README.md

---
pretty_name: Decision models × evaluation documents
license: other
task_categories:
- text-classification
size_categories:
- 100K<n<1M
tags:
- international-development
- evaluation
- multi-label
- decision-models
language:
- en
dataset_info:
- config_name: documents
  features:
  - name: document_id
    dtype: string
  - name: title
    dtype: string
  - name: text
    dtype: string
  - name: n_chars
    dtype: int64
  - name: truncated
    dtype: bool
  - name: evaluation_approach
    dtype: string
  - name: evaluation_type
    dtype: string
  - name: temporality
    dtype: string
  - name: themes
    list: string
  - name: countries
    list: string
  - name: evaluation_approach_pipeline
    dtype: string
  - name: evaluation_type_pipeline
    dtype: string
  - name: temporality_pipeline
    dtype: string
  - name: themes_pipeline
    list: string
  - name: countries_pipeline
    list: string
  - name: regions_pipeline
    list: string
  - name: label_source_pipeline
    dtype: string
  splits:
  - name: train
    num_bytes: 13068338
    num_examples: 1148
  - name: validation
    num_bytes: 1610901
    num_examples: 138
  - name: test
    num_bytes: 1547444
    num_examples: 134
  download_size: 8048167
  dataset_size: 16226683
- config_name: excerpts
  features:
  - name: excerpt_id
    dtype: large_string
  - name: document_id
    dtype: large_string
  - name: type
    dtype: large_string
  - name: section_category
    dtype: large_string
  - name: page
    dtype: int32
  - name: text
    dtype: large_string
  - name: themes
    list: string
  - name: regions
    list: string
  - name: countries
    list: string
  - name: methods
    list: string
  - name: eval_sample
    dtype: bool
  - name: themes_glm
    list: string
  - name: themes_deepseek
    list: string
  - name: themes_qwen
    list: string
  - name: themes_majority
    list: string
  - name: regions_glm
    list: string
  - name: regions_deepseek
    list: string
  - name: regions_qwen
    list: string
  - name: regions_majority
    list: string
  - name: countries_glm
    list: string
  - name: countries_deepseek
    list: string
  - name: countries_qwen
    list: string
  - name: countries_majority
    list: string
  - name: methods_glm
    list: string
  - name: methods_deepseek
    list: string
  - name: methods_qwen
    list: string
  - name: methods_majority
    list: string
  splits:
  - name: train
    num_bytes: 105407893
    num_examples: 157302
  - name: validation
    num_bytes: 12401757
    num_examples: 18000
  - name: test
    num_bytes: 11036653
    num_examples: 15834
  download_size: 52684858
  dataset_size: 128846303
- config_name: taxonomy
  features:
  - name: field
    dtype: large_string
  - name: code
    dtype: large_string
  - name: label
    dtype: large_string
  - name: definition
    dtype: large_string
  - name: definition_excerpts
    dtype: large_string
  - name: region
    dtype: large_string
  - name: in_documents
    dtype: bool
  - name: in_excerpts
    dtype: bool
  splits:
  - name: train
    num_bytes: 38866
    num_examples: 334
  download_size: 19512
  dataset_size: 38866
configs:
- config_name: documents
  data_files:
  - split: train
    path: documents/train-*
  - split: validation
    path: documents/validation-*
  - split: test
    path: documents/test-*
- config_name: excerpts
  data_files:
  - split: train
    path: excerpts/train-*
  - split: validation
    path: excerpts/validation-*
  - split: test
    path: excerpts/test-*
- config_name: taxonomy
  data_files:
  - split: train
    path: taxonomy/train-*
---

# Decision models × evaluation documents

1,420 publicly published international development evaluation reports (about 40 organisations), with document labels and excerpt tags. Document labels come in two sets: silver labels from GLM-5.3-Flash (default columns) and the original pipeline labels (`*_pipeline`). Built for [baobab-tech/decision-models-experiments](https://github.com/baobab-tech/decision-models-experiments): experiment [01](https://github.com/baobab-tech/decision-models-experiments/tree/main/experiments/01-many-option-classification) (zero-shot many-option classification) and [02](https://github.com/baobab-tech/decision-models-experiments/tree/main/experiments/02-fine-tuning) (fine-tuning).

