Upload folder using huggingface_hub

README.md

@@ -1,125 +1,15 @@
 ---
-dataset_info:
-  features:
-  - name: incident_id
-    dtype: string
-  - name: flight_id
-    dtype: string
-  - name: dataset_kind
-    dtype: string
-  - name: window_index
-    dtype: int64
-  - name: window_seconds
-    dtype: int64
-  - name: flight_phase
-    dtype: string
-  - name: fault_class
-    dtype: string
-  - name: flight_fault_class
-    dtype: string
-  - name: severity
-    dtype: string
-  - name: severity_ordinal
-    dtype: int64
-  - name: severity_score
-    dtype: float64
-  - name: active_mechanisms_json
-    dtype: string
-  - name: active_families_json
-    dtype: string
-  - name: scheduled_events_json
-    dtype: string
-  - name: fault_onset_window
-    dtype: int64
-  - name: is_compound
-    dtype: bool
-  - name: domain_shift
-    dtype: bool
-  - name: source_type
-    dtype: string
-  - name: approved_mitigation
-    dtype: string
-  - name: approved_inspection
-    dtype: string
-  - name: physics_validation_status
-    dtype: string
-  - name: action_training_eligible
-    dtype: bool
-  - name: consequence_control
-    dtype: float64
-  - name: consequence_power
-    dtype: float64
-  - name: consequence_navigation
-    dtype: float64
-  - name: consequence_altitude
-    dtype: float64
-  - name: altitude_mean_m
-    dtype: float64
-  - name: commanded_speed_mean_mps
-    dtype: float64
-  - name: voltage_mean_v
-    dtype: float64
-  - name: voltage_min_v
-    dtype: float64
-  - name: voltage_slope_v_s
-    dtype: float64
-  - name: current_mean_a
-    dtype: float64
-  - name: motor_output_mean
-    dtype: float64
-  - name: motor_output_std
-    dtype: float64
-  - name: motor_imbalance
-    dtype: float64
-  - name: accel_rms_g
-    dtype: float64
-  - name: gyro_rms_dps
-    dtype: float64
-  - name: attitude_error_mean_deg
-    dtype: float64
-  - name: vibration_rms_g
-    dtype: float64
-  - name: gps_satellites_mean
-    dtype: int64
-  - name: gps_hdop_mean
-    dtype: float64
-  - name: position_innovation_mean_m
-    dtype: float64
-  - name: baro_altitude_std_m
-    dtype: float64
-  - name: altitude_residual_mean_m
-    dtype: float64
-  - name: vertical_speed_error_mean_mps
-    dtype: float64
-  - name: wind_estimate_mean_mps
-    dtype: float64
-  - name: qwen_source_evidence_json
-    dtype: string
-  - name: evidence_hash
-    dtype: string
-  - name: symptom_report
-    dtype: string
-  - name: report_generation_method
-    dtype: string
-  - name: report_validator_pass
-    dtype: bool
-  - name: report_retry_count
-    dtype: int64
-  - name: report_model
-    dtype: string
-  - name: report_model_revision
-    dtype: string
-  - name: prompt_version
-    dtype: string
-  splits:
-  - name: benchmark
-    num_bytes: 13946536
-    num_examples: 10500
-  - name: challenge
-    num_bytes: 4237903
-    num_examples: 3500
-  download_size: 5678523
-  dataset_size: 18184439
+language:
+- en
+pretty_name: UAV Fault Symptom Reports (synthetic)
+size_categories:
+- 10K<n<100K
+task_categories:
+- text-classification
+tags:
+- synthetic
+- uav
+- telemetry
 configs:
 - config_name: default
   data_files:
@@ -128,3 +18,155 @@ configs:
   - split: challenge
     path: data/challenge-*
 ---
+
+# UAV Fault Symptom Reports
+
+A synthetic dataset of UAV flight telemetry paired with operator-style symptom reports written by a
+language model. Each row is one five-second window of a flight: 20 telemetry channels, the fault
+class, a severity derived from simulated consequences, and a one-sentence report.
+
+| split | rows | flights | model-written reports | unique reports |
+| --- | --- | --- | --- | --- |
+| benchmark | 10,500 | 2,100 | 91.5% | 82.6% |
+| challenge | 3,500 | 700 | 94.7% | 91.9% |
+
+**benchmark** is balanced across seven classes (six fault families plus normal).
+**challenge** is harder: mostly normal flights, random fault onset, and some flights with two faults.
+
+## How it was generated
+
+1. A mechanistic simulator produces telemetry from sampled aircraft, battery, sensor and
+   environment state. Fault events change those latent parameters, never the telemetry directly.
+2. Labels are read from the event log afterwards, so the fault class never builds the telemetry.
+3. Facts are derived from the telemetry against fixed thresholds.
+4. `Qwen/Qwen2.5-0.5B-Instruct` rewrites those facts as an operator report, using a
+   system prompt, few-shot examples, a JSON output format and explicit constraints.
+5. Every report is validated: length, no numbers, no diagnosis, no actions, no inferred causes, no
+   signals the model was not given, no chat-style text. Failures are retried twice, then replaced
+   by a templated sentence built from the same facts.
