0,0,0,0
-troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,0,0
+troubleshooting,0,0,0,0,0,0,0,0,0,0,0,1,0,1,0,0,0,0
 account_help,0,0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,0
 billing_help,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-follow_up,1,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,8,0
+follow_up,0,0,0,0,2,0,0,0,0,0,0,1,0,0,0,0,8,0
 emotional_reflection,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,5
diff --git a/artifacts/evaluation/latest/intent_subtype_test_report.json b/artifacts/evaluation/latest/intent_subtype_test_report.json
index 124bd1866a7fc31b6e804433ae076df68ddca2ad..120cbf10fc566e55ada2da80fc7a088f28800006 100644
--- a/artifacts/evaluation/latest/intent_subtype_test_report.json
+++ b/artifacts/evaluation/latest/intent_subtype_test_report.json
@@ -1,13 +1,13 @@
 {
-  "accepted_accuracy": 0.9,
+  "accepted_accuracy": 0.8714,
   "accepted_coverage": 1.0,
-  "accuracy": 0.9,
+  "accuracy": 0.8714,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_test_confusion_matrix.csv",
   "count": 70,
   "dataset_path": "/content/agentic-intent-classifier/data/subtype/test.jsonl",
   "fallback_rate": 0.0,
   "head": "intent_subtype",
-  "macro_f1": 0.8531,
+  "macro_f1": 0.8317,
   "per_class_metrics": {
     "account_help": {
       "f1-score": 1.0,
@@ -15,7 +15,7 @@
       "recall": 1.0,
       "support": 2.0
     },
-    "accuracy": 0.9,
+    "accuracy": 0.8714285714285714,
     "billing_help": {
       "f1-score": 0.0,
       "precision": 0.0,
@@ -41,9 +41,9 @@
       "support": 0.0
     },
     "deal_seeking": {
-      "f1-score": 0.6666666666666666,
-      "precision": 0.5,
-      "recall": 1.0,
+      "f1-score": 0.3333333333333333,
+      "precision": 0.25,
+      "recall": 0.5,
       "support": 2.0
     },
     "download": {
@@ -65,9 +65,9 @@
       "support": 5.0
     },
     "evaluation": {
-      "f1-score": 0.0,
-      "precision": 0.0,
-      "recall": 0.0,
+      "f1-score": 0.5,
+      "precision": 0.5,
+      "recall": 0.5,
       "support": 2.0
     },
     "follow_up": {
@@ -77,9 +77,9 @@
       "support": 11.0
     },
     "macro avg": {
-      "f1-score": 0.6635221022395855,
-      "precision": 0.6578042328042328,
-      "recall": 0.6885521885521885,
+      "f1-score": 0.646896135613619,
+      "precision": 0.6657407407407407,
+      "recall": 0.6538299663299663,
       "support": 70.0
     },
     "onboarding_setup": {
@@ -89,9 +89,9 @@
       "support": 4.0
     },
     "product_discovery": {
-      "f1-score": 1.0,
+      "f1-score": 0.9333333333333333,
       "precision": 1.0,
-      "recall": 1.0,
+      "recall": 0.875,
       "support": 8.0
     },
     "provider_selection": {
@@ -113,21 +113,21 @@
       "support": 2.0
     },
     "task_execution": {
-      "f1-score": 0.9230769230769231,
-      "precision": 0.8571428571428571,
+      "f1-score": 0.8571428571428571,
+      "precision": 0.75,
       "recall": 1.0,
       "support": 6.0
     },
     "troubleshooting": {
-      "f1-score": 1.0,
+      "f1-score": 0.6666666666666666,
       "precision": 1.0,
-      "recall": 1.0,
+      "recall": 0.5,
       "support": 2.0
     },
     "weighted avg": {
-      "f1-score": 0.8939882610403741,
-      "precision": 0.9094217687074829,
-      "recall": 0.9,
+      "f1-score": 0.8759558172936446,
+      "precision": 0.9073809523809524,
+      "recall": 0.8714285714285714,
       "support": 70.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_subtype_train_confusion_matrix.csv b/artifacts/evaluation/latest/intent_subtype_train_confusion_matrix.csv
index cec2aad62af2ad3c09e31dc591f77602bc7467e0..c1467c69fdefabb0c451845a88e5e12f4d2325ab 100644
--- a/artifacts/evaluation/latest/intent_subtype_train_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_subtype_train_confusion_matrix.csv
@@ -1,19 +1,19 @@
 ,education,product_discovery,comparison,evaluation,deal_seeking,provider_selection,signup,purchase,booking,download,contact_sales,task_execution,onboarding_setup,troubleshooting,account_help,billing_help,follow_up,emotional_reflection
 education,52,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-product_discovery,0,29,1,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0
-comparison,0,0,15,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-evaluation,4,4,0,9,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+product_discovery,0,30,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0
+comparison,0,0,14,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+evaluation,3,5,0,9,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 deal_seeking,0,0,0,0,10,1,0,0,0,0,0,0,0,0,0,0,0,0
 provider_selection,0,0,0,0,0,23,0,0,0,0,0,0,0,0,0,0,2,0
 signup,0,0,0,0,0,0,16,0,0,0,0,0,0,0,0,0,0,0
 purchase,0,0,0,0,0,0,2,4,0,0,0,0,0,0,0,0,0,0
 booking,0,0,0,0,0,0,0,0,5,0,0,0,0,0,0,0,0,0
 download,0,0,0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0
-contact_sales,0,0,0,0,0,0,2,0,1,0,6,0,0,0,0,0,0,0
-task_execution,0,0,0,0,0,0,1,0,0,1,0,17,0,0,0,0,0,0
+contact_sales,0,0,0,0,0,0,2,0,1,0,5,1,0,0,0,0,0,0
+task_execution,0,0,0,0,0,0,0,1,0,0,0,18,0,0,0,0,0,0
 onboarding_setup,0,0,0,0,0,0,0,0,1,0,0,0,16,0,0,0,0,0
-troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,10,2,0,1,0
-account_help,0,0,0,0,0,0,0,0,0,0,0,0,0,2,5,0,0,0
+troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,9,3,0,1,0
+account_help,0,0,0,0,0,0,0,0,0,0,0,0,0,1,6,0,0,0
 billing_help,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,6,0,0
-follow_up,0,0,0,0,1,0,0,0,0,0,0,5,0,0,0,0,30,0
+follow_up,0,0,0,0,0,0,0,0,0,0,0,4,0,0,0,0,32,0
 emotional_reflection,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,20
diff --git a/artifacts/evaluation/latest/intent_subtype_train_report.json b/artifacts/evaluation/latest/intent_subtype_train_report.json
index 96bb5763fdceb3d304d5baafb3cb725f5674bda9..8e4ed55f68a12ff748f66a2c1d102c84dae18381 100644
--- a/artifacts/evaluation/latest/intent_subtype_train_report.json
+++ b/artifacts/evaluation/latest/intent_subtype_train_report.json
@@ -1,21 +1,21 @@
 {
-  "accepted_accuracy": 0.8978,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.8978,
+  "accepted_accuracy": 0.9068,
+  "accepted_coverage": 0.9936,
+  "accuracy": 0.9042,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_train_confusion_matrix.csv",
   "count": 313,
   "dataset_path": "/content/agentic-intent-classifier/data/subtype/train.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.0064,
   "head": "intent_subtype",
-  "macro_f1": 0.877,
+  "macro_f1": 0.8787,
   "per_class_metrics": {
     "account_help": {
-      "f1-score": 0.7142857142857143,
-      "precision": 0.7142857142857143,
-      "recall": 0.7142857142857143,
+      "f1-score": 0.75,
+      "precision": 0.6666666666666666,
+      "recall": 0.8571428571428571,
       "support": 7.0
     },
-    "accuracy": 0.8977635782747604,
+    "accuracy": 0.9041533546325878,
     "billing_help": {
       "f1-score": 1.0,
       "precision": 1.0,
@@ -29,32 +29,32 @@
       "support": 5.0
     },
     "comparison": {
-      "f1-score": 0.967741935483871,
-      "precision": 0.9375,
-      "recall": 1.0,
+      "f1-score": 0.9655172413793104,
+      "precision": 1.0,
+      "recall": 0.9333333333333333,
       "support": 15.0
     },
     "contact_sales": {
-      "f1-score": 0.8,
+      "f1-score": 0.7142857142857143,
       "precision": 1.0,
-      "recall": 0.6666666666666666,
+      "recall": 0.5555555555555556,
       "support": 9.0
     },
     "deal_seeking": {
-      "f1-score": 0.9090909090909091,
-      "precision": 0.9090909090909091,
+      "f1-score": 0.9523809523809523,
+      "precision": 1.0,
       "recall": 0.9090909090909091,
       "support": 11.0
     },
     "download": {
-      "f1-score": 0.9411764705882353,
-      "precision": 0.8888888888888888,
+      "f1-score": 1.0,
+      "precision": 1.0,
       "recall": 1.0,
       "support": 8.0
     },
     "education": {
-      "f1-score": 0.9629629629629629,
-      "precision": 0.9285714285714286,
+      "f1-score": 0.9719626168224299,
+      "precision": 0.9454545454545454,
       "recall": 1.0,
       "support": 52.0
     },
@@ -65,21 +65,21 @@
       "support": 20.0
     },
     "evaluation": {
-      "f1-score": 0.6923076923076923,
-      "precision": 1.0,
+      "f1-score": 0.6666666666666666,
+      "precision": 0.9,
       "recall": 0.5294117647058824,
       "support": 17.0
     },
     "follow_up": {
-      "f1-score": 0.8571428571428571,
-      "precision": 0.8823529411764706,
-      "recall": 0.8333333333333334,
+      "f1-score": 0.8888888888888888,
+      "precision": 0.8888888888888888,
+      "recall": 0.8888888888888888,
       "support": 36.0
     },
     "macro avg": {
-      "f1-score": 0.8770498618135788,
-      "precision": 0.8988923431325393,
-      "recall": 0.876671278202288,
+      "f1-score": 0.8786951095392168,
+      "precision": 0.9007433723013433,
+      "recall": 0.8782602497120292,
       "support": 313.0
     },
     "onboarding_setup": {
@@ -89,9 +89,9 @@
       "support": 17.0
     },
     "product_discovery": {
-      "f1-score": 0.90625,
-      "precision": 0.8787878787878788,
-      "recall": 0.9354838709677419,
+      "f1-score": 0.9090909090909091,
+      "precision": 0.8571428571428571,
+      "recall": 0.967741935483871,
       "support": 31.0
     },
     "provider_selection": {
@@ -101,33 +101,33 @@
       "support": 25.0
     },
     "purchase": {
-      "f1-score": 0.8,
-      "precision": 1.0,
+      "f1-score": 0.7272727272727273,
+      "precision": 0.8,
       "recall": 0.6666666666666666,
       "support": 6.0
     },
     "signup": {
-      "f1-score": 0.8648648648648649,
-      "precision": 0.7619047619047619,
+      "f1-score": 0.8888888888888888,
+      "precision": 0.8,
       "recall": 1.0,
       "support": 16.0
     },
     "task_execution": {
-      "f1-score": 0.8292682926829268,
-      "precision": 0.7727272727272727,
-      "recall": 0.8947368421052632,
+      "f1-score": 0.8571428571428571,
+      "precision": 0.782608695652174,
+      "recall": 0.9473684210526315,
       "support": 19.0
     },
     "troubleshooting": {
-      "f1-score": 0.8,
-      "precision": 0.8333333333333334,
-      "recall": 0.7692307692307693,
+      "f1-score": 0.782608695652174,
+      "precision": 0.9,
+      "recall": 0.6923076923076923,
       "support": 13.0
     },
     "weighted avg": {
-      "f1-score": 0.894423568060199,
-      "precision": 0.9063956713482179,
-      "recall": 0.8977635782747604,
+      "f1-score": 0.9005147505975646,
+      "precision": 0.9118244687664052,
+      "recall": 0.9041533546325878,
       "support": 313.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_subtype_val_confusion_matrix.csv b/artifacts/evaluation/latest/intent_subtype_val_confusion_matrix.csv
index 833eb8abc8d04eb31f19748c94edb8012823b2a6..27f1d7777abb47fc69ba3374373ec65f8b8ee72a 100644
--- a/artifacts/evaluation/latest/intent_subtype_val_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_subtype_val_confusion_matrix.csv
@@ -1,12 +1,12 @@
 ,education,product_discovery,comparison,evaluation,deal_seeking,provider_selection,signup,purchase,booking,download,contact_sales,task_execution,onboarding_setup,troubleshooting,account_help,billing_help,follow_up,emotional_reflection
 education,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-product_discovery,0,9,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
-comparison,0,0,2,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+product_discovery,0,11,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
+comparison,0,0,1,2,1,0,0,0,0,0,0,0,0,0,0,0,0,0
 evaluation,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 deal_seeking,0,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0,0,0
 provider_selection,0,0,0,0,0,7,0,0,0,0,0,0,0,0,0,0,0,0
 signup,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0
-purchase,0,0,0,0,0,0,1,0,0,0,0,1,0,0,0,0,0,0
+purchase,0,0,0,0,0,0,2,0,0,0,0,0,0,0,0,0,0,0
 booking,0,0,0,0,0,0,0,0,3,0,0,0,0,0,0,0,0,0
 download,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
 contact_sales,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
@@ -14,6 +14,6 @@ task_execution,0,0,0,0,0,0,0,0,0,0,0,8,0,0,0,0,0,0
 onboarding_setup,0,1,0,0,0,0,0,0,0,0,0,0,4,0,0,0,0,0
 troubleshooting,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0
 account_help,0,0,0,0,0,0,0,0,0,0,0,0,0,1,1,0,0,0
-billing_help,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0
+billing_help,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0
 follow_up,0,0,0,0,1,0,0,0,0,0,0,1,0,0,0,0,9,0
 emotional_reflection,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,5
diff --git a/artifacts/evaluation/latest/intent_subtype_val_report.json b/artifacts/evaluation/latest/intent_subtype_val_report.json
index 30a623834fc4e573b7c309f7df3df57eb565d43d..96ad039e77173ac9c1c7f75f25768d1c1ec2923a 100644
--- a/artifacts/evaluation/latest/intent_subtype_val_report.json
+++ b/artifacts/evaluation/latest/intent_subtype_val_report.json
@@ -1,13 +1,13 @@
 {
-  "accepted_accuracy": 0.8608,
-  "accepted_coverage": 0.9875,
-  "accuracy": 0.85,
+  "accepted_accuracy": 0.8625,
+  "accepted_coverage": 1.0,
+  "accuracy": 0.8625,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_val_confusion_matrix.csv",
   "count": 80,
   "dataset_path": "/content/agentic-intent-classifier/data/subtype/val.jsonl",
-  "fallback_rate": 0.0125,
+  "fallback_rate": 0.0,
   "head": "intent_subtype",
-  "macro_f1": 0.6722,
+  "macro_f1": 0.6561,
   "per_class_metrics": {
     "account_help": {
       "f1-score": 0.5,
@@ -15,7 +15,7 @@
       "recall": 0.5,
       "support": 2.0
     },
-    "accuracy": 0.85,
+    "accuracy": 0.8625,
     "billing_help": {
       "f1-score": 0.0,
       "precision": 0.0,
@@ -29,9 +29,9 @@
       "support": 3.0
     },
     "comparison": {
-      "f1-score": 0.5,
-      "precision": 0.5,
-      "recall": 0.5,
+      "f1-score": 0.4,
+      "precision": 1.0,
+      "recall": 0.25,
       "support": 4.0
     },
     "contact_sales": {
@@ -41,8 +41,8 @@
       "support": 0.0
     },
     "deal_seeking": {
-      "f1-score": 0.8,
-      "precision": 0.6666666666666666,
+      "f1-score": 0.6666666666666666,
+      "precision": 0.5,
       "recall": 1.0,
       "support": 2.0
     },
@@ -77,21 +77,21 @@
       "support": 11.0
     },
     "macro avg": {
-      "f1-score": 0.5974890931031281,
-      "precision": 0.5811447811447812,
-      "recall": 0.6353535353535353,
+      "f1-score": 0.5832054560954817,
+      "precision": 0.5891975308641975,
+      "recall": 0.6315656565656567,
       "support": 80.0
     },
     "onboarding_setup": {
-      "f1-score": 0.8888888888888888,
-      "precision": 1.0,
+      "f1-score": 0.8,
+      "precision": 0.8,
       "recall": 0.8,
       "support": 5.0
     },
     "product_discovery": {
-      "f1-score": 0.8571428571428571,
-      "precision": 0.9,
-      "recall": 0.8181818181818182,
+      "f1-score": 0.9565217391304348,
+      "precision": 0.9166666666666666,
+      "recall": 1.0,
       "support": 11.0
     },
     "provider_selection": {
@@ -107,14 +107,14 @@
       "support": 2.0
     },
     "signup": {
-      "f1-score": 0.8,
-      "precision": 0.6666666666666666,
+      "f1-score": 0.6666666666666666,
+      "precision": 0.5,
       "recall": 1.0,
       "support": 2.0
     },
     "task_execution": {
-      "f1-score": 0.8421052631578947,
-      "precision": 0.7272727272727273,
+      "f1-score": 0.9411764705882353,
+      "precision": 0.8888888888888888,
       "recall": 1.0,
       "support": 8.0
     },
@@ -125,9 +125,9 @@
       "support": 1.0
     },
     "weighted avg": {
-      "f1-score": 0.8380398913951546,
-      "precision": 0.8423106060606059,
-      "recall": 0.85,
+      "f1-score": 0.8443893861892583,
+      "precision": 0.8649305555555555,
+      "recall": 0.8625,
       "support": 80.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_type_hard_cases_report.json b/artifacts/evaluation/latest/intent_type_hard_cases_report.json
index d4434a92551f3a1dc00712b9dff4fb3a342f3af7..775a5de8fea1b2d20db0781e665a289184e6cc22 100644
--- a/artifacts/evaluation/latest/intent_type_hard_cases_report.json
+++ b/artifacts/evaluation/latest/intent_type_hard_cases_report.json
@@ -1,11 +1,11 @@
 {
   "accepted_accuracy": 1.0,
-  "accepted_coverage": 0.9836,
+  "accepted_coverage": 1.0,
   "accuracy": 1.0,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_hard_cases_confusion_matrix.csv",
   "count": 61,
   "dataset_path": "/content/agentic-intent-classifier/data/hard_cases.jsonl",
-  "fallback_rate": 0.0164,
+  "fallback_rate": 0.0,
   "head": "intent_type",
   "macro_f1": 1.0,
   "per_class_metrics": {
diff --git a/artifacts/evaluation/latest/intent_type_test_confusion_matrix.csv b/artifacts/evaluation/latest/intent_type_test_confusion_matrix.csv
index 8ed8bdaf8c350860a22261b3cfd68c081e6415ae..e0fba415dd7603e364d3843fdcafaaf0437f5158 100644
--- a/artifacts/evaluation/latest/intent_type_test_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_type_test_confusion_matrix.csv
@@ -1,11 +1,11 @@
 ,informational,exploratory,commercial,transactional,support,personal_reflection,creative_generation,chit_chat,ambiguous,prohibited
 informational,8,0,0,0,0,0,0,0,0,0
 exploratory,0,1,0,0,0,0,0,0,0,0
-commercial,0,0,10,0,0,0,0,0,0,0
-transactional,0,0,0,8,0,0,0,0,0,0
+commercial,1,0,9,0,0,0,0,0,0,0
+transactional,0,0,0,7,0,0,1,0,0,0
 support,0,0,0,0,2,0,0,0,0,1
 personal_reflection,0,0,0,0,0,5,0,0,0,0
-creative_generation,0,0,0,0,0,0,1,0,0,0
+creative_generation,0,0,0,1,0,0,0,0,0,0
 chit_chat,0,0,0,0,0,0,0,1,0,0
 ambiguous,1,0,1,0,0,0,0,0,7,0
 prohibited,0,0,0,0,0,0,0,0,0,1
diff --git a/artifacts/evaluation/latest/intent_type_test_report.json b/artifacts/evaluation/latest/intent_type_test_report.json
index 03178f8f35b3c34944ada9a61a79ab034359731a..2065e59475ab2ac96d486ccbbb15046371eb3203 100644
--- a/artifacts/evaluation/latest/intent_type_test_report.json
+++ b/artifacts/evaluation/latest/intent_type_test_report.json
@@ -1,15 +1,15 @@
 {
-  "accepted_accuracy": 0.9362,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.9362,
+  "accepted_accuracy": 0.8889,
+  "accepted_coverage": 0.9574,
+  "accuracy": 0.8723,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_test_confusion_matrix.csv",
   "count": 47,
   "dataset_path": "/content/agentic-intent-classifier/data/test.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.0426,
   "head": "intent_type",
-  "macro_f1": 0.9235,
+  "macro_f1": 0.8006,
   "per_class_metrics": {
-    "accuracy": 0.9361702127659575,
+    "accuracy": 0.8723404255319149,
     "ambiguous": {
       "f1-score": 0.875,
       "precision": 1.0,
@@ -23,15 +23,15 @@
       "support": 1.0
     },
     "commercial": {
-      "f1-score": 0.9523809523809523,
-      "precision": 0.9090909090909091,
-      "recall": 1.0,
+      "f1-score": 0.9,
+      "precision": 0.9,
+      "recall": 0.9,
       "support": 10.0
     },
     "creative_generation": {
-      "f1-score": 1.0,
-      "precision": 1.0,
-      "recall": 1.0,
+      "f1-score": 0.0,
+      "precision": 0.0,
+      "recall": 0.0,
       "support": 1.0
     },
     "exploratory": {
@@ -41,15 +41,15 @@
       "support": 1.0
     },
     "informational": {
-      "f1-score": 0.9411764705882353,
-      "precision": 0.8888888888888888,
+      "f1-score": 0.8888888888888888,
+      "precision": 0.8,
       "recall": 1.0,
       "support": 8.0
     },
     "macro avg": {
-      "f1-score": 0.9235224089635853,
-      "precision": 0.9297979797979797,
-      "recall": 0.9444444444444444,
+      "f1-score": 0.8005555555555555,
+      "precision": 0.8074999999999999,
+      "recall": 0.8219444444444445,
       "support": 47.0
     },
     "personal_reflection": {
@@ -71,15 +71,15 @@
       "support": 3.0
     },
     "transactional": {
-      "f1-score": 1.0,
-      "precision": 1.0,
-      "recall": 1.0,
+      "f1-score": 0.875,
+      "precision": 0.875,
+      "recall": 0.875,
       "support": 8.0
     },
     "weighted avg": {
-      "f1-score": 0.9360614458549377,
-      "precision": 0.9511068128089405,
-      "recall": 0.9361702127659575,
+      "f1-score": 0.8734633569739952,
+      "precision": 0.8914893617021277,
+      "recall": 0.8723404255319149,
       "support": 47.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_type_third_wave_cases_confusion_matrix.csv b/artifacts/evaluation/latest/intent_type_third_wave_cases_confusion_matrix.csv
index dbbd11b91a1bf127e1eb34fd2392aca759cfb272..fffa2bbab6f4d5807c46fb306e41cc880aafc2c5 100644
--- a/artifacts/evaluation/latest/intent_type_third_wave_cases_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_type_third_wave_cases_confusion_matrix.csv
@@ -1,7 +1,7 @@
 ,informational,exploratory,commercial,transactional,support,personal_reflection,creative_generation,chit_chat,ambiguous,prohibited
 informational,0,0,0,0,0,0,0,0,0,0
 exploratory,0,1,0,0,0,0,0,0,0,0
-commercial,0,0,12,0,0,0,0,0,0,0
+commercial,1,0,11,0,0,0,0,0,0,0
 transactional,0,0,0,0,0,0,0,0,0,0
 support,0,0,0,0,0,0,0,0,0,0
 personal_reflection,0,0,0,0,0,0,0,0,0,0
diff --git a/artifacts/evaluation/latest/intent_type_third_wave_cases_report.json b/artifacts/evaluation/latest/intent_type_third_wave_cases_report.json
index 6265d844145b5f42455cb87b9657901a58fea714..af143bc00eed6d5535eb597e699c2bc3fdb3be6f 100644
--- a/artifacts/evaluation/latest/intent_type_third_wave_cases_report.json
+++ b/artifacts/evaluation/latest/intent_type_third_wave_cases_report.json
@@ -1,15 +1,15 @@
 {
-  "accepted_accuracy": 0.8846,
+  "accepted_accuracy": 0.8462,
   "accepted_coverage": 1.0,
-  "accuracy": 0.8846,
+  "accuracy": 0.8462,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_third_wave_cases_confusion_matrix.csv",
   "count": 26,
   "dataset_path": "/content/agentic-intent-classifier/data/third_wave_cases.jsonl",
   "fallback_rate": 0.0,
   "head": "intent_type",
-  "macro_f1": 0.8209,
+  "macro_f1": 0.8148,
   "per_class_metrics": {
-    "accuracy": 0.8846153846153846,
+    "accuracy": 0.8461538461538461,
     "ambiguous": {
       "f1-score": 0.8235294117647058,
       "precision": 1.0,
@@ -23,9 +23,9 @@
       "support": 1.0
     },
     "commercial": {
-      "f1-score": 0.9230769230769231,
-      "precision": 0.8571428571428571,
-      "recall": 1.0,
+      "f1-score": 0.88,
+      "precision": 0.8461538461538461,
+      "recall": 0.9166666666666666,
       "support": 12.0
     },
     "creative_generation": {
@@ -47,9 +47,9 @@
       "support": 0.0
     },
     "macro avg": {
-      "f1-score": 0.5746606334841629,
-      "precision": 0.5857142857142857,
-      "recall": 0.5700000000000001,
+      "f1-score": 0.5703529411764705,
+      "precision": 0.5846153846153846,
+      "recall": 0.5616666666666666,
       "support": 26.0
     },
     "personal_reflection": {
@@ -77,9 +77,9 @@
       "support": 0.0
     },
     "weighted avg": {
-      "f1-score": 0.8966237382526975,
-      "precision": 0.9340659340659341,
-      "recall": 0.8846153846153846,
+      "f1-score": 0.8767420814479638,
+      "precision": 0.9289940828402367,
+      "recall": 0.8461538461538461,
       "support": 26.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_type_train_confusion_matrix.csv b/artifacts/evaluation/latest/intent_type_train_confusion_matrix.csv
index 0bccb99598127decfb008208f7a31553bf527feb..8dc8d14c82ea9c51e9cfd3c11514c27bc8fe0968 100644
--- a/artifacts/evaluation/latest/intent_type_train_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_type_train_confusion_matrix.csv
@@ -7,5 +7,5 @@ support,0,0,0,0,10,0,0,0,0,0
 personal_reflection,0,0,0,0,0,20,0,0,0,0
 creative_generation,0,0,0,0,0,0,5,0,0,0
 chit_chat,0,0,0,0,0,0,0,5,0,0
-ambiguous,0,0,0,0,0,0,0,0,31,0
+ambiguous,0,0,0,0,1,0,0,0,30,0
 prohibited,0,0,0,0,0,0,0,0,0,5