Reports were converted from PDF with Docling and labelled by an LLM ingestion pipeline. The dataset is built by [`build_evaluation_docs_dataset.py`](https://github.com/baobab-tech/decision-models-experiments/blob/main/experiments/common/build_evaluation_docs_dataset.py) from `baobabtech/evalexplorer-data` revision `5315eab`; label names and definitions come from the pipeline's taxonomy and its document and excerpt prompts (eval-explorer commit `6ae90f7` for the excerpt prompts).

```python
from datasets import load_dataset
docs = load_dataset("baobabtech/decision-models-evaluation-docs", "documents", split="test")
excerpts = load_dataset("baobabtech/decision-models-evaluation-docs", "excerpts", split="test").filter(lambda r: r["eval_sample"])
taxonomy = load_dataset("baobabtech/decision-models-evaluation-docs", "taxonomy", split="train")
```

## Configs

| Config | Rows (train / validation / test) | One row per |
|---|---|---|
| `documents` | 1,148 / 138 / 134 | report: first pages + document labels |
| `excerpts` | 157,302 / 18,000 / 15,834 | finding, recommendation or methodology excerpt + tags |
| `taxonomy` | 334 (`train`) | label code |

Splits are by document (80/10/10 on a hash of `document_id`), shared by both configs.

### `documents`

| Column | Contents |
|---|---|
| `document_id`, `title` | Report ID and title |
| `text` | First pages, exactly as the `classify_codes` config fed them to classifiers: 2 pages for reports under 10 pages, otherwise 5; cut at 24,000 characters. Median 1,935 tokens |
| `n_chars`, `truncated` | Length before truncation; whether it was cut (92 of 1,420) |
| `evaluation_approach`, `evaluation_type`, `temporality` | Silver labels from GLM-5.3-Flash (high reasoning effort, codes with one-line definitions, full first pages): one code or null. Test nulls: 24 / 10 / 37 of 134 |
| `themes`, `countries` | Silver labels: themes (1–4 of 18 codes) and ISO 3166-1 alpha-2 countries |
| `*_pipeline` | The ingestion pipeline's labels (Gemini 2.5 Flash, gpt-oss-120b, Qwen 3 235B): `evaluation_approach_pipeline`, `evaluation_type_pipeline`, `temporality_pipeline`, `themes_pipeline`, `countries_pipeline`, `regions_pipeline` |
| `label_source_pipeline` | `ai`, or `manual` for the 36 documents whose pipeline labels were corrected by hand |

### `excerpts`

| Column | Contents |
|---|---|
| `excerpt_id`, `document_id`, `type`, `section_category`, `page` | Identity and position |
| `text` | Verbatim excerpt; median 34 words, p90 107 |
| `themes` (22 codes), `regions` (17), `countries` (198) | Pipeline labels, findings and recommendations. Tagged with the whole section as input |
| `methods` (24 codes) | Pipeline labels, methodology excerpts |
| `eval_sample` | `true` for the fixed experiment-01 test sample: 300 findings, 150 recommendations, 150 methodology, seed 0 |
| `<field>_glm`, `<field>_deepseek`, `<field>_qwen` | `eval_sample` rows only: labels from GLM-5.3-Flash, DeepSeek-V4.1-Flash and Qwen3.8-Flash-Next, given the excerpt alone, with classify-only prompts rebuilt from the pipeline's ([prompts](https://github.com/baobab-tech/decision-models-experiments/blob/main/experiments/common/prompts/excerpt-tagging.md)). Generated 2026-10-03 |
| `<field>_majority` | `eval_sample` rows only: labels chosen by at least 2 of the 3 LLMs. The experiment-01 reference |

### `taxonomy`

`field`, `code`, `label`, `definition` (document fields only, from the document classification prompt), `definition_excerpts` (themes and methods, from the excerpt extract-and-classify prompts; longer theme definitions), `region` (countries only), `in_documents`, `in_excerpts`. The two flags mark the codes that occur in the silver or pipeline labels; experiments ask over those sets.

## Labels

- **Silver (default):** GLM-5.3-Flash relabelled every document through HF Inference Providers. It returns `null` more often than the pipeline (test: approach 18%, temporality 28%).
- **Pipeline (`*_pipeline`):** the ingestion pipeline's LLM output; 36 document classifications corrected by hand.
- Pipeline labels score 76.2 mean field score against silver on test (exact match 9.0%).
- Excerpt tags are pipeline output only.
- Codes seen in fewer than 20 documents were dropped upstream.

## Baselines on the same test split

Earlier Baobab Tech classifier runs on `documents` test (134), mean field score 0–100, against pipeline labels / against silver:

| Model | Zero-shot | Fine-tuned (pipeline / silver) |
|---|---|---|
| Gemma 4 26B-A4B | 70.1 / 72.9 | 84.4 / 80.3 |
| Qwen3.5 4B | 67.1 / 66.2 | 84.7 / 77.8 |
| Gemma 4 E4B | 72.3 / 72.1 | 83.0 / 76.8 |
| GLiNER2.5 base | 45.4 / 45.2 | 58.4 / 57.2 |

## Licence

The reports are publicly published by their organisations, which keep their rights; check each report's terms before reusing its text beyond research. Labels, splits and taxonomy are released by Baobab Tech. No labels in this dataset are human gold labels except the 36 hand-corrected documents (`label_source_pipeline == "manual"`); all others are LLM output.