+
+The model never sees the fault class, the severity or the recommended action.
+
+## Exploratory data analysis
+
+### Instruction-following failures
+
+91.5% of reports were written by the model and accepted. The rest failed
+validation three times. Rejection reasons across all attempts:
+
+| reason                              |   rejections |   share_of_attempts |
+|:------------------------------------|-------------:|--------------------:|
+| Too short / too long                |         2642 |               0.192 |
+| Inferred a cause                    |         1208 |               0.088 |
+| Other                               |          271 |               0.02  |
+| Mentioned a signal it was not given |          216 |               0.016 |
+| Ignored the given signals           |          147 |               0.011 |
+| Invented a number                   |           61 |               0.004 |
+| Named a diagnosis                   |            1 |               0     |
+
+![Why model outputs were rejected](figures/e2_rejection_reasons.png)
+
+*Why model outputs were rejected*
+
+![Model-written share per fault class](figures/e1_model_share_by_class.png)
+
+*Model-written share per fault class*
+
+
+### Hallucination
+
+Accepted reports naming a signal they were not given: **0**. The validator blocks these.
+31.6% of accepted reports add no content word beyond their facts.
+
+### Inconsistency between text and label
+
+Each report describes one five-second window, and never sees the label. **36.4%** of fault
+windows show no threshold crossing and are described as steady; **15.8%** of normal
+windows cross one during ordinary manoeuvres and are described as irregular. These rows are where
+the text alone cannot identify the fault.
+
+![How each class is described by its text](figures/e4_text_vs_label.png)
+
+*How each class is described by its text*
+
+![Reports in TF-IDF space, coloured by class](figures/e8_tfidf_pca.png)
+
+*Reports in TF-IDF space, coloured by class*
+
+
+### Duplication, and how the EDA changed the generation
+
+The first full run had **42.8%** duplicate reports. The cause was the input:
+the evidence builder drew from a small fixed pool of facts, which capped how many different reports
+were possible. Widening the evidence phrasings brought duplicates down to
+**17.4%**. Counting reports that differ only in word order or function
+words as the same, the near-duplicate rate is
+**25.9%**.
+
+![First run vs. final run](figures/f_first_vs_final_run.png)
+
+*First run vs. final run*
+
+
+### Text statistics
+
+![Words per report](figures/e6_report_length.png)
+
+*Words per report*
+
+![Top content words and Zipf plot](figures/e7_word_frequency.png)
+
+*Top content words and Zipf plot*
+
+![Word cloud](figures/e7_wordcloud.png)
+
+*Word cloud*
+
+![Signal families mentioned together](figures/e7_cooccurrence.png)
+
+*Signal families mentioned together*
+
+
+### Telemetry
+
+Severity is scored per window, so it can escalate within a flight:
+**43.6%** of flights change severity across their five windows, and
+92.5% of those only escalate.
+
+![Telemetry correlations](figures/correlation_heatmap_readable.png)
+
+*Telemetry correlations*
+
+![Similarity between fault signatures](figures/rq2_ranked_fault_signature_similarity.png)
+
+*Similarity between fault signatures*
+
+![Persistence of fault evidence across windows](figures/rq6_persistence_by_fault_and_severity.png)
+
+*Persistence of fault evidence across windows*
+
+
+### Flagged issues
+
+| issue                                                                     | type                           |   rows affected |   share of rows | handling                                                                                |
+|:--------------------------------------------------------------------------|:-------------------------------|----------------:|----------------:|:----------------------------------------------------------------------------------------|
+| Model failed its instructions and fell back to a template                 | instruction following          |             896 |           0.085 | Kept, labelled canonical_grounded_fallback; the sentence states the same facts          |
+| Accepted report names a signal it was not given                           | hallucination                  |               0 |           0     | Blocked by the validator before acceptance                                              |
+| Fault window described as steady operation                                | inconsistency (text vs. label) |            2785 |           0.265 | Intended: the fault was not visible in that window. Flagged for the recommendation step |
+| Normal window described as irregular                                      | inconsistency (text vs. label) |             201 |           0.019 | Intended: ordinary manoeuvres cross thresholds. Flagged for the recommendation step     |
+| Exact duplicate report text                                               | duplication                    |            1823 |           0.174 | Reduced from the first run by widening the evidence (Part F)                            |
+| Near-duplicate report text (differs only in word order or function words) | duplication                    |            2719 |           0.259 | Flagged; each row still carries its own telemetry, but the text adds little on its own  |
+| Chat residue in a report                                                  | instruction following          |               0 |           0     | Blocked by the validator                                                                |
+
+## Limitations
+
+- The simulator is a transparent mechanistic prototype, not a validated flight-dynamics model.
+- Severity is a declared academic rubric, not a certified safety scale.
+- `approved_mitigation` and `approved_inspection` are fixed academic placeholders, one per fault
+  family; they are not maintenance procedures.
+
+## Files
+
+- `01_synthetic_data_generation.ipynb` - the generator
+- `02_dataset_and_eda.ipynb` - this analysis
+- `manifest.json` - seed, model revision, row counts and SHA-256 of each file