diff --git a/artifacts/evaluation/latest/intent_type_train_report.json b/artifacts/evaluation/latest/intent_type_train_report.json
index ac713faea97a231464bb9f073a89fc3fde69f1be..95149ca22fc9759b238bf770ddab843399d52f5c 100644
--- a/artifacts/evaluation/latest/intent_type_train_report.json
+++ b/artifacts/evaluation/latest/intent_type_train_report.json
@@ -1,19 +1,19 @@
 {
-  "accepted_accuracy": 1.0,
-  "accepted_coverage": 0.9945,
-  "accuracy": 1.0,
+  "accepted_accuracy": 0.9945,
+  "accepted_coverage": 1.0,
+  "accuracy": 0.9945,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_train_confusion_matrix.csv",
   "count": 183,
   "dataset_path": "/content/agentic-intent-classifier/data/train.jsonl",
-  "fallback_rate": 0.0055,
+  "fallback_rate": 0.0,
   "head": "intent_type",
-  "macro_f1": 1.0,
+  "macro_f1": 0.9936,
   "per_class_metrics": {
-    "accuracy": 1.0,
+    "accuracy": 0.994535519125683,
     "ambiguous": {
-      "f1-score": 1.0,
+      "f1-score": 0.9836065573770492,
       "precision": 1.0,
-      "recall": 1.0,
+      "recall": 0.967741935483871,
       "support": 31.0
     },
     "chit_chat": {
@@ -47,9 +47,9 @@
       "support": 38.0
     },
     "macro avg": {
-      "f1-score": 1.0,
-      "precision": 1.0,
-      "recall": 1.0,
+      "f1-score": 0.9935987509758002,
+      "precision": 0.990909090909091,
+      "recall": 0.9967741935483871,
       "support": 183.0
     },
     "personal_reflection": {
@@ -65,8 +65,8 @@
       "support": 5.0
     },
     "support": {
-      "f1-score": 1.0,
-      "precision": 1.0,
+      "f1-score": 0.9523809523809523,
+      "precision": 0.9090909090909091,
       "recall": 1.0,
       "support": 10.0
     },
@@ -77,9 +77,9 @@
       "support": 28.0
     },
     "weighted avg": {
-      "f1-score": 1.0,
-      "precision": 1.0,
-      "recall": 1.0,
+      "f1-score": 0.9946208349863281,
+      "precision": 0.9950322901142573,
+      "recall": 0.994535519125683,
       "support": 183.0
     }
   },
diff --git a/artifacts/evaluation/latest/intent_type_val_confusion_matrix.csv b/artifacts/evaluation/latest/intent_type_val_confusion_matrix.csv
index 15483dd9c56df7dc4959e423e64aea34c7bf95a8..e729131ddbc7ee65939f5606aae08f28a8bb16aa 100644
--- a/artifacts/evaluation/latest/intent_type_val_confusion_matrix.csv
+++ b/artifacts/evaluation/latest/intent_type_val_confusion_matrix.csv
@@ -2,8 +2,8 @@
 informational,8,0,0,0,0,0,0,0,0,0
 exploratory,0,1,0,0,0,0,0,0,0,0
 commercial,0,1,9,0,0,0,0,0,0,0
-transactional,0,0,0,7,0,0,1,0,0,0
-support,0,0,0,0,3,0,0,0,0,0
+transactional,0,0,0,5,0,0,3,0,0,0
+support,0,0,0,0,2,0,0,0,0,1
 personal_reflection,0,0,0,0,0,5,0,0,0,0
 creative_generation,0,0,0,0,0,0,1,0,0,0
 chit_chat,0,0,0,0,0,0,0,1,0,0
diff --git a/artifacts/evaluation/latest/intent_type_val_report.json b/artifacts/evaluation/latest/intent_type_val_report.json
index a24406a8d18fb7daea6251a5ad22a8b1507743cf..9015d7ddf61a7c57db7f8f54178ed8b15f9a6ee0 100644
--- a/artifacts/evaluation/latest/intent_type_val_report.json
+++ b/artifacts/evaluation/latest/intent_type_val_report.json
@@ -1,15 +1,15 @@
 {
-  "accepted_accuracy": 0.9362,
-  "accepted_coverage": 1.0,
-  "accuracy": 0.9362,
+  "accepted_accuracy": 0.8913,
+  "accepted_coverage": 0.9787,
+  "accuracy": 0.8723,
   "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_val_confusion_matrix.csv",
   "count": 47,
   "dataset_path": "/content/agentic-intent-classifier/data/val.jsonl",
-  "fallback_rate": 0.0,
+  "fallback_rate": 0.0213,
   "head": "intent_type",
-  "macro_f1": 0.9108,
+  "macro_f1": 0.8144,
   "per_class_metrics": {
-    "accuracy": 0.9361702127659575,
+    "accuracy": 0.8723404255319149,
     "ambiguous": {
       "f1-score": 0.9411764705882353,
       "precision": 1.0,
@@ -29,8 +29,8 @@
       "support": 10.0
     },
     "creative_generation": {
-      "f1-score": 0.6666666666666666,
-      "precision": 0.5,
+      "f1-score": 0.4,
+      "precision": 0.25,
       "recall": 1.0,
       "support": 1.0
     },
@@ -47,9 +47,9 @@
       "support": 8.0
     },
     "macro avg": {
-      "f1-score": 0.9107843137254902,
-      "precision": 0.89,
-      "recall": 0.966388888888889,
+      "f1-score": 0.8143740573152337,
+      "precision": 0.8150000000000001,
+      "recall": 0.9080555555555556,
       "support": 47.0
     },
     "personal_reflection": {
@@ -59,27 +59,27 @@
       "support": 5.0
     },
     "prohibited": {
-      "f1-score": 1.0,
-      "precision": 1.0,
+      "f1-score": 0.6666666666666666,
+      "precision": 0.5,
       "recall": 1.0,
       "support": 1.0
     },
     "support": {
-      "f1-score": 1.0,
+      "f1-score": 0.8,
       "precision": 1.0,
-      "recall": 1.0,
+      "recall": 0.6666666666666666,
       "support": 3.0
     },
     "transactional": {
-      "f1-score": 0.9333333333333333,
+      "f1-score": 0.7692307692307693,
       "precision": 1.0,
-      "recall": 0.875,
+      "recall": 0.625,
       "support": 8.0
     },
     "weighted avg": {
-      "f1-score": 0.9419274092615769,
-      "precision": 0.9574468085106383,
-      "recall": 0.9361702127659575,
+      "f1-score": 0.8884631430313531,
+      "precision": 0.9414893617021277,
+      "recall": 0.8723404255319149,
       "support": 47.0
     }
   },
diff --git a/artifacts/evaluation/latest/summary.json b/artifacts/evaluation/latest/summary.json
index fb6fb4007d7953cf009ecfa3909939ce5231639e..fc3bdef671cdb58f12083bb6ed562eb2544f5a51 100644
--- a/artifacts/evaluation/latest/summary.json
+++ b/artifacts/evaluation/latest/summary.json
@@ -21,7 +21,7 @@
       "results": [
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Automotive",
             "model_output.classification.iab_content.tier2.label": null
           },
@@ -32,11 +32,6 @@
           },
           "id": "car-buying-maps-to-automotive-buying",
           "mismatches": [
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
             {
               "actual": null,
               "expected": "Auto Type",
@@ -114,8 +109,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -129,6 +124,16 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Software and Applications",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "CRM education should resolve to the closest business/sales path, not generic software.",
@@ -138,9 +143,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Robotics",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -152,12 +157,7 @@
           "id": "crm-comparison-maps-to-sales",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Robotics",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -174,9 +174,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Careers",
-            "model_output.classification.iab_content.tier2.label": "Job Search",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -193,12 +193,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Job Search",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -216,7 +211,7 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-            "model_output.classification.iab_content.tier1.label": "Science"
+            "model_output.classification.iab_content.tier1.label": "Real Estate"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -225,7 +220,7 @@
           "id": "ml-explanation-maps-to-ai",
           "mismatches": [
             {
-              "actual": "Science",
+              "actual": "Real Estate",
               "expected": "Technology & Computing",
               "path": "model_output.classification.iab_content.tier1.label"
             }
@@ -237,10 +232,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Personal Finance",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -251,12 +246,17 @@
           "id": "support-credential-help-maps-to-business-it",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Personal Finance",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Information and Network Security",
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
               "expected": "Internet",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -292,9 +292,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-            "model_output.classification.iab_content.tier2.label": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Sports",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -306,17 +306,12 @@
           "id": "trial-signup-maps-to-software",
           "mismatches": [
             {
-              "actual": "Sensitive Topics",
+              "actual": "Sports",
               "expected": "Hobbies & Interests",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+              "actual": null,
               "expected": "Content Production",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -406,21 +401,21 @@
     "iab_cross_vertical_behavior_lock_regression": {
       "by_status": {
         "must_fix": {
-          "failed": 90,
-          "passed": 0,
+          "failed": 88,
+          "passed": 2,
           "total": 90
         }
       },
       "cases_path": "/content/agentic-intent-classifier/examples/iab_cross_vertical_behavior_lock_cases.json",
       "count": 90,
-      "failed": 90,
-      "passed": 0,
+      "failed": 88,
+      "passed": 2,
       "results": [
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Personal Finance",
-            "model_output.classification.iab_content.tier2.label": "Insurance"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Travel",
+            "model_output.classification.iab_content.tier2.label": "Travel Type"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -428,31 +423,15 @@
             "model_output.classification.iab_content.tier2.label": "Travel Type"
           },
           "id": "auto-buying-easy",
-          "mismatches": [
-            {
-              "actual": "Personal Finance",
-              "expected": "Travel",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Insurance",
-              "expected": "Travel Type",
-              "path": "model_output.classification.iab_content.tier2.label"
-            }
-          ],
+          "mismatches": [],
           "notes": "Cross-vertical easy IAB mapping case for Automotive > Auto Buying and Selling.",
-          "pass": false,
+          "pass": true,
           "status": "must_fix",
           "text": "Which car should I buy for commuting?"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Automotive",
             "model_output.classification.iab_content.tier2.label": "Auto Body Styles"
           },
@@ -462,23 +441,17 @@
             "model_output.classification.iab_content.tier2.label": "Auto Body Styles"
           },
           "id": "auto-buying-medium",
-          "mismatches": [
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            }
-          ],
+          "mismatches": [],
           "notes": "Cross-vertical medium IAB mapping case for Automotive > Auto Buying and Selling.",
-          "pass": false,
+          "pass": true,
           "status": "must_fix",
           "text": "Best used SUV for a family of four"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Automotive",
-            "model_output.classification.iab_content.tier2.label": "Car Culture"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -488,12 +461,7 @@
           "id": "auto-buying-hard",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Car Culture",
+              "actual": null,
               "expected": "Auto Type",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -507,8 +475,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -522,6 +490,16 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Software and Applications",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical easy IAB mapping case for Business and Finance > Business > Sales.",
@@ -531,9 +509,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Robotics",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -545,12 +523,7 @@
           "id": "sales-crm-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Robotics",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -570,7 +543,7 @@
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Business and Finance",
             "model_output.classification.iab_content.tier2.label": "Business",
-            "model_output.classification.iab_content.tier3.label": "Sales"
+            "model_output.classification.iab_content.tier3.label": "Startups"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -584,6 +557,11 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": "Startups",
+              "expected": "Sales",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical hard IAB mapping case for Business and Finance > Business > Sales.",
@@ -593,9 +571,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Careers",
-            "model_output.classification.iab_content.tier2.label": "Job Search",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -607,17 +585,7 @@
           "id": "marketing-tools-easy",
           "mismatches": [
             {
-              "actual": "Careers",
-              "expected": "Technology & Computing",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Job Search",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -634,9 +602,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-            "model_output.classification.iab_content.tier2.label": "Terrorism"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -646,17 +614,12 @@
           "id": "marketing-tools-medium",
           "mismatches": [
             {
-              "actual": "Sensitive Topics",
+              "actual": "Technology & Computing",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Terrorism",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -668,9 +631,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Personal Finance",
-            "model_output.classification.iab_content.tier2.label": "Home Utilities",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -682,17 +645,12 @@
           "id": "marketing-tools-hard",
           "mismatches": [
             {
-              "actual": "Personal Finance",
+              "actual": "Careers",
               "expected": "Technology & Computing",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Home Utilities",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -709,10 +667,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -723,12 +681,17 @@
           "id": "business-it-easy",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Careers",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Information and Network Security",
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
               "expected": "Internet",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -740,9 +703,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Careers",
-            "model_output.classification.iab_content.tier2.label": "Job Search"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Personal Finance",
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -752,9 +715,14 @@
           "id": "business-it-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Personal Finance",
+              "expected": "Careers",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": null,
+              "expected": "Job Search",
+              "path": "model_output.classification.iab_content.tier2.label"
             }
           ],
           "notes": "Cross-vertical medium IAB mapping case for Business and Finance > Business > Business I.T..",
@@ -820,8 +788,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Attractions",
-            "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+            "model_output.classification.iab_content.tier1.label": "Food & Drink",
+            "model_output.classification.iab_content.tier2.label": "Dining Out"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -830,20 +798,10 @@
           },
           "id": "dining-out-medium",
           "mismatches": [
-            {
-              "actual": "Attractions",
-              "expected": "Food & Drink",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
             {
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Bars & Restaurants",
-              "expected": "Dining Out",
-              "path": "model_output.classification.iab_content.tier2.label"
             }
           ],
           "notes": "Cross-vertical medium IAB mapping case for Food & Drink > Dining Out.",
@@ -854,8 +812,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Attractions",
-            "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+            "model_output.classification.iab_content.tier1.label": "Food & Drink",
+            "model_output.classification.iab_content.tier2.label": "Dining Out"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -864,20 +822,10 @@
           },
           "id": "dining-out-hard",
           "mismatches": [
-            {
-              "actual": "Attractions",
-              "expected": "Food & Drink",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
             {
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Bars & Restaurants",
-              "expected": "Dining Out",
-              "path": "model_output.classification.iab_content.tier2.label"
             }
           ],
           "notes": "Cross-vertical hard IAB mapping case for Food & Drink > Dining Out.",
@@ -954,7 +902,7 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-            "model_output.classification.iab_content.tier1.label": "Science"
+            "model_output.classification.iab_content.tier1.label": "Real Estate"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -963,7 +911,7 @@
           "id": "artificial-intelligence-easy",
           "mismatches": [
             {
-              "actual": "Science",
+              "actual": "Real Estate",
               "expected": "Technology & Computing",
               "path": "model_output.classification.iab_content.tier1.label"
             }
@@ -975,9 +923,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Education",
-            "model_output.classification.iab_content.tier2.label": "Language Learning"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -987,11 +935,11 @@
           "id": "artificial-intelligence-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            }
-          ],
+              "actual": null,
+              "expected": "Language Learning",
+              "path": "model_output.classification.iab_content.tier2.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Artificial Intelligence.",
           "pass": false,
           "status": "must_fix",
@@ -1024,8 +972,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing"
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": "Job Search"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1035,7 +983,7 @@
           "id": "software-apps-easy",
           "mismatches": [
             {
-              "actual": "Technology & Computing",
+              "actual": "Careers",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
@@ -1045,7 +993,7 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Computing",
+              "actual": "Job Search",
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -1059,8 +1007,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null,
             "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
@@ -1078,7 +1026,12 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Software and Applications",
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
               "expected": "Internet",
               "path": "model_output.classification.iab_content.tier3.label"
             },
@@ -1096,8 +1049,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": "Job Search",
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -1108,13 +1061,18 @@
           },
           "id": "software-apps-hard",
           "mismatches": [
+            {
+              "actual": "Careers",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
             {
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Virtual Reality",
+              "actual": "Job Search",
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -1179,10 +1137,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Information and Network Security",
+            "model_output.classification.iab_content.tier3.label": null,
             "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
@@ -1195,12 +1153,7 @@
           "id": "communication-software-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Information and Network Security",
+              "actual": null,
               "expected": "Software and Applications",
               "path": "model_output.classification.iab_content.tier3.label"
             },
@@ -1217,9 +1170,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null,
             "model_output.classification.iab_content.tier4.label": null
           },
@@ -1233,12 +1186,12 @@
           "id": "communication-software-hard",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Careers",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Virtual Reality",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -1263,8 +1216,8 @@
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Internet",
-            "model_output.classification.iab_content.tier4.label": "Web Hosting"
+            "model_output.classification.iab_content.tier3.label": "Data Storage and Warehousing",
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1279,6 +1232,16 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": "Data Storage and Warehousing",
+              "expected": "Internet",
+              "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": null,
+              "expected": "Web Hosting",
+              "path": "model_output.classification.iab_content.tier4.label"
             }
           ],
           "notes": "Cross-vertical easy IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
@@ -1288,11 +1251,11 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Internet",
-            "model_output.classification.iab_content.tier4.label": "Web Hosting"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null,
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1304,9 +1267,19 @@
           "id": "web-hosting-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Internet",
+              "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": null,
+              "expected": "Web Hosting",
+              "path": "model_output.classification.iab_content.tier4.label"
             }
           ],
           "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
@@ -1316,11 +1289,11 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Internet",
-            "model_output.classification.iab_content.tier4.label": "Web Hosting"
+            "model_output.classification.iab_content.tier3.label": null,
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1332,9 +1305,14 @@
           "id": "web-hosting-hard",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": null,
+              "expected": "Internet",
+              "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": null,
+              "expected": "Web Hosting",
+              "path": "model_output.classification.iab_content.tier4.label"
             }
           ],
           "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
@@ -1398,8 +1376,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Laptops"
+            "model_output.classification.iab_content.tier2.label": "Consumer Electronics",
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1415,12 +1393,7 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Computing",
-              "expected": "Consumer Electronics",
-              "path": "model_output.classification.iab_content.tier2.label"
-            },
-            {
