README.md

---
license: apache-2.0
language:
- en
pretty_name: MemoryATHENA Results
tags:
- experiment-metadata
- question-answering
- text-classification
- memory-augmented-language-model
---

# MemoryATHENA

This dataset repository is a compact release ledger of MemoryATHENA experiment metadata and aggregate results. It follows the release style of [OLAResearchX/XMemTransfer-Results](https://huggingface.co/datasets/OLAResearchX/XMemTransfer-Results). It is **not** a training corpus, benchmark mirror, or redistribution of task examples, labels, or prediction logs.

## What is released

- results.jsonl: 54 compact aggregate rows (30 QA condition/task rows and 24 general-NLP condition/task rows).
- metadata.json: release provenance, dataset references, model links, and exclusions.
- schema.json: row schema.

The repository intentionally excludes ARTIFACTS.md, internal filesystem paths, raw datasets, labels, predictions, optimizer states, and credentials.

## Training data

The model stages use causal text and a token budget. The reported training and validation counts below are **processed token positions**, not the number of raw documents. Downstream QA/NLP labels are never training targets.

| Component | Training source and split | Held-out validation used for checkpoint selection | Budget / selection | Source |
|---|---|---|---|---|
| QA reader/router | English Wikipedia 2021 causal-text stream | Held-out causal-text validation from the same Wikipedia-2021 training stream; this is not NQ/WebQuestions/TriviaQA validation | 20M processed positions per stage; 2M validation positions; QA router checkpoint selected at 8.192M processed positions | [Wikimedia English Wikipedia dumps](https://dumps.wikimedia.org/enwiki/20210101/) |
| General-NLP reader/router | Equal-token mixture of WikiText-103, Amazon Polarity, CC-News, and IMDB causal text | Held-out causal-text validation for the same general-text protocol; benchmark labels are not used for model selection | 19,998,720 processed positions per stage (20M budget); 2M validation positions; sequence length 2,048; general-NLP router checkpoint selected at 16.384M positions | [WikiText-103](https://huggingface.co/datasets/Salesforce/wikitext), [Amazon Polarity](https://huggingface.co/datasets/fancyzhx/amazon_polarity), [CC-News](https://huggingface.co/datasets/cc_news), [IMDB](https://huggingface.co/datasets/stanfordnlp/imdb) |
| Imported source memory | Released Llama-2 source-memory artifact from XMemTransfer | Not re-trained in this release | 20M-source-memory artifact; used as an input dependency for the QA line | [XMemTransfer result release](https://huggingface.co/datasets/OLAResearchX/XMemTransfer-Results) |

### Validation versus downstream evaluation

There are two different uses of the word “validation” in this release:

1. **Training validation** is held-out causal text used to select memory/router checkpoints. It uses no downstream benchmark labels.
2. **Downstream benchmark splits** are listed below. Their labels are used only for final metrics after inference, not to train memory, readers, or the router.

This distinction matters: a QA validation split is an evaluation benchmark split, whereas the 2M-position causal-text validation stream is the checkpoint-selection set.

## Downstream evaluation datasets

All released QA and general-NLP results are inference-only evaluations with frozen model components. The exact split and the number of examples actually scored are listed here.

### Five-task QA evaluation

Open-QA tasks report exact match (EM) and token F1. TruthfulQA reports MC1, MC2, MC3, and their arithmetic mean. NQ has 3,610 rows in the public validation split; the evaluator excludes one malformed answer-only row, so the released result scores 3,609 examples.