-              "actual": "Laptops",
+              "actual": null,
               "expected": "Smartphones",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -1432,11 +1405,11 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications",
-            "model_output.classification.iab_content.tier4.label": "Computer Animation"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null,
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1448,12 +1421,17 @@
           "id": "desktops-easy",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Software and Applications",
+              "path": "model_output.classification.iab_content.tier3.label"
             },
             {
-              "actual": "Computer Animation",
+              "actual": null,
               "expected": "Photo Editing Software",
               "path": "model_output.classification.iab_content.tier4.label"
             }
@@ -1465,10 +1443,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Desktops"
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1479,9 +1457,9 @@
           "id": "desktops-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": null,
+              "expected": "Desktops",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Desktops.",
@@ -1491,10 +1469,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Desktops"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1505,9 +1483,14 @@
           "id": "desktops-hard",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Desktops",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Desktops.",
@@ -1593,10 +1576,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Men's Fashion",
-            "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Shopping",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1607,17 +1590,17 @@
           "id": "style-fashion-parent-easy",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Shopping",
+              "expected": "Style & Fashion",
+              "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Men's Fashion",
+              "actual": null,
               "expected": "Women's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": "Men's Shoes and Footwear",
+              "actual": null,
               "expected": "Women's Shoes and Footwear",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -1691,9 +1674,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sports",
-            "model_output.classification.iab_content.tier2.label": "Bodybuilding",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -1705,17 +1688,7 @@
           "id": "womens-shoes-easy",
           "mismatches": [
             {
-              "actual": "Sports",
-              "expected": "Style & Fashion",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Bodybuilding",
+              "actual": null,
               "expected": "Women's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -1732,9 +1705,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sports",
-            "model_output.classification.iab_content.tier2.label": "Walking",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -1746,17 +1719,7 @@
           "id": "womens-shoes-medium",
           "mismatches": [
             {
-              "actual": "Sports",
-              "expected": "Style & Fashion",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Walking",
+              "actual": null,
               "expected": "Women's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -1801,8 +1764,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-            "model_output.classification.iab_content.tier3.label": null
+            "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+            "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1818,12 +1781,12 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Children's Clothing",
+              "actual": "Women's Fashion",
               "expected": "Men's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": null,
+              "actual": "Women's Shoes and Footwear",
               "expected": "Men's Clothing",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -1837,8 +1800,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Men's Fashion",
-            "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
+            "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+            "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1854,7 +1817,12 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Men's Shoes and Footwear",
+              "actual": "Women's Fashion",
+              "expected": "Men's Fashion",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": "Women's Shoes and Footwear",
               "expected": "Men's Clothing",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -1868,8 +1836,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-            "model_output.classification.iab_content.tier3.label": null
+            "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+            "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1885,12 +1853,12 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Children's Clothing",
+              "actual": "Women's Fashion",
               "expected": "Men's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": null,
+              "actual": "Women's Shoes and Footwear",
               "expected": "Men's Shoes and Footwear",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -1936,10 +1904,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Travel",
-            "model_output.classification.iab_content.tier2.label": "Travel Type",
-            "model_output.classification.iab_content.tier3.label": "Hotels and Motels"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1950,9 +1918,14 @@
           "id": "hotels-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": null,
+              "expected": "Travel Type",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Hotels and Motels",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical medium IAB mapping case for Travel > Travel Type > Hotels and Motels.",
@@ -1964,7 +1937,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Travel",
-            "model_output.classification.iab_content.tier2.label": "Travel Type"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -1977,6 +1950,11 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Travel Type",
+              "path": "model_output.classification.iab_content.tier2.label"
             }
           ],
           "notes": "Cross-vertical hard IAB mapping case for Travel > Travel Type > Hotels and Motels.",
@@ -2077,8 +2055,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Healthy Living",
-            "model_output.classification.iab_content.tier2.label": "Fitness and Exercise",
-            "model_output.classification.iab_content.tier3.label": "Running and Jogging"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2099,12 +2077,12 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Fitness and Exercise",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": "Running and Jogging",
+              "actual": null,
               "expected": "Green Solutions",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -2118,7 +2096,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Sports",
-            "model_output.classification.iab_content.tier2.label": "Walking",
+            "model_output.classification.iab_content.tier2.label": "Bodybuilding",
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -2140,7 +2118,7 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Walking",
+              "actual": "Bodybuilding",
               "expected": "Fitness and Exercise",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -2302,9 +2280,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Travel",
-            "model_output.classification.iab_content.tier2.label": "Travel Type"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2319,12 +2297,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Travel Type",
+              "actual": null,
               "expected": "Fiction",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -2337,7 +2310,7 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Books and Literature"
+            "model_output.classification.iab_content.tier1.label": "Genres"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2345,6 +2318,11 @@
           },
           "id": "fiction-hard",
           "mismatches": [
+            {
+              "actual": "Genres",
+              "expected": "Books and Literature",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
             {
               "actual": "exact",
               "expected": "nearest_equivalent",
@@ -2360,7 +2338,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Home & Garden",
-            "model_output.classification.iab_content.tier2.label": "Interior Decorating"
+            "model_output.classification.iab_content.tier2.label": "Remodeling & Construction"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2373,11 +2351,6 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Interior Decorating",
-              "expected": "Remodeling & Construction",
-              "path": "model_output.classification.iab_content.tier2.label"
             }
           ],
           "notes": "Cross-vertical easy IAB mapping case for Home & Garden > Home Improvement.",
@@ -2388,9 +2361,9 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Home & Garden",
-            "model_output.classification.iab_content.tier2.label": "Interior Decorating",
-            "model_output.classification.iab_content.tier3.label": null
+            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+            "model_output.classification.iab_content.tier2.label": "Personal Care",
+            "model_output.classification.iab_content.tier3.label": "Bath and Shower"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2400,25 +2373,10 @@
           },
           "id": "home-improvement-medium",
           "mismatches": [
-            {
-              "actual": "Home & Garden",
-              "expected": "Style & Fashion",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
             {
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Interior Decorating",
-              "expected": "Personal Care",
-              "path": "model_output.classification.iab_content.tier2.label"
-            },
-            {
-              "actual": null,
-              "expected": "Bath and Shower",
-              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical medium IAB mapping case for Home & Garden > Home Improvement.",
@@ -2462,9 +2420,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Augmented Reality"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2474,17 +2432,12 @@
           "id": "online-education-easy",
           "mismatches": [
             {
-              "actual": "Technology & Computing",
+              "actual": "Careers",
               "expected": "Education",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Augmented Reality",
+              "actual": null,
               "expected": "Language Learning",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -2623,10 +2576,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Medical Health",
-            "model_output.classification.iab_content.tier2.label": "Diseases and Conditions",
-            "model_output.classification.iab_content.tier3.label": "Allergies"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Food & Drink",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2637,9 +2590,19 @@
           "id": "medical-health-easy",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Food & Drink",
+              "expected": "Medical Health",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": null,
+              "expected": "Diseases and Conditions",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Allergies",
+              "path": "model_output.classification.iab_content.tier3.label"
             }
           ],
           "notes": "Cross-vertical easy IAB mapping case for Medical Health.",
@@ -2649,10 +2612,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Medical Health",
             "model_output.classification.iab_content.tier2.label": "Diseases and Conditions",
-            "model_output.classification.iab_content.tier3.label": "Bone and Joint Conditions",
+            "model_output.classification.iab_content.tier3.label": null,
             "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
@@ -2665,12 +2628,7 @@
           "id": "medical-health-medium",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Bone and Joint Conditions",
+              "actual": null,
               "expected": "Injuries",
               "path": "model_output.classification.iab_content.tier3.label"
             },
@@ -2689,7 +2647,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Medical Health",
-            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier2.label": "Surgery",
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -2711,7 +2669,7 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": null,
+              "actual": "Surgery",
               "expected": "Wellness",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -2812,9 +2770,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Holidays",
-            "model_output.classification.iab_content.tier2.label": "National & Civic Holidays"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2829,12 +2787,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "National & Civic Holidays",
+              "actual": null,
               "expected": "Food Movements",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -2935,9 +2888,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Genres",
-            "model_output.classification.iab_content.tier2.label": "Family/Children"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Shopping",
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2947,17 +2900,12 @@
           "id": "parenting-medium",
           "mismatches": [
             {
-              "actual": "Genres",
+              "actual": "Shopping",
               "expected": "Family and Relationships",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Family/Children",
+              "actual": null,
               "expected": "Parenting",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -2972,7 +2920,7 @@
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Family and Relationships",
             "model_output.classification.iab_content.tier2.label": "Parenting",
-            "model_output.classification.iab_content.tier3.label": null
+            "model_output.classification.iab_content.tier3.label": "Parenting Babies and Toddlers"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -2988,7 +2936,7 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": null,
+              "actual": "Parenting Babies and Toddlers",
               "expected": "Special Needs Kids",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -3070,7 +3018,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Entertainment",
-            "model_output.classification.iab_content.tier2.label": "Movies"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -3083,6 +3031,11 @@
               "actual": "exact",
               "expected": "nearest_equivalent",
               "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Movies",
+              "path": "model_output.classification.iab_content.tier2.label"
             }
           ],
           "notes": "Cross-vertical easy IAB mapping case for Entertainment > Movies.",
@@ -3093,8 +3046,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Entertainment",
-            "model_output.classification.iab_content.tier2.label": "Movies",
+            "model_output.classification.iab_content.tier1.label": "Genres",
+            "model_output.classification.iab_content.tier2.label": "Horror",
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3106,7 +3059,7 @@
           "id": "movies-medium",
           "mismatches": [
             {
-              "actual": "Entertainment",
+              "actual": "Genres",
               "expected": "Video Gaming",
               "path": "model_output.classification.iab_content.tier1.label"
             },
@@ -3116,7 +3069,7 @@
               "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Movies",
+              "actual": "Horror",
               "expected": "Video Game Genres",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3172,21 +3125,21 @@
     "iab_cross_vertical_quality_target_eval": {
       "by_status": {
         "must_fix": {
-          "failed": 49,
-          "passed": 41,
+          "failed": 64,
+          "passed": 26,
           "total": 90
         }
       },
       "cases_path": "/content/agentic-intent-classifier/examples/iab_cross_vertical_mapping_cases.json",
       "count": 90,
-      "failed": 49,
-      "passed": 41,
+      "failed": 64,
+      "passed": 26,
       "results": [
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Personal Finance",
-            "model_output.classification.iab_content.tier2.label": "Insurance"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Travel",
+            "model_output.classification.iab_content.tier2.label": "Travel Type"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -3196,17 +3149,12 @@
           "id": "auto-buying-easy",
           "mismatches": [
             {
-              "actual": "Personal Finance",
+              "actual": "Travel",
               "expected": "Automotive",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Insurance",
+              "actual": "Travel Type",
               "expected": "Auto Buying and Selling",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -3218,7 +3166,7 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Automotive",
             "model_output.classification.iab_content.tier2.label": "Auto Body Styles"
           },
@@ -3229,11 +3177,6 @@
           },
           "id": "auto-buying-medium",
           "mismatches": [
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
             {
               "actual": "Auto Body Styles",
               "expected": "Auto Buying and Selling",
@@ -3247,9 +3190,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Automotive",
-            "model_output.classification.iab_content.tier2.label": "Car Culture"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -3259,12 +3202,7 @@
           "id": "auto-buying-hard",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Car Culture",
+              "actual": null,
               "expected": "Auto Buying and Selling",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -3278,8 +3216,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3295,12 +3233,12 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Computing",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": "Software and Applications",
+              "actual": null,
               "expected": "Sales",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -3312,9 +3250,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Robotics",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3331,7 +3269,12 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Robotics",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3351,7 +3294,7 @@
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Business and Finance",
             "model_output.classification.iab_content.tier2.label": "Business",
-            "model_output.classification.iab_content.tier3.label": "Sales"
+            "model_output.classification.iab_content.tier3.label": "Startups"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3360,17 +3303,23 @@
             "model_output.classification.iab_content.tier3.label": "Sales"
           },
           "id": "sales-crm-hard",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Startups",
+              "expected": "Sales",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical hard IAB mapping case for Business and Finance > Business > Sales.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Need software to manage leads and pipeline for a startup sales team"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Careers",
-            "model_output.classification.iab_content.tier2.label": "Job Search",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3382,12 +3331,17 @@
           "id": "marketing-tools-easy",
           "mismatches": [
             {
-              "actual": "Careers",
+              "actual": "Technology & Computing",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Job Search",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3404,9 +3358,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-            "model_output.classification.iab_content.tier2.label": "Terrorism",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3418,12 +3372,17 @@
           "id": "marketing-tools-medium",
           "mismatches": [
             {
-              "actual": "Sensitive Topics",
+              "actual": "Technology & Computing",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Terrorism",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3440,9 +3399,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Personal Finance",
-            "model_output.classification.iab_content.tier2.label": "Home Utilities",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3454,12 +3413,17 @@
           "id": "marketing-tools-hard",
           "mismatches": [
             {
-              "actual": "Personal Finance",
+              "actual": "Careers",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Home Utilities",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3476,10 +3440,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3490,17 +3454,22 @@
           "id": "business-it-easy",
           "mismatches": [
             {
-              "actual": "Technology & Computing",
+              "actual": "Careers",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Computing",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": "Information and Network Security",
+              "actual": null,
               "expected": "Business I.T.",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -3512,9 +3481,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Careers",
-            "model_output.classification.iab_content.tier2.label": "Job Search",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Personal Finance",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3526,12 +3495,17 @@
           "id": "business-it-medium",
           "mismatches": [
             {
-              "actual": "Careers",
+              "actual": "Personal Finance",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Job Search",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3603,8 +3577,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Attractions",
-            "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+            "model_output.classification.iab_content.tier1.label": "Food & Drink",
+            "model_output.classification.iab_content.tier2.label": "Dining Out"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3612,28 +3586,17 @@
             "model_output.classification.iab_content.tier2.label": "Dining Out"
           },
           "id": "dining-out-medium",
-          "mismatches": [
-            {
-              "actual": "Attractions",
-              "expected": "Food & Drink",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
-            {
-              "actual": "Bars & Restaurants",
-              "expected": "Dining Out",
-              "path": "model_output.classification.iab_content.tier2.label"
-            }
-          ],
+          "mismatches": [],
           "notes": "Cross-vertical medium IAB mapping case for Food & Drink > Dining Out.",
-          "pass": false,
+          "pass": true,
           "status": "must_fix",
           "text": "Good restaurants for a client dinner downtown"
         },
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Attractions",
-            "model_output.classification.iab_content.tier2.label": "Bars & Restaurants"
+            "model_output.classification.iab_content.tier1.label": "Food & Drink",
+            "model_output.classification.iab_content.tier2.label": "Dining Out"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3641,20 +3604,9 @@
             "model_output.classification.iab_content.tier2.label": "Dining Out"
           },
           "id": "dining-out-hard",
-          "mismatches": [
-            {
-              "actual": "Attractions",
-              "expected": "Food & Drink",
-              "path": "model_output.classification.iab_content.tier1.label"
-            },
-            {
-              "actual": "Bars & Restaurants",
-              "expected": "Dining Out",
-              "path": "model_output.classification.iab_content.tier2.label"
-            }
-          ],
+          "mismatches": [],
           "notes": "Cross-vertical hard IAB mapping case for Food & Drink > Dining Out.",
-          "pass": false,
+          "pass": true,
           "status": "must_fix",
           "text": "Need a place to eat tonight where I can make a reservation online"
         },
@@ -3721,7 +3673,7 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-            "model_output.classification.iab_content.tier1.label": "Science",
+            "model_output.classification.iab_content.tier1.label": "Real Estate",
             "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
@@ -3732,7 +3684,7 @@
           "id": "artificial-intelligence-easy",
           "mismatches": [
             {
-              "actual": "Science",
+              "actual": "Real Estate",
               "expected": "Technology & Computing",
               "path": "model_output.classification.iab_content.tier1.label"
             },
@@ -3754,9 +3706,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Education",
-            "model_output.classification.iab_content.tier2.label": "Language Learning"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3771,7 +3723,12 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Language Learning",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Artificial Intelligence",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -3813,9 +3770,9 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": "Job Search",
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3824,9 +3781,25 @@
             "model_output.classification.iab_content.tier3.label": "Software and Applications"
           },
           "id": "software-apps-easy",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Careers",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "Job Search",
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Software and Applications",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical easy IAB mapping case for Technology & Computing > Computing > Computer Software and Applications.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Best workflow software for a small operations team"
         },
@@ -3834,8 +3807,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -3844,17 +3817,28 @@
             "model_output.classification.iab_content.tier3.label": "Software and Applications"
           },
           "id": "software-apps-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Software and Applications",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Computer Software and Applications.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Need project management software for a distributed team"
         },
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": "Job Search",
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -3866,7 +3850,12 @@
           "id": "software-apps-hard",
           "mismatches": [
             {
-              "actual": "Virtual Reality",
+              "actual": "Careers",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "Job Search",
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -3926,10 +3915,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Information and Network Security",
+            "model_output.classification.iab_content.tier3.label": null,
             "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
@@ -3942,7 +3931,12 @@
           "id": "communication-software-medium",
           "mismatches": [
             {
-              "actual": "Information and Network Security",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Software and Applications",
               "path": "model_output.classification.iab_content.tier3.label"
             },
@@ -3959,9 +3953,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Virtual Reality",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null,
             "model_output.classification.iab_content.tier4.label": null
           },
@@ -3975,7 +3969,17 @@
           "id": "communication-software-hard",
           "mismatches": [
             {
-              "actual": "Virtual Reality",
+              "actual": "Careers",
+              "expected": "Technology & Computing",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -4000,8 +4004,8 @@
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Internet",
-            "model_output.classification.iab_content.tier4.label": "Web Hosting"
+            "model_output.classification.iab_content.tier3.label": "Data Storage and Warehousing",
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4011,19 +4015,30 @@
             "model_output.classification.iab_content.tier4.label": "Web Hosting"
           },
           "id": "web-hosting-easy",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Data Storage and Warehousing",
+              "expected": "Internet",
+              "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": null,
+              "expected": "Web Hosting",
+              "path": "model_output.classification.iab_content.tier4.label"
+            }
+          ],
           "notes": "Cross-vertical easy IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Vercel vs Netlify for website hosting"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Internet",
-            "model_output.classification.iab_content.tier4.label": "Web Hosting"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null,
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4033,19 +4048,40 @@
             "model_output.classification.iab_content.tier4.label": "Web Hosting"
           },
           "id": "web-hosting-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Internet",
+              "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": null,
+              "expected": "Web Hosting",
+              "path": "model_output.classification.iab_content.tier4.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Best hosting platform for a startup website"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Internet",
-            "model_output.classification.iab_content.tier4.label": "Web Hosting"
+            "model_output.classification.iab_content.tier3.label": null,
+            "model_output.classification.iab_content.tier4.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4055,9 +4091,25 @@
             "model_output.classification.iab_content.tier4.label": "Web Hosting"
           },
           "id": "web-hosting-hard",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Internet",
+              "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": null,
+              "expected": "Web Hosting",
+              "path": "model_output.classification.iab_content.tier4.label"
+            }
+          ],
           "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Internet > Web Hosting.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Need a managed hosting provider to deploy and run our marketing site"
         },
@@ -4105,8 +4157,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Laptops"
+            "model_output.classification.iab_content.tier2.label": "Consumer Electronics",
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4115,18 +4167,29 @@
             "model_output.classification.iab_content.tier3.label": "Laptops"
           },
           "id": "laptops-hard",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Consumer Electronics",
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Laptops",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Laptops.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Need a portable computer with good battery life for everyday work"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4137,7 +4200,17 @@
           "id": "desktops-easy",
           "mismatches": [
             {
-              "actual": "Software and Applications",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
               "expected": "Desktops",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -4149,10 +4222,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
             "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Desktops"
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4161,18 +4234,29 @@
             "model_output.classification.iab_content.tier3.label": "Desktops"
           },
           "id": "desktops-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Desktops",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Technology & Computing > Computing > Desktops.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Which desktop computer should I buy for a home office?"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Desktops"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4181,9 +4265,25 @@
             "model_output.classification.iab_content.tier3.label": "Desktops"
           },
           "id": "desktops-hard",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Computing",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Desktops",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical hard IAB mapping case for Technology & Computing > Computing > Desktops.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Need a desktop PC with strong performance for creative work"
         },
@@ -4249,8 +4349,8 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Style & Fashion"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Shopping"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -4259,9 +4359,9 @@
           "id": "style-fashion-parent-easy",
           "mismatches": [
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
+              "actual": "Shopping",
+              "expected": "Style & Fashion",
+              "path": "model_output.classification.iab_content.tier1.label"
             }
           ],
           "notes": "Cross-vertical easy IAB mapping case for Style & Fashion.",
@@ -4315,9 +4415,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sports",
-            "model_output.classification.iab_content.tier2.label": "Bodybuilding",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -4329,12 +4429,12 @@
           "id": "womens-shoes-easy",
           "mismatches": [
             {
-              "actual": "Sports",
-              "expected": "Style & Fashion",
-              "path": "model_output.classification.iab_content.tier1.label"
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Bodybuilding",
+              "actual": null,
               "expected": "Women's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -4351,9 +4451,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sports",
-            "model_output.classification.iab_content.tier2.label": "Walking",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -4365,12 +4465,12 @@
           "id": "womens-shoes-medium",
           "mismatches": [
             {
-              "actual": "Sports",
-              "expected": "Style & Fashion",
-              "path": "model_output.classification.iab_content.tier1.label"
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
             },
             {
-              "actual": "Walking",
+              "actual": null,
               "expected": "Women's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -4409,8 +4509,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-            "model_output.classification.iab_content.tier3.label": null
+            "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+            "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4421,12 +4521,12 @@
           "id": "mens-shoes-easy",
           "mismatches": [
             {
-              "actual": "Children's Clothing",
+              "actual": "Women's Fashion",
               "expected": "Men's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": null,
+              "actual": "Women's Shoes and Footwear",
               "expected": "Men's Shoes and Footwear",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -4440,8 +4540,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Men's Fashion",
-            "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
+            "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+            "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4450,9 +4550,20 @@
             "model_output.classification.iab_content.tier3.label": "Men's Shoes and Footwear"
           },
           "id": "mens-shoes-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Women's Fashion",
+              "expected": "Men's Fashion",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": "Women's Shoes and Footwear",
+              "expected": "Men's Shoes and Footwear",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Style & Fashion > Men's Fashion > Men's Shoes and Footwear.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Good men's dress shoes for office use"
         },
@@ -4460,8 +4571,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Style & Fashion",
-            "model_output.classification.iab_content.tier2.label": "Children's Clothing",
-            "model_output.classification.iab_content.tier3.label": null
+            "model_output.classification.iab_content.tier2.label": "Women's Fashion",
+            "model_output.classification.iab_content.tier3.label": "Women's Shoes and Footwear"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4472,12 +4583,12 @@
           "id": "mens-shoes-hard",
           "mismatches": [
             {
-              "actual": "Children's Clothing",
+              "actual": "Women's Fashion",
               "expected": "Men's Fashion",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": null,
+              "actual": "Women's Shoes and Footwear",
               "expected": "Men's Shoes and Footwear",
               "path": "model_output.classification.iab_content.tier3.label"
             }
@@ -4509,10 +4620,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Travel",
-            "model_output.classification.iab_content.tier2.label": "Travel Type",
-            "model_output.classification.iab_content.tier3.label": "Hotels and Motels"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4521,9 +4632,25 @@
             "model_output.classification.iab_content.tier3.label": "Hotels and Motels"
           },
           "id": "hotels-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
+              "expected": "Travel Type",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Hotels and Motels",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Travel > Travel Type > Hotels and Motels.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Best hotels near Times Square for a weekend trip"
         },
@@ -4531,7 +4658,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Travel",
-            "model_output.classification.iab_content.tier2.label": "Travel Type",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -4542,6 +4669,11 @@
           },
           "id": "hotels-hard",
           "mismatches": [
+            {
+              "actual": null,
+              "expected": "Travel Type",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
             {
               "actual": null,
               "expected": "Hotels and Motels",
@@ -4628,8 +4760,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Healthy Living",
-            "model_output.classification.iab_content.tier2.label": "Fitness and Exercise",
-            "model_output.classification.iab_content.tier3.label": "Running and Jogging"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4638,9 +4770,20 @@
             "model_output.classification.iab_content.tier3.label": "Running and Jogging"
           },
           "id": "running-and-jogging-easy",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": null,
+              "expected": "Fitness and Exercise",
+              "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": null,
+              "expected": "Running and Jogging",
+              "path": "model_output.classification.iab_content.tier3.label"
+            }
+          ],
           "notes": "Cross-vertical easy IAB mapping case for Healthy Living > Fitness and Exercise > Running and Jogging.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Best running plan for a first 10k"
         },
@@ -4648,7 +4791,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Sports",
-            "model_output.classification.iab_content.tier2.label": "Walking",
+            "model_output.classification.iab_content.tier2.label": "Bodybuilding",
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -4665,7 +4808,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Walking",
+              "actual": "Bodybuilding",
               "expected": "Fitness and Exercise",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -4801,9 +4944,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Travel",
-            "model_output.classification.iab_content.tier2.label": "Travel Type"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -4818,12 +4961,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
-            },
-            {
-              "actual": "Travel Type",
+              "actual": null,
               "expected": "Fiction",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -4836,8 +4974,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Books and Literature",
-            "model_output.classification.iab_content.tier2.label": "Fiction"
+            "model_output.classification.iab_content.tier1.label": "Genres",
+            "model_output.classification.iab_content.tier2.label": "Romance"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4845,9 +4983,20 @@
             "model_output.classification.iab_content.tier2.label": "Fiction"
           },
           "id": "fiction-hard",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Genres",
+              "expected": "Books and Literature",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "Romance",
+              "expected": "Fiction",
+              "path": "model_output.classification.iab_content.tier2.label"
+            }
+          ],
           "notes": "Cross-vertical hard IAB mapping case for Books and Literature > Fiction.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Looking for a character-driven novel, not comics or poetry"
         },
@@ -4855,7 +5004,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Home & Garden",
-            "model_output.classification.iab_content.tier2.label": "Interior Decorating"
+            "model_output.classification.iab_content.tier2.label": "Remodeling & Construction"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4865,7 +5014,7 @@
           "id": "home-improvement-easy",
           "mismatches": [
             {
-              "actual": "Interior Decorating",
+              "actual": "Remodeling & Construction",
               "expected": "Home Improvement",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -4878,8 +5027,8 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Home & Garden",
-            "model_output.classification.iab_content.tier2.label": "Interior Decorating"
+            "model_output.classification.iab_content.tier1.label": "Style & Fashion",
+            "model_output.classification.iab_content.tier2.label": "Personal Care"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4889,7 +5038,12 @@
           "id": "home-improvement-medium",
           "mismatches": [
             {
-              "actual": "Interior Decorating",
+              "actual": "Style & Fashion",
+              "expected": "Home & Garden",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "Personal Care",
               "expected": "Home Improvement",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -4930,9 +5084,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Augmented Reality"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Careers",
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -4942,12 +5096,17 @@
           "id": "online-education-easy",
           "mismatches": [
             {
-              "actual": "Technology & Computing",
+              "actual": "Careers",
               "expected": "Education",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Augmented Reality",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Online Education",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -5077,23 +5236,34 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Medical Health"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Food & Drink"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Medical Health"
           },
           "id": "medical-health-easy",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Food & Drink",
+              "expected": "Medical Health",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            }
+          ],
           "notes": "Cross-vertical easy IAB mapping case for Medical Health.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "what do these allergy symptoms mean"
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Medical Health"
           },
           "expected": {
@@ -5101,9 +5271,15 @@
             "model_output.classification.iab_content.tier1.label": "Medical Health"
           },
           "id": "medical-health-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Medical Health.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "when should i see a doctor for persistent knee pain"
         },
@@ -5207,9 +5383,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Holidays",
-            "model_output.classification.iab_content.tier2.label": "National & Civic Holidays"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -5224,7 +5400,12 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "National & Civic Holidays",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Financial Planning",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -5302,9 +5483,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Genres",
-            "model_output.classification.iab_content.tier2.label": "Family/Children"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Shopping",
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -5314,12 +5495,17 @@
           "id": "parenting-medium",
           "mismatches": [
             {
-              "actual": "Genres",
+              "actual": "Shopping",
               "expected": "Family and Relationships",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Family/Children",
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
+            },
+            {
+              "actual": null,
               "expected": "Parenting",