| Task | Dataset / configuration | Split actually evaluated | Examples scored | Metric | Dataset link |
|---|---|---:|---:|---|---|
| NQ | google-research-datasets/nq_open | validation | 3,609 | EM / F1 | [Natural Questions Open](https://huggingface.co/datasets/google-research-datasets/nq_open) |
| WebQA | Stanford/web_questions | test | 2,032 | EM / F1 | [WebQuestions](https://huggingface.co/datasets/Stanford/web_questions) |
| TriviaQA | mandarjoshi/trivia_qa, rc.nocontext | validation | 17,944 | EM / F1 | [TriviaQA](https://huggingface.co/datasets/mandarjoshi/trivia_qa) |
| TruthfulQA | truthfulqa/truthful_qa, multiple_choice | validation | 817 | MC1 / MC2 / MC3 / mean | [TruthfulQA](https://huggingface.co/datasets/truthfulqa/truthful_qa) |
| HotpotQA | hotpotqa/hotpot_qa, distractor | validation | 7,405 | EM / F1 | [HotpotQA](https://huggingface.co/datasets/hotpotqa/hotpot_qa) |

For the case-study and RAG diagnostics, HotpotQA uses distractor context. Supporting-fact annotations are not provided to the router or used to configure routing.

### Six-task general-NLP evaluation

The primary six-task table follows the exact public kNN-Prompt task files used by the evaluator. This is important for reproducibility: MR, CR, and RT are local protocol files rather than a claim that the HF SetFit/CR or another replacement dataset is identical.

| Task | Exact release used by the primary evaluator | Split / file used | Examples scored | Metric | Dataset/source link |
|---|---|---:|---:|---|---|
| SST2 | kNN-Prompt task data; GLUE SST-2 development set | dev.tsv / GLUE validation | 872 | Accuracy | [GLUE on HF](https://huggingface.co/datasets/nyu-mll/glue), [exact task files](https://github.com/swj0419/kNN_prompt/tree/main/task_data) |
| MR | kNN-Prompt task data | test.csv | 2,000 | Accuracy | [Exact task files](https://github.com/swj0419/kNN_prompt/tree/main/task_data) |
| CR | kNN-Prompt task data | test.csv | 2,000 | Accuracy | [Exact task files](https://github.com/swj0419/kNN_prompt/tree/main/task_data) |
| RT | kNN-Prompt task data / Rotten Tomatoes | test.jsonl / test | 1,066 | Accuracy | [Rotten Tomatoes on HF](https://huggingface.co/datasets/rotten_tomatoes), [exact task files](https://github.com/swj0419/kNN_prompt/tree/main/task_data) |
| AGN | AG News | test | 7,600 | Accuracy | [AG News](https://huggingface.co/datasets/fancyzhx/ag_news) |
| Yahoo | Yahoo Answers Topics | test | 60,000 | Accuracy | [Yahoo Answers Topics](https://huggingface.co/datasets/yahoo_answers_topics) |

The six-task score is an unweighted mean of the six accuracies. General-NLP scoring uses domain-conditional PMI with next-token log-probability sums over label synonyms. Labels are consumed only for final accuracy.

### Secondary HaluEval stress test

HaluEval is kept separate from the six-task macro-average because it uses a different binary factuality-classification protocol. The evaluator uses pminervini/HaluEval, configurations dialogue_samples, qa_samples, and summarization_samples, split data, with 10,000 examples per configuration.

- [HaluEval](https://huggingface.co/datasets/pminervini/HaluEval)

### Yahoo threshold note

The default router evaluation uses the configured threshold tau=0. A separate Yahoo tau=0.9/1.0 sweep is a post-hoc test-set diagnostic and must not be interpreted as an independently validation-selected threshold. It does not retrain the model and is reported separately from the default evaluation.

## Links

- Project page: [MemoryATHENA](https://www.olaresearch.org/MemoryATHENA)
- Code: [OLAResearch/ATHENA](https://github.com/OLAResearch/ATHENA)
- Collection: [MemoryATHENA](https://huggingface.co/collections/OLAResearchX/memoryathena)
- QA model: [OLAResearchX/memoryathena-qa-20260922](https://huggingface.co/OLAResearchX/memoryathena-qa-20260922)
- General-NLP model: [OLAResearchX/memoryathena-general-nlp-20260922](https://huggingface.co/OLAResearchX/memoryathena-general-nlp-20260922)
- Reference release: [OLAResearchX/XMemTransfer-Results](https://huggingface.co/datasets/OLAResearchX/XMemTransfer-Results)