               "path": "model_output.classification.iab_content.tier2.label"
             }
@@ -5405,7 +5591,7 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Entertainment",
-            "model_output.classification.iab_content.tier2.label": "Movies"
+            "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -5413,17 +5599,23 @@
             "model_output.classification.iab_content.tier2.label": "Movies"
           },
           "id": "movies-easy",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": null,
+              "expected": "Movies",
+              "path": "model_output.classification.iab_content.tier2.label"
+            }
+          ],
           "notes": "Cross-vertical easy IAB mapping case for Entertainment > Movies.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "What movie should we watch tonight?"
         },
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Entertainment",
-            "model_output.classification.iab_content.tier2.label": "Movies"
+            "model_output.classification.iab_content.tier1.label": "Genres",
+            "model_output.classification.iab_content.tier2.label": "Horror"
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "exact",
@@ -5431,9 +5623,20 @@
             "model_output.classification.iab_content.tier2.label": "Movies"
           },
           "id": "movies-medium",
-          "mismatches": [],
+          "mismatches": [
+            {
+              "actual": "Genres",
+              "expected": "Entertainment",
+              "path": "model_output.classification.iab_content.tier1.label"
+            },
+            {
+              "actual": "Horror",
+              "expected": "Movies",
+              "path": "model_output.classification.iab_content.tier2.label"
+            }
+          ],
           "notes": "Cross-vertical medium IAB mapping case for Entertainment > Movies.",
-          "pass": true,
+          "pass": false,
           "status": "must_fix",
           "text": "Best thriller movies from the last few years"
         },
@@ -5472,7 +5675,7 @@
       "results": [
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Automotive",
             "model_output.classification.iab_content.tier2.label": null
           },
@@ -5487,6 +5690,11 @@
               "actual": null,
               "expected": "Auto Buying and Selling",
               "path": "model_output.classification.iab_content.tier2.label"
+            },
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
             }
           ],
           "notes": "Vehicle shopping queries should map into the automotive buying branch, not business sales.",
@@ -5549,8 +5757,8 @@
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "exact",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Software and Applications"
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -5566,12 +5774,12 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Computing",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": "Software and Applications",
+              "actual": null,
               "expected": "Sales",
               "path": "model_output.classification.iab_content.tier3.label"
             },
@@ -5588,9 +5796,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Robotics",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -5607,7 +5815,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Robotics",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -5615,6 +5823,11 @@
               "actual": null,
               "expected": "Sales",
               "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
             }
           ],
           "notes": "Direct CRM vendor comparison should map cleanly into the sales domain.",
@@ -5624,9 +5837,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
             "model_output.classification.iab_content.tier1.label": "Careers",
-            "model_output.classification.iab_content.tier2.label": "Job Search",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -5643,7 +5856,7 @@
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Job Search",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -5651,6 +5864,11 @@
               "actual": null,
               "expected": "Marketing and Advertising",
               "path": "model_output.classification.iab_content.tier3.label"
+            },
+            {
+              "actual": "nearest_equivalent",
+              "expected": "exact",
+              "path": "model_output.classification.iab_content.mapping_mode"
             }
           ],
           "notes": "Marketing tool discovery should map to the marketing and advertising branch.",
@@ -5661,7 +5879,7 @@
         {
           "actual": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
-            "model_output.classification.iab_content.tier1.label": "Science",
+            "model_output.classification.iab_content.tier1.label": "Real Estate",
             "model_output.classification.iab_content.tier2.label": null
           },
           "expected": {
@@ -5672,7 +5890,7 @@
           "id": "ml-explanation-maps-to-ai",
           "mismatches": [
             {
-              "actual": "Science",
+              "actual": "Real Estate",
               "expected": "Technology & Computing",
               "path": "model_output.classification.iab_content.tier1.label"
             },
@@ -5694,10 +5912,10 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Technology & Computing",
-            "model_output.classification.iab_content.tier2.label": "Computing",
-            "model_output.classification.iab_content.tier3.label": "Information and Network Security"
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Personal Finance",
+            "model_output.classification.iab_content.tier2.label": null,
+            "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
             "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
@@ -5708,24 +5926,19 @@
           "id": "support-credential-help-maps-to-business-it",
           "mismatches": [
             {
-              "actual": "Technology & Computing",
+              "actual": "Personal Finance",
               "expected": "Business and Finance",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Computing",
+              "actual": null,
               "expected": "Business",
               "path": "model_output.classification.iab_content.tier2.label"
             },
             {
-              "actual": "Information and Network Security",
+              "actual": null,
               "expected": "Business I.T.",
               "path": "model_output.classification.iab_content.tier3.label"
-            },
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
             }
           ],
           "notes": "Credential and account help should map to business IT rather than generic business.",
@@ -5753,9 +5966,9 @@
         },
         {
           "actual": {
-            "model_output.classification.iab_content.mapping_mode": "exact",
-            "model_output.classification.iab_content.tier1.label": "Sensitive Topics",
-            "model_output.classification.iab_content.tier2.label": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+            "model_output.classification.iab_content.mapping_mode": "nearest_equivalent",
+            "model_output.classification.iab_content.tier1.label": "Sports",
+            "model_output.classification.iab_content.tier2.label": null,
             "model_output.classification.iab_content.tier3.label": null
           },
           "expected": {
@@ -5767,12 +5980,12 @@
           "id": "trial-signup-maps-to-software",
           "mismatches": [
             {
-              "actual": "Sensitive Topics",
+              "actual": "Sports",
               "expected": "Technology & Computing",
               "path": "model_output.classification.iab_content.tier1.label"
             },
             {
-              "actual": "Crime & Harmful Acts to Individuals, Society & Human Right Violations",
+              "actual": null,
               "expected": "Computing",
               "path": "model_output.classification.iab_content.tier2.label"
             },
@@ -5780,11 +5993,6 @@
               "actual": null,
               "expected": "Software and Applications",
               "path": "model_output.classification.iab_content.tier3.label"
-            },
-            {
-              "actual": "exact",
-              "expected": "nearest_equivalent",
-              "path": "model_output.classification.iab_content.mapping_mode"
             }
           ],
           "notes": "Software action queries should map to the software/application branch.",
@@ -6261,9 +6469,9 @@
   "heads": {
     "decision_phase": {
       "difficulty_benchmark": {
-        "accepted_accuracy": 0.9524,
+        "accepted_accuracy": 0.9619,
         "accepted_coverage": 1.0,
-        "accuracy": 0.9524,
+        "accuracy": 0.9619,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_difficulty_benchmark_confusion_matrix.csv",
         "count": 105,
         "dataset_path": "/content/agentic-intent-classifier/data/decision_phase_benchmark.jsonl",
@@ -6277,12 +6485,12 @@
             "macro_f1": 0.9711
           },
           "hard": {
-            "accepted_accuracy": 0.8857,
+            "accepted_accuracy": 0.9143,
             "accepted_coverage": 1.0,
-            "accuracy": 0.8857,
+            "accuracy": 0.9143,
             "count": 35,
             "fallback_rate": 0.0,
-            "macro_f1": 0.8908
+            "macro_f1": 0.9194
           },
           "medium": {
             "accepted_accuracy": 1.0,
@@ -6295,9 +6503,9 @@
         },
         "fallback_rate": 0.0,
         "head": "decision_phase",
-        "macro_f1": 0.9536,
+        "macro_f1": 0.9635,
         "per_class_metrics": {
-          "accuracy": 0.9523809523809523,
+          "accuracy": 0.9619047619047619,
           "action": {
             "f1-score": 0.9655172413793104,
             "precision": 1.0,
@@ -6311,21 +6519,21 @@
             "support": 15.0
           },
           "consideration": {
-            "f1-score": 0.9285714285714286,
+            "f1-score": 0.9655172413793104,
             "precision": 1.0,
-            "recall": 0.8666666666666667,
+            "recall": 0.9333333333333333,
             "support": 15.0
           },
           "decision": {
-            "f1-score": 0.9333333333333333,
-            "precision": 0.9333333333333333,
+            "f1-score": 0.9655172413793104,
+            "precision": 1.0,
             "recall": 0.9333333333333333,
             "support": 15.0
           },
           "macro avg": {
-            "f1-score": 0.9536131694056934,
-            "precision": 0.9604010025062657,
-            "recall": 0.9523809523809524,
+            "f1-score": 0.9634888438133874,
+            "precision": 0.9699248120300752,
+            "recall": 0.9619047619047619,
             "support": 105.0
           },
           "post_purchase": {
@@ -6347,26 +6555,26 @@
             "support": 15.0
           },
           "weighted avg": {
-            "f1-score": 0.9536131694056934,
-            "precision": 0.9604010025062656,
-            "recall": 0.9523809523809523,
+            "f1-score": 0.9634888438133875,
+            "precision": 0.9699248120300752,
+            "recall": 0.9619047619047619,
             "support": 105.0
           }
         },
         "suite": "difficulty_benchmark"
       },
       "final_wave_cases": {
-        "accepted_accuracy": 0.8889,
+        "accepted_accuracy": 0.8519,
         "accepted_coverage": 1.0,
-        "accuracy": 0.8889,
+        "accuracy": 0.8519,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_final_wave_cases_confusion_matrix.csv",
         "count": 27,
         "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/final_wave_cases.jsonl",
         "fallback_rate": 0.0,
         "head": "decision_phase",
-        "macro_f1": 0.8876,
+        "macro_f1": 0.8319,
         "per_class_metrics": {
-          "accuracy": 0.8888888888888888,
+          "accuracy": 0.8518518518518519,
           "action": {
             "f1-score": 0.0,
             "precision": 0.0,
@@ -6374,15 +6582,15 @@
             "support": 0.0
           },
           "awareness": {
-            "f1-score": 0.9090909090909091,
-            "precision": 0.8333333333333334,
+            "f1-score": 0.7692307692307693,
+            "precision": 0.625,
             "recall": 1.0,
             "support": 5.0
           },
           "consideration": {
-            "f1-score": 0.75,
+            "f1-score": 0.8888888888888888,
             "precision": 1.0,
-            "recall": 0.6,
+            "recall": 0.8,
             "support": 5.0
           },
           "decision": {
@@ -6392,9 +6600,9 @@
             "support": 5.0
           },
           "macro avg": {
-            "f1-score": 0.7608225108225108,
-            "precision": 0.7761904761904762,
-            "recall": 0.7642857142857142,
+            "f1-score": 0.7130647130647131,
+            "precision": 0.7321428571428571,
+            "recall": 0.7214285714285714,
             "support": 27.0
           },
           "post_purchase": {
@@ -6404,9 +6612,9 @@
             "support": 4.0
           },
           "research": {
-            "f1-score": 0.6666666666666666,
-            "precision": 0.6,
-            "recall": 0.75,
+            "f1-score": 0.3333333333333333,
+            "precision": 0.5,
+            "recall": 0.25,
             "support": 4.0
           },
           "support": {
@@ -6416,9 +6624,9 @@
             "support": 4.0
           },
           "weighted avg": {
-            "f1-score": 0.8874859708193041,
-            "precision": 0.9098765432098765,
-            "recall": 0.8888888888888888,
+            "f1-score": 0.8379233934789491,
+            "precision": 0.8564814814814815,
+            "recall": 0.8518518518518519,
             "support": 27.0
           }
         },
@@ -6502,7 +6710,7 @@
         "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/test.jsonl",
         "fallback_rate": 0.0,
         "head": "decision_phase",
-        "macro_f1": 0.7724,
+        "macro_f1": 0.7637,
         "per_class_metrics": {
           "accuracy": 0.7586206896551724,
           "action": {
@@ -6512,8 +6720,8 @@
             "support": 3.0
           },
           "awareness": {
-            "f1-score": 0.6666666666666666,
-            "precision": 0.5,
+            "f1-score": 0.6,
+            "precision": 0.42857142857142855,
             "recall": 1.0,
             "support": 3.0
           },
@@ -6524,27 +6732,27 @@
             "support": 5.0
           },
           "decision": {
-            "f1-score": 0.8888888888888888,
+            "f1-score": 1.0,
             "precision": 1.0,
-            "recall": 0.8,
+            "recall": 1.0,
             "support": 5.0
           },
           "macro avg": {
-            "f1-score": 0.7724489795918367,
-            "precision": 0.8095238095238095,
+            "f1-score": 0.763718820861678,
+            "precision": 0.7945578231292517,
             "recall": 0.7928571428571428,
             "support": 29.0
           },
           "post_purchase": {
-            "f1-score": 0.8,
-            "precision": 0.6666666666666666,
+            "f1-score": 0.8888888888888888,
+            "precision": 0.8,
             "recall": 1.0,
             "support": 4.0
           },
           "research": {
-            "f1-score": 0.4444444444444444,
-            "precision": 0.5,
-            "recall": 0.4,
+            "f1-score": 0.25,
+            "precision": 0.3333333333333333,
+            "recall": 0.2,
             "support": 5.0
           },
           "support": {
@@ -6554,8 +6762,8 @@
             "support": 4.0
           },
           "weighted avg": {
-            "f1-score": 0.7601806239737274,
-            "precision": 0.8160919540229885,
+            "f1-score": 0.7511767925561028,
+            "precision": 0.7983579638752052,
             "recall": 0.7586206896551724,
             "support": 29.0
           }
@@ -6563,17 +6771,17 @@
         "suite": "test"
       },
       "train": {
-        "accepted_accuracy": 0.9412,
+        "accepted_accuracy": 0.9706,
         "accepted_coverage": 1.0,
-        "accuracy": 0.9412,
+        "accuracy": 0.9706,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_train_confusion_matrix.csv",
         "count": 102,
         "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/train.jsonl",
         "fallback_rate": 0.0,
         "head": "decision_phase",
-        "macro_f1": 0.9464,
+        "macro_f1": 0.9729,
         "per_class_metrics": {
-          "accuracy": 0.9411764705882353,
+          "accuracy": 0.9705882352941176,
           "action": {
             "f1-score": 1.0,
             "precision": 1.0,
@@ -6587,21 +6795,21 @@
             "support": 16.0
           },
           "consideration": {
-            "f1-score": 0.8387096774193549,
-            "precision": 0.9285714285714286,
-            "recall": 0.7647058823529411,
+            "f1-score": 0.9375,
+            "precision": 1.0,
+            "recall": 0.8823529411764706,
             "support": 17.0
           },
           "decision": {
-            "f1-score": 0.967741935483871,
+            "f1-score": 1.0,
             "precision": 1.0,
-            "recall": 0.9375,
+            "recall": 1.0,
             "support": 16.0
           },
           "macro avg": {
-            "f1-score": 0.9463762044407206,
-            "precision": 0.9496465252767774,
-            "recall": 0.9479341736694679,
+            "f1-score": 0.9729175394497975,
+            "precision": 0.9737394957983193,
+            "recall": 0.9736694677871148,
             "support": 102.0
           },
           "post_purchase": {
@@ -6611,8 +6819,8 @@
             "support": 14.0
           },
           "research": {
-            "f1-score": 0.8484848484848485,
-            "precision": 0.7777777777777778,
+            "f1-score": 0.9032258064516129,
+            "precision": 0.875,
             "recall": 0.9333333333333333,
             "support": 15.0
           },
@@ -6623,35 +6831,35 @@
             "support": 14.0
           },
           "weighted avg": {
-            "f1-score": 0.9410231345715216,
-            "precision": 0.946188279233262,
-            "recall": 0.9411764705882353,
+            "f1-score": 0.9705984177639775,
+            "precision": 0.9723904267589389,
+            "recall": 0.9705882352941176,
             "support": 102.0
           }
         },
         "suite": "train"
       },
       "val": {
-        "accepted_accuracy": 0.8621,
+        "accepted_accuracy": 0.8276,
         "accepted_coverage": 1.0,
-        "accuracy": 0.8621,
+        "accuracy": 0.8276,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/decision_phase_val_confusion_matrix.csv",
         "count": 29,
         "dataset_path": "/content/agentic-intent-classifier/data/decision_phase/val.jsonl",
         "fallback_rate": 0.0,
         "head": "decision_phase",
-        "macro_f1": 0.8567,
+        "macro_f1": 0.8254,
         "per_class_metrics": {
-          "accuracy": 0.8620689655172413,
+          "accuracy": 0.8275862068965517,
           "action": {
-            "f1-score": 0.8571428571428571,
-            "precision": 0.75,
+            "f1-score": 1.0,
+            "precision": 1.0,
             "recall": 1.0,
             "support": 3.0
           },
           "awareness": {
-            "f1-score": 0.9090909090909091,
-            "precision": 0.8333333333333334,
+            "f1-score": 0.8333333333333334,
+            "precision": 0.7142857142857143,
             "recall": 1.0,
             "support": 5.0
           },
@@ -6668,21 +6876,21 @@
             "support": 4.0
           },
           "macro avg": {
-            "f1-score": 0.8566790352504637,
-            "precision": 0.880952380952381,
-            "recall": 0.8571428571428571,
+            "f1-score": 0.8254483611626469,
+            "precision": 0.8520408163265306,
+            "recall": 0.8214285714285714,
             "support": 29.0
           },
           "post_purchase": {
-            "f1-score": 0.8571428571428571,
-            "precision": 1.0,
+            "f1-score": 0.75,
+            "precision": 0.75,
             "recall": 0.75,
             "support": 4.0
           },
           "research": {
-            "f1-score": 0.75,
-            "precision": 0.75,
-            "recall": 0.75,
+            "f1-score": 0.5714285714285714,
+            "precision": 0.6666666666666666,
+            "recall": 0.5,
             "support": 4.0
           },
           "support": {
@@ -6692,9 +6900,9 @@
             "support": 4.0
           },
           "weighted avg": {
-            "f1-score": 0.8602776533811015,
-            "precision": 0.8821839080459771,
-            "recall": 0.8620689655172413,
+            "f1-score": 0.822585460516495,
+            "precision": 0.8415435139573071,
+            "recall": 0.8275862068965517,
             "support": 29.0
           }
         },
@@ -6703,92 +6911,92 @@
     },
     "iab_content": {
       "cross_vertical_benchmark": {
-        "accepted_accuracy": 0.427,
-        "accepted_coverage": 0.9889,
-        "accuracy": 0.4222,
+        "accepted_accuracy": 0.3108,
+        "accepted_coverage": 0.8222,
+        "accuracy": 0.2556,
         "count": 90,
         "dataset_path": "/content/agentic-intent-classifier/data/iab_cross_vertical_benchmark.jsonl",
         "difficulty_breakdown": {
           "easy": {
-            "accepted_accuracy": 0.4138,
-            "accepted_coverage": 0.9667,
-            "accuracy": 0.4,
+            "accepted_accuracy": 0.3636,
+            "accepted_coverage": 0.7333,
+            "accuracy": 0.2667,
             "count": 30,
-            "fallback_rate": 0.0333,
-            "macro_f1": 0.2727
+            "fallback_rate": 0.2667,
+            "macro_f1": 0.1778
           },
           "hard": {
-            "accepted_accuracy": 0.4667,
-            "accepted_coverage": 1.0,
-            "accuracy": 0.4667,
+            "accepted_accuracy": 0.3077,
+            "accepted_coverage": 0.8667,
+            "accuracy": 0.2667,
             "count": 30,
-            "fallback_rate": 0.0,
-            "macro_f1": 0.3106
+            "fallback_rate": 0.1333,
+            "macro_f1": 0.1562
           },
           "medium": {
-            "accepted_accuracy": 0.4,
-            "accepted_coverage": 1.0,
-            "accuracy": 0.4,
+            "accepted_accuracy": 0.2692,
+            "accepted_coverage": 0.8667,
+            "accuracy": 0.2333,
             "count": 30,
-            "fallback_rate": 0.0,
-            "macro_f1": 0.2667
+            "fallback_rate": 0.1333,
+            "macro_f1": 0.1591
           }
         },
-        "fallback_rate": 0.0111,
+        "fallback_rate": 0.1778,
         "head": "iab_content",
-        "macro_f1": 0.227,
+        "macro_f1": 0.1228,
         "primary_source": "supervised_classifier",
         "suite": "cross_vertical_benchmark",
         "tier_metrics": {
-          "average_prediction_depth": 2.4,
+          "average_prediction_depth": 1.9222,
           "error_buckets": {
-            "exact_match": 38,
-            "parent_safe_stop": 1,
-            "right_tier1_wrong_tier2": 14,
+            "exact_match": 23,
+            "parent_safe_stop": 3,
+            "right_tier1_wrong_tier2": 23,
             "wrong_deep_leaf": 8,
-            "wrong_tier1": 29
+            "wrong_tier1": 33
           },
-          "exact_path_accuracy": 0.4222,
-          "parent_safe_accuracy": 0.4444,
-          "tier1_accuracy": 0.6778,
-          "tier2_accuracy": 0.4881,
-          "tier3_accuracy": 0.5238,
-          "tier4_accuracy": 0.5
+          "exact_path_accuracy": 0.2556,
+          "parent_safe_accuracy": 0.4222,
+          "tier1_accuracy": 0.6333,
+          "tier2_accuracy": 0.3571,
+          "tier3_accuracy": 0.2381,
+          "tier4_accuracy": 0.0
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.4,
+            "average_prediction_depth": 1.9222,
             "error_buckets": {
-              "exact_match": 37,
-              "parent_safe_stop": 1,
-              "right_tier1_wrong_tier2": 14,
-              "wrong_deep_leaf": 9,
-              "wrong_tier1": 29
-            },
-            "exact_path_accuracy": 0.4111,
-            "parent_safe_accuracy": 0.4333,
-            "tier1_accuracy": 0.6778,
-            "tier2_accuracy": 0.4881,
-            "tier3_accuracy": 0.4762,
-            "tier4_accuracy": 0.5
+              "exact_match": 23,
+              "parent_safe_stop": 3,
+              "right_tier1_wrong_tier2": 23,
+              "wrong_deep_leaf": 8,
+              "wrong_tier1": 33
+            },
+            "exact_path_accuracy": 0.2556,
+            "parent_safe_accuracy": 0.4222,
+            "tier1_accuracy": 0.6333,
+            "tier2_accuracy": 0.3571,
+            "tier3_accuracy": 0.2381,
+            "tier4_accuracy": 0.0
           },
           "combined_path": {
-            "average_prediction_depth": 2.4,
+            "average_prediction_depth": 1.9222,
             "error_buckets": {
-              "exact_match": 37,
-              "parent_safe_stop": 1,
-              "right_tier1_wrong_tier2": 14,
-              "wrong_deep_leaf": 9,
-              "wrong_tier1": 29
-            },
-            "exact_path_accuracy": 0.4111,
-            "fallback_overuse_count": 25,
-            "fallback_rate": 0.2778,
-            "parent_safe_accuracy": 0.4333,
-            "tier1_accuracy": 0.6778,
-            "tier2_accuracy": 0.4881,
-            "tier3_accuracy": 0.4762,
-            "tier4_accuracy": 0.5
+              "exact_match": 23,
+              "parent_safe_stop": 3,
+              "right_tier1_wrong_tier2": 23,
+              "wrong_deep_leaf": 8,
+              "wrong_tier1": 33
+            },
+            "exact_path_accuracy": 0.2556,
+            "fallback_overuse_count": 19,
+            "fallback_rate": 0.2111,
+            "parent_safe_accuracy": 0.4222,
+            "tier1_accuracy": 0.6333,
+            "tier2_accuracy": 0.3571,
+            "tier3_accuracy": 0.2381,
+            "tier4_accuracy": 0.0
           },
           "disagreements": {
             "classifier_vs_combined": 0
@@ -6801,92 +7009,92 @@
         }
       },
       "difficulty_benchmark": {
-        "accepted_accuracy": 0.4231,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.4231,
+        "accepted_accuracy": 0.32,
+        "accepted_coverage": 0.8013,
+        "accuracy": 0.2564,
         "count": 156,
         "dataset_path": "/content/agentic-intent-classifier/data/iab_benchmark.jsonl",
         "difficulty_breakdown": {
           "easy": {
-            "accepted_accuracy": 0.4615,
-            "accepted_coverage": 1.0,
-            "accuracy": 0.4615,
+            "accepted_accuracy": 0.35,
+            "accepted_coverage": 0.7692,
+            "accuracy": 0.2692,
             "count": 52,
-            "fallback_rate": 0.0,
-            "macro_f1": 0.2359
+            "fallback_rate": 0.2308,
+            "macro_f1": 0.153
           },
           "hard": {
-            "accepted_accuracy": 0.3654,
-            "accepted_coverage": 1.0,
-            "accuracy": 0.3654,
+            "accepted_accuracy": 0.275,
+            "accepted_coverage": 0.7692,
+            "accuracy": 0.2115,
             "count": 52,
-            "fallback_rate": 0.0,
-            "macro_f1": 0.1892
+            "fallback_rate": 0.2308,
+            "macro_f1": 0.1108
           },
           "medium": {
-            "accepted_accuracy": 0.4423,
-            "accepted_coverage": 1.0,
-            "accuracy": 0.4423,
+            "accepted_accuracy": 0.3333,
+            "accepted_coverage": 0.8654,
+            "accuracy": 0.2885,
             "count": 52,
-            "fallback_rate": 0.0,
-            "macro_f1": 0.2338
+            "fallback_rate": 0.1346,
+            "macro_f1": 0.1491
           }
         },
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.1987,
         "head": "iab_content",
-        "macro_f1": 0.1524,
+        "macro_f1": 0.105,
         "primary_source": "supervised_classifier",
         "suite": "difficulty_benchmark",
         "tier_metrics": {
-          "average_prediction_depth": 2.4103,
+          "average_prediction_depth": 1.7564,
           "error_buckets": {
-            "exact_match": 66,
-            "parent_safe_stop": 1,
-            "right_tier1_wrong_tier2": 42,
-            "wrong_deep_leaf": 8,
-            "wrong_tier1": 39
+            "exact_match": 40,
+            "parent_safe_stop": 11,
+            "right_tier1_wrong_tier2": 58,
+            "wrong_deep_leaf": 1,
+            "wrong_tier1": 46
           },
-          "exact_path_accuracy": 0.4231,
-          "parent_safe_accuracy": 0.5385,
-          "tier1_accuracy": 0.75,
-          "tier2_accuracy": 0.4808,
-          "tier3_accuracy": 0.5093,
-          "tier4_accuracy": 0.4583
+          "exact_path_accuracy": 0.2564,
+          "parent_safe_accuracy": 0.6218,
+          "tier1_accuracy": 0.7051,
+          "tier2_accuracy": 0.3333,
+          "tier3_accuracy": 0.2315,
+          "tier4_accuracy": 0.0
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.4103,
+            "average_prediction_depth": 1.7564,
             "error_buckets": {
-              "exact_match": 59,
-              "parent_safe_stop": 1,
-              "right_tier1_wrong_tier2": 42,
-              "wrong_deep_leaf": 15,
-              "wrong_tier1": 39
-            },
-            "exact_path_accuracy": 0.3782,
-            "parent_safe_accuracy": 0.4936,
-            "tier1_accuracy": 0.75,
-            "tier2_accuracy": 0.4808,
-            "tier3_accuracy": 0.4259,
-            "tier4_accuracy": 0.1667
+              "exact_match": 40,
+              "parent_safe_stop": 11,
+              "right_tier1_wrong_tier2": 58,
+              "wrong_deep_leaf": 1,
+              "wrong_tier1": 46
+            },
+            "exact_path_accuracy": 0.2564,
+            "parent_safe_accuracy": 0.6218,
+            "tier1_accuracy": 0.7051,
+            "tier2_accuracy": 0.3333,
+            "tier3_accuracy": 0.2315,
+            "tier4_accuracy": 0.0
           },
           "combined_path": {
-            "average_prediction_depth": 2.4103,
+            "average_prediction_depth": 1.7564,
             "error_buckets": {
-              "exact_match": 59,
-              "parent_safe_stop": 1,
-              "right_tier1_wrong_tier2": 42,
-              "wrong_deep_leaf": 15,
-              "wrong_tier1": 39
-            },
-            "exact_path_accuracy": 0.3782,
-            "fallback_overuse_count": 15,
-            "fallback_rate": 0.0962,
-            "parent_safe_accuracy": 0.4936,
-            "tier1_accuracy": 0.75,
-            "tier2_accuracy": 0.4808,
-            "tier3_accuracy": 0.4259,
-            "tier4_accuracy": 0.1667
+              "exact_match": 40,
+              "parent_safe_stop": 11,
+              "right_tier1_wrong_tier2": 58,
+              "wrong_deep_leaf": 1,
+              "wrong_tier1": 46
+            },
+            "exact_path_accuracy": 0.2564,
+            "fallback_overuse_count": 13,
+            "fallback_rate": 0.0833,
+            "parent_safe_accuracy": 0.6218,
+            "tier1_accuracy": 0.7051,
+            "tier2_accuracy": 0.3333,
+            "tier3_accuracy": 0.2315,
+            "tier4_accuracy": 0.0
           },
           "disagreements": {
             "classifier_vs_combined": 0
@@ -6899,60 +7107,60 @@
         }
       },
       "extended_cases": {
-        "accepted_accuracy": 0.5,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.5,
+        "accepted_accuracy": 0.6,
+        "accepted_coverage": 0.625,
+        "accuracy": 0.375,
         "count": 8,
         "dataset_path": "/content/agentic-intent-classifier/data/iab/extended_cases.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.375,
         "head": "iab_content",
-        "macro_f1": 0.3333,
+        "macro_f1": 0.2308,
         "primary_source": "supervised_classifier",
         "suite": "extended_cases",
         "tier_metrics": {
-          "average_prediction_depth": 2.125,
+          "average_prediction_depth": 1.75,
           "error_buckets": {
-            "exact_match": 4,
-            "right_tier1_wrong_tier2": 2,
+            "exact_match": 3,
+            "right_tier1_wrong_tier2": 1,
             "wrong_deep_leaf": 1,
-            "wrong_tier1": 1
+            "wrong_tier1": 3
           },
-          "exact_path_accuracy": 0.5,
-          "parent_safe_accuracy": 0.5,
-          "tier1_accuracy": 0.875,
+          "exact_path_accuracy": 0.375,
+          "parent_safe_accuracy": 0.375,
+          "tier1_accuracy": 0.625,
           "tier2_accuracy": 0.5714,
           "tier3_accuracy": 0.0,
           "tier4_accuracy": 0.0
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.125,
+            "average_prediction_depth": 1.75,
             "error_buckets": {
-              "exact_match": 4,
-              "right_tier1_wrong_tier2": 2,
+              "exact_match": 3,
+              "right_tier1_wrong_tier2": 1,
               "wrong_deep_leaf": 1,
-              "wrong_tier1": 1
+              "wrong_tier1": 3
             },
-            "exact_path_accuracy": 0.5,
-            "parent_safe_accuracy": 0.5,
-            "tier1_accuracy": 0.875,
+            "exact_path_accuracy": 0.375,
+            "parent_safe_accuracy": 0.375,
+            "tier1_accuracy": 0.625,
             "tier2_accuracy": 0.5714,
             "tier3_accuracy": 0.0,
             "tier4_accuracy": 0.0
           },
           "combined_path": {
-            "average_prediction_depth": 2.125,
+            "average_prediction_depth": 1.75,
             "error_buckets": {
-              "exact_match": 4,
-              "right_tier1_wrong_tier2": 2,
+              "exact_match": 3,
+              "right_tier1_wrong_tier2": 1,
               "wrong_deep_leaf": 1,
-              "wrong_tier1": 1
+              "wrong_tier1": 3
             },
-            "exact_path_accuracy": 0.5,
+            "exact_path_accuracy": 0.375,
             "fallback_overuse_count": 2,
             "fallback_rate": 0.25,
-            "parent_safe_accuracy": 0.5,
-            "tier1_accuracy": 0.875,
+            "parent_safe_accuracy": 0.375,
+            "tier1_accuracy": 0.625,
             "tier2_accuracy": 0.5714,
             "tier3_accuracy": 0.0,
             "tier4_accuracy": 0.0
@@ -6968,18 +7176,18 @@
         }
       },
       "hard_cases": {
-        "accepted_accuracy": 0.4286,
-        "accepted_coverage": 0.875,
+        "accepted_accuracy": 0.6,
+        "accepted_coverage": 0.625,
         "accuracy": 0.375,
         "count": 8,
         "dataset_path": "/content/agentic-intent-classifier/data/iab/hard_cases.jsonl",
-        "fallback_rate": 0.125,
+        "fallback_rate": 0.375,
         "head": "iab_content",
         "macro_f1": 0.2308,
         "primary_source": "supervised_classifier",
         "suite": "hard_cases",
         "tier_metrics": {
-          "average_prediction_depth": 2.25,
+          "average_prediction_depth": 1.75,
           "error_buckets": {
             "exact_match": 3,
             "right_tier1_wrong_tier2": 1,
@@ -6994,7 +7202,7 @@
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.25,
+            "average_prediction_depth": 1.75,
             "error_buckets": {
               "exact_match": 3,
               "right_tier1_wrong_tier2": 1,
@@ -7008,7 +7216,7 @@
             "tier4_accuracy": 0.0
           },
           "combined_path": {
-            "average_prediction_depth": 2.25,
+            "average_prediction_depth": 1.75,
             "error_buckets": {
               "exact_match": 3,
               "right_tier1_wrong_tier2": 1,
@@ -7034,48 +7242,48 @@
         }
       },
       "test": {
-        "accepted_accuracy": 0.943,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.943,
+        "accepted_accuracy": 0.9278,
+        "accepted_coverage": 0.996,
+        "accuracy": 0.9247,
         "count": 3282,
         "dataset_path": "/content/agentic-intent-classifier/data/iab/test.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.004,
         "head": "iab_content",
-        "macro_f1": 0.911,
+        "macro_f1": 0.8814,
         "primary_source": "supervised_classifier",
         "suite": "test",
         "tier_metrics": {
-          "average_prediction_depth": 2.213,
+          "average_prediction_depth": 2.1706,
           "error_buckets": {
-            "exact_match": 3095,
-            "parent_safe_stop": 45,
-            "right_tier1_wrong_tier2": 41,
-            "wrong_deep_leaf": 72,
-            "wrong_tier1": 29
-          },
-          "exact_path_accuracy": 0.943,
-          "parent_safe_accuracy": 0.958,
-          "tier1_accuracy": 0.9912,
-          "tier2_accuracy": 0.9776,
-          "tier3_accuracy": 0.9078,
-          "tier4_accuracy": 0.7
+            "exact_match": 3035,
+            "parent_safe_stop": 87,
+            "right_tier1_wrong_tier2": 56,
+            "wrong_deep_leaf": 69,
+            "wrong_tier1": 35
+          },
+          "exact_path_accuracy": 0.9247,
+          "parent_safe_accuracy": 0.961,
+          "tier1_accuracy": 0.9893,
+          "tier2_accuracy": 0.9707,
+          "tier3_accuracy": 0.8487,
+          "tier4_accuracy": 0.5714
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.213,
+            "average_prediction_depth": 2.1706,
             "error_buckets": {
-              "exact_match": 3052,
-              "parent_safe_stop": 44,
-              "right_tier1_wrong_tier2": 53,
-              "wrong_deep_leaf": 104,
-              "wrong_tier1": 29
-            },
-            "exact_path_accuracy": 0.9299,
-            "parent_safe_accuracy": 0.9445,
-            "tier1_accuracy": 0.9912,
-            "tier2_accuracy": 0.9734,
-            "tier3_accuracy": 0.8725,
-            "tier4_accuracy": 0.5
+              "exact_match": 3004,
+              "parent_safe_stop": 84,
+              "right_tier1_wrong_tier2": 68,
+              "wrong_deep_leaf": 91,
+              "wrong_tier1": 35
+            },
+            "exact_path_accuracy": 0.9153,
+            "parent_safe_accuracy": 0.9506,
+            "tier1_accuracy": 0.9893,
+            "tier2_accuracy": 0.9665,
+            "tier3_accuracy": 0.8259,
+            "tier4_accuracy": 0.4429
           },
           "combined_path": {
             "count": 3282,
@@ -7097,48 +7305,48 @@
         }
       },
       "train": {
-        "accepted_accuracy": 0.9459,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.9459,
+        "accepted_accuracy": 0.9314,
+        "accepted_coverage": 0.9972,
+        "accuracy": 0.9295,
         "count": 13211,
         "dataset_path": "/content/agentic-intent-classifier/data/iab/train.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.0028,
         "head": "iab_content",
-        "macro_f1": 0.9194,
+        "macro_f1": 0.8927,
         "primary_source": "supervised_classifier",
         "suite": "train",
         "tier_metrics": {
-          "average_prediction_depth": 2.2105,
+          "average_prediction_depth": 2.1683,
           "error_buckets": {
-            "exact_match": 12496,
-            "parent_safe_stop": 162,
-            "right_tier1_wrong_tier2": 144,
-            "wrong_deep_leaf": 284,
-            "wrong_tier1": 125
-          },
-          "exact_path_accuracy": 0.9459,
-          "parent_safe_accuracy": 0.9585,
-          "tier1_accuracy": 0.9905,
-          "tier2_accuracy": 0.9805,
-          "tier3_accuracy": 0.9135,
-          "tier4_accuracy": 0.7268
+            "exact_match": 12280,
+            "parent_safe_stop": 312,
+            "right_tier1_wrong_tier2": 215,
+            "wrong_deep_leaf": 288,
+            "wrong_tier1": 116
+          },
+          "exact_path_accuracy": 0.9295,
+          "parent_safe_accuracy": 0.9618,
+          "tier1_accuracy": 0.9912,
+          "tier2_accuracy": 0.9737,
+          "tier3_accuracy": 0.8557,
+          "tier4_accuracy": 0.6107
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.2105,
+            "average_prediction_depth": 2.1683,
             "error_buckets": {
-              "exact_match": 12323,
-              "parent_safe_stop": 157,
-              "right_tier1_wrong_tier2": 192,
-              "wrong_deep_leaf": 414,
-              "wrong_tier1": 125
-            },
-            "exact_path_accuracy": 0.9328,
-            "parent_safe_accuracy": 0.945,
-            "tier1_accuracy": 0.9905,
-            "tier2_accuracy": 0.9764,
-            "tier3_accuracy": 0.8777,
-            "tier4_accuracy": 0.525
+              "exact_match": 12145,
+              "parent_safe_stop": 300,
+              "right_tier1_wrong_tier2": 263,
+              "wrong_deep_leaf": 387,
+              "wrong_tier1": 116
+            },
+            "exact_path_accuracy": 0.9193,
+            "parent_safe_accuracy": 0.9507,
+            "tier1_accuracy": 0.9912,
+            "tier2_accuracy": 0.9695,
+            "tier3_accuracy": 0.8301,
+            "tier4_accuracy": 0.475
           },
           "combined_path": {
             "count": 13211,
@@ -7160,48 +7368,48 @@
         }
       },
       "val": {
-        "accepted_accuracy": 0.9442,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.9442,
+        "accepted_accuracy": 0.9273,
+        "accepted_coverage": 0.9973,
+        "accuracy": 0.9254,
         "count": 3282,
         "dataset_path": "/content/agentic-intent-classifier/data/iab/val.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.0027,
         "head": "iab_content",
-        "macro_f1": 0.9166,
+        "macro_f1": 0.8864,
         "primary_source": "supervised_classifier",
         "suite": "val",
         "tier_metrics": {
-          "average_prediction_depth": 2.2151,
+          "average_prediction_depth": 2.1709,
           "error_buckets": {
-            "exact_match": 3099,
-            "parent_safe_stop": 35,
-            "right_tier1_wrong_tier2": 45,
-            "wrong_deep_leaf": 72,
-            "wrong_tier1": 31
-          },
-          "exact_path_accuracy": 0.9442,
-          "parent_safe_accuracy": 0.9576,
-          "tier1_accuracy": 0.9906,
-          "tier2_accuracy": 0.9769,
-          "tier3_accuracy": 0.9088,
-          "tier4_accuracy": 0.7286
+            "exact_match": 3037,
+            "parent_safe_stop": 80,
+            "right_tier1_wrong_tier2": 55,
+            "wrong_deep_leaf": 74,
+            "wrong_tier1": 36
+          },
+          "exact_path_accuracy": 0.9254,
+          "parent_safe_accuracy": 0.9613,
+          "tier1_accuracy": 0.989,
+          "tier2_accuracy": 0.9713,
+          "tier3_accuracy": 0.8549,
+          "tier4_accuracy": 0.6071
         },
         "view_metrics": {
           "classifier": {
-            "average_prediction_depth": 2.2151,
+            "average_prediction_depth": 2.1709,
             "error_buckets": {
-              "exact_match": 3056,
-              "parent_safe_stop": 34,
-              "right_tier1_wrong_tier2": 57,
-              "wrong_deep_leaf": 104,
-              "wrong_tier1": 31
-            },
-            "exact_path_accuracy": 0.9311,
-            "parent_safe_accuracy": 0.9442,
-            "tier1_accuracy": 0.9906,
-            "tier2_accuracy": 0.9727,
-            "tier3_accuracy": 0.8736,
-            "tier4_accuracy": 0.5286
+              "exact_match": 3002,
+              "parent_safe_stop": 78,
+              "right_tier1_wrong_tier2": 67,
+              "wrong_deep_leaf": 99,
+              "wrong_tier1": 36
+            },
+            "exact_path_accuracy": 0.9147,
+            "parent_safe_accuracy": 0.95,
+            "tier1_accuracy": 0.989,
+            "tier2_accuracy": 0.9672,
+            "tier3_accuracy": 0.829,
+            "tier4_accuracy": 0.4643
           },
           "combined_path": {
             "count": 3282,
@@ -7225,9 +7433,9 @@
     },
     "intent_subtype": {
       "difficulty_benchmark": {
-        "accepted_accuracy": 0.9104,
-        "accepted_coverage": 0.9675,
-        "accuracy": 0.8917,
+        "accepted_accuracy": 0.8901,
+        "accepted_coverage": 0.9856,
+        "accuracy": 0.8845,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_difficulty_benchmark_confusion_matrix.csv",
         "count": 277,
         "dataset_path": "/content/agentic-intent-classifier/data/subtype_benchmark.jsonl",
@@ -7238,52 +7446,52 @@
             "accuracy": 0.913,
             "count": 92,
             "fallback_rate": 0.0,
-            "macro_f1": 0.9109
+            "macro_f1": 0.9124
           },
           "hard": {
-            "accepted_accuracy": 0.8554,
-            "accepted_coverage": 0.9121,
-            "accuracy": 0.8132,
+            "accepted_accuracy": 0.8295,
+            "accepted_coverage": 0.967,
+            "accuracy": 0.8242,
             "count": 91,
-            "fallback_rate": 0.0879,
-            "macro_f1": 0.8025
+            "fallback_rate": 0.033,
+            "macro_f1": 0.8183
           },
           "medium": {
-            "accepted_accuracy": 0.957,
+            "accepted_accuracy": 0.9247,
             "accepted_coverage": 0.9894,
-            "accuracy": 0.9468,
+            "accuracy": 0.9149,
             "count": 94,
             "fallback_rate": 0.0106,
-            "macro_f1": 0.9469
+            "macro_f1": 0.9117
           }
         },
-        "fallback_rate": 0.0325,
+        "fallback_rate": 0.0144,
         "head": "intent_subtype",
-        "macro_f1": 0.8886,
+        "macro_f1": 0.8824,
         "per_class_metrics": {
           "account_help": {
-            "f1-score": 0.64,
-            "precision": 0.8,
-            "recall": 0.5333333333333333,
+            "f1-score": 0.7142857142857143,
+            "precision": 0.7692307692307693,
+            "recall": 0.6666666666666666,
             "support": 15.0
           },
-          "accuracy": 0.8916967509025271,
+          "accuracy": 0.8844765342960289,
           "billing_help": {
-            "f1-score": 0.8387096774193549,
-            "precision": 0.8125,
+            "f1-score": 0.8666666666666667,
+            "precision": 0.8666666666666667,
             "recall": 0.8666666666666667,
             "support": 15.0
           },
           "booking": {
-            "f1-score": 0.9285714285714286,
+            "f1-score": 0.75,
             "precision": 1.0,
-            "recall": 0.8666666666666667,
+            "recall": 0.6,
             "support": 15.0
           },
           "comparison": {
-            "f1-score": 0.8148148148148148,
-            "precision": 0.9166666666666666,
-            "recall": 0.7333333333333333,
+            "f1-score": 0.8888888888888888,
+            "precision": 1.0,
+            "recall": 0.8,
             "support": 15.0
           },
           "contact_sales": {
@@ -7293,15 +7501,15 @@
             "support": 15.0
           },
           "deal_seeking": {
-            "f1-score": 0.896551724137931,
-            "precision": 0.9285714285714286,
-            "recall": 0.8666666666666667,
+            "f1-score": 0.8888888888888888,
+            "precision": 1.0,
+            "recall": 0.8,
             "support": 15.0
           },
           "download": {
-            "f1-score": 0.9285714285714286,
+            "f1-score": 0.9655172413793104,
             "precision": 1.0,
-            "recall": 0.8666666666666667,
+            "recall": 0.9333333333333333,
             "support": 15.0
           },
           "education": {
@@ -7311,15 +7519,15 @@
             "support": 15.0
           },
           "emotional_reflection": {
-            "f1-score": 1.0,
-            "precision": 1.0,
+            "f1-score": 0.9375,
+            "precision": 0.8823529411764706,
             "recall": 1.0,
             "support": 15.0
           },
           "evaluation": {
-            "f1-score": 0.896551724137931,
-            "precision": 0.9285714285714286,
-            "recall": 0.8666666666666667,
+            "f1-score": 0.9333333333333333,
+            "precision": 0.9333333333333333,
+            "recall": 0.9333333333333333,
             "support": 15.0
           },
           "follow_up": {
@@ -7329,21 +7537,21 @@
             "support": 15.0
           },
           "macro avg": {
-            "f1-score": 0.8886471209737711,
-            "precision": 0.8965122159975102,
-            "recall": 0.8895561002178651,
+            "f1-score": 0.8824228919733669,
+            "precision": 0.8968567719420234,
+            "recall": 0.8825617283950618,
             "support": 277.0
           },
           "onboarding_setup": {
-            "f1-score": 0.8648648648648649,
-            "precision": 0.8,
-            "recall": 0.9411764705882353,
+            "f1-score": 0.918918918918919,
+            "precision": 0.85,
+            "recall": 1.0,
             "support": 17.0
           },
           "product_discovery": {
-            "f1-score": 0.9032258064516129,
-            "precision": 0.875,
-            "recall": 0.9333333333333333,
+            "f1-score": 0.9375,
+            "precision": 0.8823529411764706,
+            "recall": 1.0,
             "support": 15.0
           },
           "provider_selection": {
@@ -7353,56 +7561,56 @@
             "support": 16.0
           },
           "purchase": {
-            "f1-score": 0.896551724137931,
-            "precision": 0.9285714285714286,
-            "recall": 0.8666666666666667,
+            "f1-score": 0.8888888888888888,
+            "precision": 1.0,
+            "recall": 0.8,
             "support": 15.0
           },
           "signup": {
-            "f1-score": 0.8823529411764706,
-            "precision": 0.8333333333333334,
-            "recall": 0.9375,
+            "f1-score": 0.8235294117647058,
+            "precision": 0.7777777777777778,
+            "recall": 0.875,
             "support": 16.0
           },
           "task_execution": {
-            "f1-score": 0.9230769230769231,
-            "precision": 0.8571428571428571,
-            "recall": 1.0,
+            "f1-score": 0.8292682926829268,
+            "precision": 0.7391304347826086,
+            "recall": 0.9444444444444444,
             "support": 18.0
           },
           "troubleshooting": {
-            "f1-score": 0.8,
-            "precision": 0.8,
-            "recall": 0.8,
+            "f1-score": 0.7586206896551724,
+            "precision": 0.7857142857142857,
+            "recall": 0.7333333333333333,
             "support": 15.0
           },
           "weighted avg": {
-            "f1-score": 0.8891181699377334,
-            "precision": 0.8953221541324111,
-            "recall": 0.8916967509025271,
+            "f1-score": 0.8822131766431675,
+            "precision": 0.8945403392673653,
+            "recall": 0.8844765342960289,
             "support": 277.0
           }
         },
         "suite": "difficulty_benchmark"
       },
       "extended_cases": {
-        "accepted_accuracy": 0.8302,
+        "accepted_accuracy": 0.8113,
         "accepted_coverage": 1.0,
-        "accuracy": 0.8302,
+        "accuracy": 0.8113,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_extended_cases_confusion_matrix.csv",
         "count": 53,
         "dataset_path": "/content/agentic-intent-classifier/data/subtype/extended_cases.jsonl",
         "fallback_rate": 0.0,
         "head": "intent_subtype",
-        "macro_f1": 0.7668,
+        "macro_f1": 0.7517,
         "per_class_metrics": {
           "account_help": {
-            "f1-score": 0.8,
-            "precision": 1.0,
+            "f1-score": 0.6666666666666666,
+            "precision": 0.6666666666666666,
             "recall": 0.6666666666666666,
             "support": 3.0
           },
-          "accuracy": 0.8301886792452831,
+          "accuracy": 0.8113207547169812,
           "billing_help": {
             "f1-score": 0.0,
             "precision": 0.0,
@@ -7416,9 +7624,9 @@
             "support": 0.0
           },
           "comparison": {
-            "f1-score": 1.0,
+            "f1-score": 0.6666666666666666,
             "precision": 1.0,
-            "recall": 1.0,
+            "recall": 0.5,
             "support": 2.0
           },
           "contact_sales": {
@@ -7428,9 +7636,9 @@
             "support": 0.0
           },
           "deal_seeking": {
-            "f1-score": 0.8571428571428571,
-            "precision": 0.75,
-            "recall": 1.0,
+            "f1-score": 0.7619047619047619,
+            "precision": 0.6666666666666666,
+            "recall": 0.8888888888888888,
             "support": 9.0
           },
           "download": {
@@ -7458,15 +7666,15 @@
             "support": 3.0
           },
           "follow_up": {
-            "f1-score": 0.7368421052631579,
+            "f1-score": 0.8,
             "precision": 1.0,
-            "recall": 0.5833333333333334,
+            "recall": 0.6666666666666666,
             "support": 12.0
           },
           "macro avg": {
-            "f1-score": 0.46858256266151,
-            "precision": 0.4565696649029982,
-            "recall": 0.513888888888889,
+            "f1-score": 0.45939292189292186,
+            "precision": 0.4611992945326278,
+            "recall": 0.48456790123456783,
             "support": 53.0
           },
           "onboarding_setup": {
@@ -7500,8 +7708,8 @@
             "support": 0.0
           },
           "task_execution": {
-            "f1-score": 0.6666666666666666,
-            "precision": 0.5,
+            "f1-score": 1.0,
+            "precision": 1.0,
             "recall": 1.0,
             "support": 1.0
           },
@@ -7512,9 +7720,9 @@
             "support": 1.0
           },
           "weighted avg": {
-            "f1-score": 0.8018611395225099,
-            "precision": 0.8208295896975142,
-            "recall": 0.8301886792452831,
+            "f1-score": 0.7861520554916781,
+            "precision": 0.7972446840371369,
+            "recall": 0.8113207547169812,
             "support": 53.0
           }
         },
@@ -7529,7 +7737,7 @@
         "dataset_path": "/content/agentic-intent-classifier/data/subtype/hard_cases.jsonl",
         "fallback_rate": 0.0,
         "head": "intent_subtype",
-        "macro_f1": 0.8447,
+        "macro_f1": 0.8426,
         "per_class_metrics": {
           "account_help": {
             "f1-score": 0.8,
@@ -7575,8 +7783,8 @@
             "support": 0.0
           },
           "education": {
-            "f1-score": 0.9508196721311475,
-            "precision": 0.90625,
+            "f1-score": 0.9666666666666667,
+            "precision": 0.9354838709677419,
             "recall": 1.0,
             "support": 29.0
           },
@@ -7599,8 +7807,8 @@
             "support": 12.0
           },
           "macro avg": {
-            "f1-score": 0.7038911013311109,
-            "precision": 0.7234953703703704,
+            "f1-score": 0.7021723995980289,
+            "precision": 0.7210790702726187,
             "recall": 0.7212962962962962,
             "support": 94.0
           },
@@ -7611,8 +7819,8 @@
             "support": 6.0
           },
           "product_discovery": {
-            "f1-score": 0.8888888888888888,
-            "precision": 0.8,
+            "f1-score": 0.8421052631578947,
+            "precision": 0.7272727272727273,
             "recall": 1.0,
             "support": 8.0
           },
@@ -7647,8 +7855,8 @@
             "support": 3.0
           },
           "weighted avg": {
-            "f1-score": 0.8798004011763282,
-            "precision": 0.8911125886524823,
+            "f1-score": 0.8807077824069889,
+            "precision": 0.8939419937189327,
             "recall": 0.8936170212765957,
             "support": 94.0
           }
@@ -7656,15 +7864,15 @@
         "suite": "hard_cases"
       },
       "test": {
-        "accepted_accuracy": 0.9,
+        "accepted_accuracy": 0.8714,
         "accepted_coverage": 1.0,
-        "accuracy": 0.9,
+        "accuracy": 0.8714,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_test_confusion_matrix.csv",
         "count": 70,
         "dataset_path": "/content/agentic-intent-classifier/data/subtype/test.jsonl",
         "fallback_rate": 0.0,
         "head": "intent_subtype",
-        "macro_f1": 0.8531,
+        "macro_f1": 0.8317,
         "per_class_metrics": {
           "account_help": {
             "f1-score": 1.0,
@@ -7672,7 +7880,7 @@
             "recall": 1.0,
             "support": 2.0
           },
-          "accuracy": 0.9,
+          "accuracy": 0.8714285714285714,
           "billing_help": {
             "f1-score": 0.0,
             "precision": 0.0,
@@ -7698,9 +7906,9 @@
             "support": 0.0
           },
           "deal_seeking": {
-            "f1-score": 0.6666666666666666,
-            "precision": 0.5,
-            "recall": 1.0,
+            "f1-score": 0.3333333333333333,
+            "precision": 0.25,
+            "recall": 0.5,
             "support": 2.0
           },
           "download": {
@@ -7722,9 +7930,9 @@
             "support": 5.0
           },
           "evaluation": {
-            "f1-score": 0.0,
-            "precision": 0.0,
-            "recall": 0.0,
+            "f1-score": 0.5,
+            "precision": 0.5,
+            "recall": 0.5,
             "support": 2.0
           },
           "follow_up": {
@@ -7734,9 +7942,9 @@
             "support": 11.0
           },
           "macro avg": {
-            "f1-score": 0.6635221022395855,
-            "precision": 0.6578042328042328,
-            "recall": 0.6885521885521885,
+            "f1-score": 0.646896135613619,
+            "precision": 0.6657407407407407,
+            "recall": 0.6538299663299663,
             "support": 70.0
           },
           "onboarding_setup": {
@@ -7746,9 +7954,9 @@
             "support": 4.0
           },
           "product_discovery": {
-            "f1-score": 1.0,
+            "f1-score": 0.9333333333333333,
             "precision": 1.0,
-            "recall": 1.0,
+            "recall": 0.875,
             "support": 8.0
           },
           "provider_selection": {
@@ -7770,44 +7978,44 @@
             "support": 2.0
           },
           "task_execution": {
-            "f1-score": 0.9230769230769231,
-            "precision": 0.8571428571428571,
+            "f1-score": 0.8571428571428571,
+            "precision": 0.75,
             "recall": 1.0,
             "support": 6.0
           },
           "troubleshooting": {
-            "f1-score": 1.0,
+            "f1-score": 0.6666666666666666,
             "precision": 1.0,
-            "recall": 1.0,
+            "recall": 0.5,
             "support": 2.0
           },
           "weighted avg": {
-            "f1-score": 0.8939882610403741,
-            "precision": 0.9094217687074829,
-            "recall": 0.9,
+            "f1-score": 0.8759558172936446,
+            "precision": 0.9073809523809524,
+            "recall": 0.8714285714285714,
             "support": 70.0
           }
         },
         "suite": "test"
       },
       "train": {
-        "accepted_accuracy": 0.8978,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.8978,
+        "accepted_accuracy": 0.9068,
+        "accepted_coverage": 0.9936,
+        "accuracy": 0.9042,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_train_confusion_matrix.csv",
         "count": 313,
         "dataset_path": "/content/agentic-intent-classifier/data/subtype/train.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.0064,
         "head": "intent_subtype",
-        "macro_f1": 0.877,
+        "macro_f1": 0.8787,
         "per_class_metrics": {
           "account_help": {
-            "f1-score": 0.7142857142857143,
-            "precision": 0.7142857142857143,
-            "recall": 0.7142857142857143,
+            "f1-score": 0.75,
+            "precision": 0.6666666666666666,
+            "recall": 0.8571428571428571,
             "support": 7.0
           },
-          "accuracy": 0.8977635782747604,
+          "accuracy": 0.9041533546325878,
           "billing_help": {
             "f1-score": 1.0,
             "precision": 1.0,
@@ -7821,32 +8029,32 @@
             "support": 5.0
           },
           "comparison": {
-            "f1-score": 0.967741935483871,
-            "precision": 0.9375,
-            "recall": 1.0,
+            "f1-score": 0.9655172413793104,
+            "precision": 1.0,
+            "recall": 0.9333333333333333,
             "support": 15.0
           },
           "contact_sales": {
-            "f1-score": 0.8,
+            "f1-score": 0.7142857142857143,
             "precision": 1.0,
-            "recall": 0.6666666666666666,
+            "recall": 0.5555555555555556,
             "support": 9.0
           },
           "deal_seeking": {
-            "f1-score": 0.9090909090909091,
-            "precision": 0.9090909090909091,
+            "f1-score": 0.9523809523809523,
+            "precision": 1.0,
             "recall": 0.9090909090909091,
             "support": 11.0
           },
           "download": {
-            "f1-score": 0.9411764705882353,
-            "precision": 0.8888888888888888,
+            "f1-score": 1.0,
+            "precision": 1.0,
             "recall": 1.0,
             "support": 8.0
           },
           "education": {
-            "f1-score": 0.9629629629629629,
-            "precision": 0.9285714285714286,
+            "f1-score": 0.9719626168224299,
+            "precision": 0.9454545454545454,
             "recall": 1.0,
             "support": 52.0
           },
@@ -7857,21 +8065,21 @@
             "support": 20.0
           },
           "evaluation": {
-            "f1-score": 0.6923076923076923,
-            "precision": 1.0,
+            "f1-score": 0.6666666666666666,
+            "precision": 0.9,
             "recall": 0.5294117647058824,
             "support": 17.0
           },
           "follow_up": {
-            "f1-score": 0.8571428571428571,
-            "precision": 0.8823529411764706,
-            "recall": 0.8333333333333334,
+            "f1-score": 0.8888888888888888,
+            "precision": 0.8888888888888888,
+            "recall": 0.8888888888888888,
             "support": 36.0
           },
           "macro avg": {
-            "f1-score": 0.8770498618135788,
-            "precision": 0.8988923431325393,
-            "recall": 0.876671278202288,
+            "f1-score": 0.8786951095392168,
+            "precision": 0.9007433723013433,
+            "recall": 0.8782602497120292,
             "support": 313.0
           },
           "onboarding_setup": {
@@ -7881,9 +8089,9 @@
             "support": 17.0
           },
           "product_discovery": {
-            "f1-score": 0.90625,
-            "precision": 0.8787878787878788,
-            "recall": 0.9354838709677419,
+            "f1-score": 0.9090909090909091,
+            "precision": 0.8571428571428571,
+            "recall": 0.967741935483871,
             "support": 31.0
           },
           "provider_selection": {
@@ -7893,48 +8101,48 @@
             "support": 25.0
           },
           "purchase": {
-            "f1-score": 0.8,
-            "precision": 1.0,
+            "f1-score": 0.7272727272727273,
+            "precision": 0.8,
             "recall": 0.6666666666666666,
             "support": 6.0
           },
           "signup": {
-            "f1-score": 0.8648648648648649,
-            "precision": 0.7619047619047619,
+            "f1-score": 0.8888888888888888,
+            "precision": 0.8,
             "recall": 1.0,
             "support": 16.0
           },
           "task_execution": {
-            "f1-score": 0.8292682926829268,
-            "precision": 0.7727272727272727,
-            "recall": 0.8947368421052632,
+            "f1-score": 0.8571428571428571,
+            "precision": 0.782608695652174,
+            "recall": 0.9473684210526315,
             "support": 19.0
           },
           "troubleshooting": {
-            "f1-score": 0.8,
-            "precision": 0.8333333333333334,
-            "recall": 0.7692307692307693,
+            "f1-score": 0.782608695652174,
+            "precision": 0.9,
+            "recall": 0.6923076923076923,
             "support": 13.0
           },
           "weighted avg": {
-            "f1-score": 0.894423568060199,
-            "precision": 0.9063956713482179,
-            "recall": 0.8977635782747604,
+            "f1-score": 0.9005147505975646,
+            "precision": 0.9118244687664052,
+            "recall": 0.9041533546325878,
             "support": 313.0
           }
         },
         "suite": "train"
       },
       "val": {
-        "accepted_accuracy": 0.8608,
-        "accepted_coverage": 0.9875,
-        "accuracy": 0.85,
+        "accepted_accuracy": 0.8625,
+        "accepted_coverage": 1.0,
+        "accuracy": 0.8625,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_subtype_val_confusion_matrix.csv",
         "count": 80,
         "dataset_path": "/content/agentic-intent-classifier/data/subtype/val.jsonl",
-        "fallback_rate": 0.0125,
+        "fallback_rate": 0.0,
         "head": "intent_subtype",
-        "macro_f1": 0.6722,
+        "macro_f1": 0.6561,
         "per_class_metrics": {
           "account_help": {
             "f1-score": 0.5,
@@ -7942,7 +8150,7 @@
             "recall": 0.5,
             "support": 2.0
           },
-          "accuracy": 0.85,
+          "accuracy": 0.8625,
           "billing_help": {
             "f1-score": 0.0,
             "precision": 0.0,
@@ -7956,9 +8164,9 @@
             "support": 3.0
           },
           "comparison": {
-            "f1-score": 0.5,
-            "precision": 0.5,
-            "recall": 0.5,
+            "f1-score": 0.4,
+            "precision": 1.0,
+            "recall": 0.25,
             "support": 4.0
           },
           "contact_sales": {
@@ -7968,8 +8176,8 @@
             "support": 0.0
           },
           "deal_seeking": {
-            "f1-score": 0.8,
-            "precision": 0.6666666666666666,
+            "f1-score": 0.6666666666666666,
+            "precision": 0.5,
             "recall": 1.0,
             "support": 2.0
           },
@@ -8004,21 +8212,21 @@
             "support": 11.0
           },
           "macro avg": {
-            "f1-score": 0.5974890931031281,
-            "precision": 0.5811447811447812,
-            "recall": 0.6353535353535353,
+            "f1-score": 0.5832054560954817,
+            "precision": 0.5891975308641975,
+            "recall": 0.6315656565656567,
             "support": 80.0
           },
           "onboarding_setup": {
-            "f1-score": 0.8888888888888888,
-            "precision": 1.0,
+            "f1-score": 0.8,
+            "precision": 0.8,
             "recall": 0.8,
             "support": 5.0
           },
           "product_discovery": {
-            "f1-score": 0.8571428571428571,
-            "precision": 0.9,
-            "recall": 0.8181818181818182,
+            "f1-score": 0.9565217391304348,
+            "precision": 0.9166666666666666,
+            "recall": 1.0,
             "support": 11.0
           },
           "provider_selection": {
@@ -8034,14 +8242,14 @@
             "support": 2.0
           },
           "signup": {
-            "f1-score": 0.8,
-            "precision": 0.6666666666666666,
+            "f1-score": 0.6666666666666666,
+            "precision": 0.5,
             "recall": 1.0,
             "support": 2.0
           },
           "task_execution": {
-            "f1-score": 0.8421052631578947,
-            "precision": 0.7272727272727273,
+            "f1-score": 0.9411764705882353,
+            "precision": 0.8888888888888888,
             "recall": 1.0,
             "support": 8.0
           },
@@ -8052,9 +8260,9 @@
             "support": 1.0
           },
           "weighted avg": {
-            "f1-score": 0.8380398913951546,
-            "precision": 0.8423106060606059,
-            "recall": 0.85,
+            "f1-score": 0.8443893861892583,
+            "precision": 0.8649305555555555,
+            "recall": 0.8625,
             "support": 80.0
           }
         },
@@ -8177,12 +8385,12 @@
       },
       "hard_cases": {
         "accepted_accuracy": 1.0,
-        "accepted_coverage": 0.9836,
+        "accepted_coverage": 1.0,
         "accuracy": 1.0,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_hard_cases_confusion_matrix.csv",
         "count": 61,
         "dataset_path": "/content/agentic-intent-classifier/data/hard_cases.jsonl",
-        "fallback_rate": 0.0164,
+        "fallback_rate": 0.0,
         "head": "intent_type",
         "macro_f1": 1.0,
         "per_class_metrics": {
@@ -8263,17 +8471,17 @@
         "suite": "hard_cases"
       },
       "test": {
-        "accepted_accuracy": 0.9362,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.9362,
+        "accepted_accuracy": 0.8889,
+        "accepted_coverage": 0.9574,
+        "accuracy": 0.8723,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_test_confusion_matrix.csv",
         "count": 47,
         "dataset_path": "/content/agentic-intent-classifier/data/test.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.0426,
         "head": "intent_type",
-        "macro_f1": 0.9235,
+        "macro_f1": 0.8006,
         "per_class_metrics": {
-          "accuracy": 0.9361702127659575,
+          "accuracy": 0.8723404255319149,
           "ambiguous": {
             "f1-score": 0.875,
             "precision": 1.0,
@@ -8287,15 +8495,15 @@
             "support": 1.0
           },
           "commercial": {
-            "f1-score": 0.9523809523809523,
-            "precision": 0.9090909090909091,
-            "recall": 1.0,
+            "f1-score": 0.9,
+            "precision": 0.9,
+            "recall": 0.9,
             "support": 10.0
           },
           "creative_generation": {
-            "f1-score": 1.0,
-            "precision": 1.0,
-            "recall": 1.0,
+            "f1-score": 0.0,
+            "precision": 0.0,
+            "recall": 0.0,
             "support": 1.0
           },
           "exploratory": {
@@ -8305,15 +8513,15 @@
             "support": 1.0
           },
           "informational": {
-            "f1-score": 0.9411764705882353,
-            "precision": 0.8888888888888888,
+            "f1-score": 0.8888888888888888,
+            "precision": 0.8,
             "recall": 1.0,
             "support": 8.0
           },
           "macro avg": {
-            "f1-score": 0.9235224089635853,
-            "precision": 0.9297979797979797,
-            "recall": 0.9444444444444444,
+            "f1-score": 0.8005555555555555,
+            "precision": 0.8074999999999999,
+            "recall": 0.8219444444444445,
             "support": 47.0
           },
           "personal_reflection": {
@@ -8335,32 +8543,32 @@
             "support": 3.0
           },
           "transactional": {
-            "f1-score": 1.0,
-            "precision": 1.0,
-            "recall": 1.0,
+            "f1-score": 0.875,
+            "precision": 0.875,
+            "recall": 0.875,
             "support": 8.0
           },
           "weighted avg": {
-            "f1-score": 0.9360614458549377,
-            "precision": 0.9511068128089405,
-            "recall": 0.9361702127659575,
+            "f1-score": 0.8734633569739952,
+            "precision": 0.8914893617021277,
+            "recall": 0.8723404255319149,
             "support": 47.0
           }
         },
         "suite": "test"
       },
       "third_wave_cases": {
-        "accepted_accuracy": 0.8846,
+        "accepted_accuracy": 0.8462,
         "accepted_coverage": 1.0,
-        "accuracy": 0.8846,
+        "accuracy": 0.8462,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_third_wave_cases_confusion_matrix.csv",
         "count": 26,
         "dataset_path": "/content/agentic-intent-classifier/data/third_wave_cases.jsonl",
         "fallback_rate": 0.0,
         "head": "intent_type",
-        "macro_f1": 0.8209,
+        "macro_f1": 0.8148,
         "per_class_metrics": {
-          "accuracy": 0.8846153846153846,
+          "accuracy": 0.8461538461538461,
           "ambiguous": {
             "f1-score": 0.8235294117647058,
             "precision": 1.0,
@@ -8374,9 +8582,9 @@
             "support": 1.0
           },
           "commercial": {
-            "f1-score": 0.9230769230769231,
-            "precision": 0.8571428571428571,
-            "recall": 1.0,
+            "f1-score": 0.88,
+            "precision": 0.8461538461538461,
+            "recall": 0.9166666666666666,
             "support": 12.0
           },
           "creative_generation": {
@@ -8398,9 +8606,9 @@
             "support": 0.0
           },
           "macro avg": {
-            "f1-score": 0.5746606334841629,
-            "precision": 0.5857142857142857,
-            "recall": 0.5700000000000001,
+            "f1-score": 0.5703529411764705,
+            "precision": 0.5846153846153846,
+            "recall": 0.5616666666666666,
             "support": 26.0
           },
           "personal_reflection": {
@@ -8428,30 +8636,30 @@
             "support": 0.0
           },
           "weighted avg": {
-            "f1-score": 0.8966237382526975,
-            "precision": 0.9340659340659341,
-            "recall": 0.8846153846153846,
+            "f1-score": 0.8767420814479638,
+            "precision": 0.9289940828402367,
+            "recall": 0.8461538461538461,
             "support": 26.0
           }
         },
         "suite": "third_wave_cases"
       },
       "train": {
-        "accepted_accuracy": 1.0,
-        "accepted_coverage": 0.9945,
-        "accuracy": 1.0,
+        "accepted_accuracy": 0.9945,
+        "accepted_coverage": 1.0,
+        "accuracy": 0.9945,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_train_confusion_matrix.csv",
         "count": 183,
         "dataset_path": "/content/agentic-intent-classifier/data/train.jsonl",
-        "fallback_rate": 0.0055,
+        "fallback_rate": 0.0,
         "head": "intent_type",
-        "macro_f1": 1.0,
+        "macro_f1": 0.9936,
         "per_class_metrics": {
-          "accuracy": 1.0,
+          "accuracy": 0.994535519125683,
           "ambiguous": {
-            "f1-score": 1.0,
+            "f1-score": 0.9836065573770492,
             "precision": 1.0,
-            "recall": 1.0,
+            "recall": 0.967741935483871,
             "support": 31.0
           },
           "chit_chat": {
@@ -8485,9 +8693,9 @@
             "support": 38.0
           },
           "macro avg": {
-            "f1-score": 1.0,
-            "precision": 1.0,
-            "recall": 1.0,
+            "f1-score": 0.9935987509758002,
+            "precision": 0.990909090909091,
+            "recall": 0.9967741935483871,
             "support": 183.0
           },
           "personal_reflection": {
@@ -8503,8 +8711,8 @@
             "support": 5.0
           },
           "support": {
-            "f1-score": 1.0,
-            "precision": 1.0,
+            "f1-score": 0.9523809523809523,
+            "precision": 0.9090909090909091,
             "recall": 1.0,
             "support": 10.0
           },
@@ -8515,26 +8723,26 @@
             "support": 28.0
           },
           "weighted avg": {
-            "f1-score": 1.0,
-            "precision": 1.0,
-            "recall": 1.0,
+            "f1-score": 0.9946208349863281,
+            "precision": 0.9950322901142573,
+            "recall": 0.994535519125683,
             "support": 183.0
           }
         },
         "suite": "train"
       },
       "val": {
-        "accepted_accuracy": 0.9362,
-        "accepted_coverage": 1.0,
-        "accuracy": 0.9362,
+        "accepted_accuracy": 0.8913,
+        "accepted_coverage": 0.9787,
+        "accuracy": 0.8723,
         "confusion_matrix_path": "/content/agentic-intent-classifier/artifacts/evaluation/latest/intent_type_val_confusion_matrix.csv",
         "count": 47,
         "dataset_path": "/content/agentic-intent-classifier/data/val.jsonl",
-        "fallback_rate": 0.0,
+        "fallback_rate": 0.0213,
         "head": "intent_type",
-        "macro_f1": 0.9108,
+        "macro_f1": 0.8144,
         "per_class_metrics": {
-          "accuracy": 0.9361702127659575,
+          "accuracy": 0.8723404255319149,
           "ambiguous": {
             "f1-score": 0.9411764705882353,
             "precision": 1.0,
@@ -8554,8 +8762,8 @@
             "support": 10.0
           },
           "creative_generation": {
-            "f1-score": 0.6666666666666666,
-            "precision": 0.5,
+            "f1-score": 0.4,
+            "precision": 0.25,
             "recall": 1.0,
             "support": 1.0
           },
@@ -8572,9 +8780,9 @@
             "support": 8.0
           },
           "macro avg": {
-            "f1-score": 0.9107843137254902,
-            "precision": 0.89,
-            "recall": 0.966388888888889,
+            "f1-score": 0.8143740573152337,
+            "precision": 0.8150000000000001,
+            "recall": 0.9080555555555556,
             "support": 47.0
           },
           "personal_reflection": {
@@ -8584,27 +8792,27 @@
             "support": 5.0
           },
           "prohibited": {
-            "f1-score": 1.0,
-            "precision": 1.0,
+            "f1-score": 0.6666666666666666,
+            "precision": 0.5,
             "recall": 1.0,
             "support": 1.0
           },
           "support": {
-            "f1-score": 1.0,
+            "f1-score": 0.8,
             "precision": 1.0,
-            "recall": 1.0,
+            "recall": 0.6666666666666666,
             "support": 3.0
           },
           "transactional": {
-            "f1-score": 0.9333333333333333,
+            "f1-score": 0.7692307692307693,
             "precision": 1.0,
-            "recall": 0.875,
+            "recall": 0.625,
             "support": 8.0
           },
           "weighted avg": {
-            "f1-score": 0.9419274092615769,
-            "precision": 0.9574468085106383,
-            "recall": 0.9361702127659575,
+            "f1-score": 0.8884631430313531,
+            "precision": 0.9414893617021277,
+            "recall": 0.8723404255319149,
             "support": 47.0
           }
         },
diff --git a/iab_classifier_model_output/train_metrics.json b/iab_classifier_model_output/train_metrics.json
index cc536ed82a216975b70bc5163e96c341f2b3eb0f..e48d14dce1881226e11549a53e5473f9bb437521 100644
--- a/iab_classifier_model_output/train_metrics.json
+++ b/iab_classifier_model_output/train_metrics.json
@@ -4,22 +4,22 @@
   "test_count": 3282,
   "test_metrics": {
     "epoch": 3.0,
-    "test_accuracy": 0.9439366240097502,
-    "test_loss": 1.8114978075027466,
-    "test_macro_f1": 0.9150587607646185,
-    "test_runtime": 9.356,
-    "test_samples_per_second": 350.793,
-    "test_steps_per_second": 22.018
+    "test_accuracy": 0.9320536258379037,
+    "test_loss": 2.1225109100341797,
+    "test_macro_f1": 0.8928275998599801,
+    "test_runtime": 11.2269,
+    "test_samples_per_second": 292.333,
+    "test_steps_per_second": 18.349
   },
   "train_count": 13211,
   "val_count": 3282,
   "val_metrics": {
     "epoch": 3.0,
-    "val_accuracy": 0.9485070079219988,
-    "val_loss": 1.8056248426437378,
-    "val_macro_f1": 0.9206922853424929,
-    "val_runtime": 9.3831,
-    "val_samples_per_second": 349.776,
-    "val_steps_per_second": 21.954
+    "val_accuracy": 0.9320536258379037,
+    "val_loss": 2.119333505630493,
+    "val_macro_f1": 0.8965787473982275,
+    "val_runtime": 11.1915,
+    "val_samples_per_second": 293.257,
+    "val_steps_per_second": 18.407
   }
 }
diff --git a/multitask_intent_model_output/multitask_intent.onnx b/multitask_intent_model_output/multitask_intent.onnx
index 3677f5e5ac4a347bed2444d824241494775befb5..28b80ff95ef7944f5a03e24159faab14908f4bf7 100644
--- a/multitask_intent_model_output/multitask_intent.onnx
+++ b/multitask_intent_model_output/multitask_intent.onnx
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:62cd7051bfe36d521d9059be8df5f782e46c5c821d9674f424c0530633da23dd
+oid sha256:d536f910f1068119861d06d698a8c4ba100173abc4b192f42beeedf14c274707
 size 61398
diff --git a/multitask_intent_model_output/multitask_intent.onnx.data b/multitask_intent_model_output/multitask_intent.onnx.data
index b17ed51470ca36cce47139b7f7c990b0464b30fe..3c6c7fc6b66497998e8e479b4504c33c94f17e3f 100644
--- a/multitask_intent_model_output/multitask_intent.onnx.data
+++ b/multitask_intent_model_output/multitask_intent.onnx.data
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6e8bd2ab2f34a2f8add41671f3686866b66db8583964ea41559255e055a45ffa
+oid sha256:ed0f4674d1b56d1cd6e42129cf6d7f0542559c24c7bbcf93fc361f87a1d89488
 size 265598976
diff --git a/multitask_intent_model_output/multitask_model.pt b/multitask_intent_model_output/multitask_model.pt
index 05282eb1cbc6b84c146a354eebfeabb697cdd5f4..eddf246b410e9f0e0ac40e7f59def8e7b932e3bc 100644
--- a/multitask_intent_model_output/multitask_model.pt
+++ b/multitask_intent_model_output/multitask_model.pt
@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:40505a22ecb1bfce5491f1728716cdbdf4aa99001a74b0d95af1255277f36a59
+oid sha256:110aa9638912221fc67dc16d1452192a280c0651d6701726abedd9a1fa60fc9f
 size 265602027
diff --git a/multitask_intent_model_output/train_metrics.json b/multitask_intent_model_output/train_metrics.json
index 8aacd10b4efd9d3458906e1bb85d2856428441ec..37f548826f2deb50a7c1ff22ade8d1adec8d9ba1 100644
--- a/multitask_intent_model_output/train_metrics.json
+++ b/multitask_intent_model_output/train_metrics.json
@@ -8,30 +8,30 @@
   "test_count": 70,
   "test_metrics": {
     "epoch": 4.0,
-    "test_decision_phase_accuracy": 0.7931034482758621,
-    "test_decision_phase_macro_f1": 0.8010204081632653,
+    "test_decision_phase_accuracy": 0.7586206896551724,
+    "test_decision_phase_macro_f1": 0.763718820861678,
     "test_intent_subtype_accuracy": 0.8714285714285714,
-    "test_intent_subtype_macro_f1": 0.7758580399233755,
-    "test_intent_type_accuracy": 0.8936170212765957,
-    "test_intent_type_macro_f1": 0.8110224089635854,
-    "test_loss": 1.376851201057434,
-    "test_runtime": 0.1497,
-    "test_samples_per_second": 467.54,
-    "test_steps_per_second": 33.396
+    "test_intent_subtype_macro_f1": 0.8317236029317956,
+    "test_intent_type_accuracy": 0.8723404255319149,
+    "test_intent_type_macro_f1": 0.8005555555555555,
+    "test_loss": 1.4344456195831299,
+    "test_runtime": 0.1682,
+    "test_samples_per_second": 416.197,
+    "test_steps_per_second": 29.728
   },
   "train_count": 1590,
   "val_count": 473,
   "val_metrics": {
     "epoch": 4.0,
-    "val_decision_phase_accuracy": 0.9512195121951219,
-    "val_decision_phase_macro_f1": 0.9431640512510706,
-    "val_intent_subtype_accuracy": 0.8888888888888888,
-    "val_intent_subtype_macro_f1": 0.876672226881405,
-    "val_intent_type_accuracy": 0.9792387543252595,
-    "val_intent_type_macro_f1": 0.9727773461133127,
-    "val_loss": 0.5223116278648376,
-    "val_runtime": 0.9733,
-    "val_samples_per_second": 485.974,
-    "val_steps_per_second": 30.823
+    "val_decision_phase_accuracy": 0.9560975609756097,
+    "val_decision_phase_macro_f1": 0.9496568779026763,
+    "val_intent_subtype_accuracy": 0.8950617283950617,
+    "val_intent_subtype_macro_f1": 0.8822267346328656,
+    "val_intent_type_accuracy": 0.9757785467128027,
+    "val_intent_type_macro_f1": 0.970010435450997,
+    "val_loss": 0.5456064343452454,
+    "val_runtime": 1.1168,
+    "val_samples_per_second": 423.544,
+    "val_steps_per_second": 26.863
   }
 }
diff --git a/training/run_full_training_pipeline.py b/training/run_full_training_pipeline.py
index 29faa08b4472bfd19b830593db641bfcf9c0b6e6..7a47ca0b3977a565324614bcb295b2256d03a00c 100644
--- a/training/run_full_training_pipeline.py
+++ b/training/run_full_training_pipeline.py
@@ -21,6 +21,51 @@ def run_step(args: list[str]) -> None:
     print(f"    end:   {ended_wall}\n    took:  {elapsed_s:.2f}s")
 
 
+def verify_local_artifacts() -> None:
+    """Ensure complete local training artifacts were generated.
+
+    This check intentionally validates *local repo paths* so a complete run
+    cannot silently rely on hub cache/snapshot paths.
+    """
+    required_dirs = {
+        "multitask model dir": BASE_DIR / "multitask_intent_model_output",
+        "iab classifier dir": BASE_DIR / "iab_classifier_model_output",
+        "calibration dir": BASE_DIR / "artifacts" / "calibration",
+    }
+    required_files = {
+        "multitask weights": BASE_DIR / "multitask_intent_model_output" / "multitask_model.pt",
+        "multitask metadata": BASE_DIR / "multitask_intent_model_output" / "metadata.json",
+        "iab model config": BASE_DIR / "iab_classifier_model_output" / "config.json",
+        "iab model weights": BASE_DIR / "iab_classifier_model_output" / "model.safetensors",
+        "calibration intent_type": BASE_DIR / "artifacts" / "calibration" / "intent_type.json",
+        "calibration intent_subtype": BASE_DIR / "artifacts" / "calibration" / "intent_subtype.json",
+        "calibration decision_phase": BASE_DIR / "artifacts" / "calibration" / "decision_phase.json",
+        "calibration iab_content": BASE_DIR / "artifacts" / "calibration" / "iab_content.json",
+    }
+
+    missing: list[str] = []
+    for name, path in required_dirs.items():
+        if not path.is_dir():
+            missing.append(f"[MISS] {name}: {path}")
+        else:
+            print(f"[OK ] {name}: {path}")
+    for name, path in required_files.items():
+        if not path.is_file():
+            missing.append(f"[MISS] {name}: {path}")
+        else:
+            print(f"[OK ] {name}: {path}")
+
+    if missing:
+        print("\nLocal artifact verification failed:")
+        for line in missing:
+            print(line)
+        raise RuntimeError(
+            "Expected local training artifacts are missing. "
+            "Training finished, but outputs were not generated in the repo paths."
+        )
+    print("\nLocal artifact verification passed.")
+
+
 def main() -> None:
     pipeline_start = time.perf_counter()
     pipeline_start_wall = datetime.now(timezone.utc).isoformat()
@@ -116,6 +161,9 @@ def main() -> None:
     if args.smoke_test:
         run_step([python, "combined_inference.py", args.smoke_test_query])
 
+    # Always enforce local artifact presence for complete training reliability.
+    verify_local_artifacts()
+
     pipeline_elapsed_s = time.perf_counter() - pipeline_start
     pipeline_end_wall = datetime.now(timezone.utc).isoformat()
     print(
diff --git a/training/upload_to_hf.py b/training/upload_to_hf.py
index cb1d465cf12bc8acae118f6356d1b4d8df6eff5d..26a40ca6f6a56f4f445f4df8aaee70d948cb648f 100644
--- a/training/upload_to_hf.py
+++ b/training/upload_to_hf.py
@@ -10,11 +10,15 @@ from __future__ import annotations
 
 import argparse
 import os
+import shutil
 import sys
+import tempfile
 import time
 from datetime import datetime, timezone
 from pathlib import Path
 
+LARGE_FILE_UPLOAD_THRESHOLD_BYTES = 100 * 1024 * 1024
+
 
 def _parse_args() -> argparse.Namespace:
     parser = argparse.ArgumentParser(description="Upload trained intent/IAB artifacts to Hugging Face Hub.")
@@ -53,6 +57,24 @@ def _parse_args() -> argparse.Namespace:
         action="store_true",
         help="Upload a Hugging Face model card file as README.md in the Hub repo root.",
     )
+    parser.add_argument(
+        "--include-serving-code",
+        action="store_true",
+        help="Upload core runtime Python/code files required for Hub trust_remote_code inference.",
+    )
+    parser.add_argument(
+        "--include-root-checkpoint",
+        action="store_true",
+        help="Upload root-level compatibility checkpoint/tokenizer files used by transformers.pipeline loader.",
+    )
+    parser.add_argument(
+        "--include-all",
+        action="store_true",
+        help=(
+            "Upload everything needed for end-to-end Hub usage: multitask + iab + calibration + "
+            "HF README + serving code + root checkpoint/tokenizer files."
+        ),
+    )
     parser.add_argument(
         "--hf-readme-path",
         default="HF_MODEL_CARD.md",
@@ -81,6 +103,64 @@ def _parse_args() -> argparse.Namespace:
     return parser.parse_args()
 
 
+def _iter_local_files(path: Path) -> list[Path]:
+    if path.is_file():
+        return [path]
+    return sorted(p for p in path.rglob("*") if p.is_file())
+
+
+def _remote_file_paths(path_in_repo: str, local_path: Path) -> list[str]:
+    if local_path.is_file():
+        return [path_in_repo]
+    return [
+        f"{path_in_repo}/{file_path.relative_to(local_path).as_posix()}"
+        for file_path in _iter_local_files(local_path)
+    ]
+
+
+def _requires_large_upload(local_path: Path) -> bool:
+    return any(file_path.stat().st_size >= LARGE_FILE_UPLOAD_THRESHOLD_BYTES for file_path in _iter_local_files(local_path))
+
+
+def _upload_via_large_folder(api, repo_id: str, repo_path: str, local_path: Path) -> None:
+    with tempfile.TemporaryDirectory(prefix="hf_large_upload_") as tmp_dir:
+        staging_root = Path(tmp_dir)
+        staged_target = staging_root / repo_path
+        staged_target.parent.mkdir(parents=True, exist_ok=True)
+        if local_path.is_file():
+            shutil.copy2(local_path, staged_target)
+        else:
+            shutil.copytree(
+                local_path,
+                staged_target,
+                ignore=shutil.ignore_patterns(".cache", "__pycache__"),
+            )
+        # Ensure resumable-upload metadata from previous local attempts does not
+        # get carried into the fresh staging directory.
+        shutil.rmtree(staged_target / ".cache", ignore_errors=True)
+        api.upload_large_folder(
+            repo_id=repo_id,
+            repo_type="model",
+            folder_path=str(staging_root),
+            print_report=False,
+        )
+
+
+def _verify_remote_upload(api, repo_id: str, repo_path: str, local_path: Path) -> None:
+    expected = set(_remote_file_paths(repo_path, local_path))
+    for attempt in range(4):
+        files = set(api.list_repo_files(repo_id=repo_id, repo_type="model"))
+        missing = sorted(expected - files)
+        if not missing:
+            return
+        if attempt == 3:
+            raise RuntimeError(
+                "Upload completed but the following remote files are still missing: "
+                + ", ".join(missing[:20])
+            )
+        time.sleep(2 * (attempt + 1))
+
+
 def main() -> int:
     started_at = time.perf_counter()
     started_wall = datetime.now(timezone.utc).isoformat()
@@ -96,6 +176,14 @@ def main() -> int:
     calibration_dir = (repo_root / args.calibration_dir).resolve()
     hf_readme_path = (repo_root / args.hf_readme_path).resolve()
 
+    if args.include_all:
+        args.include_multitask = True
+        args.include_iab = True
+        args.include_calibration = True
+        args.include_hf_readme = True
+        args.include_serving_code = True
+        args.include_root_checkpoint = True
+
     to_upload: list[tuple[str, Path]] = []
     if args.include_multitask:
         to_upload.append(("multitask_intent_model_output", multitask_dir))
@@ -106,8 +194,43 @@ def main() -> int:
     if args.include_hf_readme:
         to_upload.append(("README.md", hf_readme_path))
 
+    if args.include_serving_code:
+        # Files needed by trust_remote_code execution path.
+        for rel in [
+            "pipeline.py",
+            "config.py",
+            "config.json",
+            "combined_inference.py",
+            "model_runtime.py",
+            "multitask_runtime.py",
+            "multitask_model.py",
+            "schemas.py",
+            "inference_intent_type.py",
+            "inference_subtype.py",
+            "inference_decision_phase.py",
+            "inference_iab_classifier.py",
+            "iab_classifier.py",
+            "iab_taxonomy.py",
+        ]:
+            to_upload.append((rel, (repo_root / rel).resolve()))
+
+    if args.include_root_checkpoint:
+        for rel in [
+            "model.safetensors",
+            "tokenizer.json",
+            "tokenizer_config.json",
+            "special_tokens_map.json",
+            "vocab.txt",
+        ]:
+            to_upload.append((rel, (repo_root / rel).resolve()))
+
     if not to_upload:
-        print("Nothing to upload. Pass --include-multitask, --include-iab, and/or --include-calibration.", file=sys.stderr)
+        print(
+            "Nothing to upload. Pass include flags (e.g. --include-all), or one/more of: "
+            "--include-multitask --include-iab --include-calibration --include-hf-readme "
+            "--include-serving-code --include-root-checkpoint.",
+            file=sys.stderr,
+        )
         return 2
 
     # Import lazily so `--dry-run` works without extra deps.
@@ -127,27 +250,26 @@ def main() -> int:
         if args.dry_run:
             print(f"[DRY] Would upload {local_dir} -> {args.repo_id}:{repo_path}")
             continue
-        # Upload single README.md file (Hub model card) vs directories
-        if repo_path == "README.md":
-            step_start = time.perf_counter()
-            print(f"[UPLOAD] {local_dir} -> {args.repo_id}:README.md")
+        step_start = time.perf_counter()
+        mode = "large-folder" if _requires_large_upload(local_dir) else "standard"
+        print(f"[UPLOAD] {local_dir} -> {args.repo_id}:{repo_path} ({mode})")
+        if mode == "large-folder":
+            _upload_via_large_folder(api, args.repo_id, repo_path, local_dir)
+        elif local_dir.is_file():
             api.upload_file(
                 repo_id=args.repo_id,
                 repo_type="model",
                 path_or_fileobj=str(local_dir),
-                path_in_repo="README.md",
+                path_in_repo=repo_path,
             )
-            print(f"[DONE ] README.md took {(time.perf_counter() - step_start):.2f}s")
-            continue
-
-        step_start = time.perf_counter()
-        print(f"[UPLOAD] {local_dir} -> {args.repo_id}:{repo_path}")
-        api.upload_folder(
-            repo_id=args.repo_id,
-            repo_type="model",
-            folder_path=str(local_dir),
-            path_in_repo=repo_path,
-        )
+        else:
+            api.upload_folder(
+                repo_id=args.repo_id,
+                repo_type="model",
+                folder_path=str(local_dir),
+                path_in_repo=repo_path,
+            )
+        _verify_remote_upload(api, args.repo_id, repo_path, local_dir)
         print(f"[DONE ] {repo_path} took {(time.perf_counter() - step_start):.2f}s")
 
     ended_wall = datetime.now(timezone.utc).isoformat()