Datasets:

ali5341
/

scitldr-chat-format

messages listlengths 2 2	meta dict
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: DppNet: Approximating Determinantal Point Processes with Deep Networks\n\nPaper content:\nDeterminantal Point Processes (DPPs) provide an...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "B1euhoAcKX", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Interactive Shape Based Brushing Technique for Trail Sets\n\nPaper content:\nBrushing techniques have a long history with the first inter...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "yaeJLwvTr", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: BERTgrid: Contextualized Embedding for 2D Document Representation and Understanding\n\nPaper content:\nFor understanding generic document...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "H1gsGaq9US", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Distilling Neural Networks for Faster and Greener Dependency Parsing\n\nPaper content:\nThe carbon footprint of natural language processi...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "Skx5ua4twS", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: \n\nPaper content:\nRecent research efforts enable study for natural language grounded navigation in photo-realistic environments, e.g., ...	{ "dataset": "scitldr", "split": "train", "source_variant": "FullText", "paper_id": "HkxzNpNtDS", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Learning Audio Features for Singer Identification and Embedding\n\nPaper content:\nThere has been an increasing use of neural networks fo...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "H1cT3NTBM", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Exploring Properties of the Deep Image Prior\n\nPaper content:\nThe Deep Image Prior (DIP, Ulyanov et al., 2017) is a fascinating recent ...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "HylijQ35IS", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Reward Design in Cooperative Multi-agent Reinforcement Learning for Packet Routing\n\nPaper content:\nIn cooperative multi-agent reinforc...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "r15kjpHa-", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Deep Imitative Models for Flexible Inference, Planning, and Control\n\nPaper content:\nImitation Learning (IL) is an appealing approach t...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "Skl4mRNYDr", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: BERT Goes to Law School: Quantifying the Competitive Advantage of Access to Large Legal Corpora in Contract Understanding\n\nPaper conten...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "rkeRMT9cLH", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Scaling shared model governance via model splitting\n\nPaper content:\nCurrently the only techniques for sharing governance of a deep lea...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "H1xEtoRqtQ", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Minimal Random Code Learning: Getting Bits Back from Compressed Model Parameters\n\nPaper content:\nWhile deep neural networks are a high...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "r1f0YiCctm", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Attention over Parameters for Dialogue Systems\n\nPaper content:\nDialogue systems require a great deal of different but complementary ex...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "BJepraEFPr", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: \n\nPaper content:\nCounterfactual Regret Minimization (CFR) is the most successful algorithm for finding approximate Nash equilibria in ...	{ "dataset": "scitldr", "split": "train", "source_variant": "FullText", "paper_id": "ByebT3EYvr", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: \n\nPaper content:\nRecent few-shot learning algorithms have enabled models to quickly adapt to new tasks based on only a few training sa...	{ "dataset": "scitldr", "split": "train", "source_variant": "FullText", "paper_id": "BJxDNxSFDH", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Discovering the mechanics of hidden neurons\n\nPaper content:\nNeural networks trained through stochastic gradient descent (SGD) have bee...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "H1srNebAZ", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Hindsight Trust Region Policy Optimization\n\nPaper content:\nAs reinforcement learning continues to drive machine intelligence beyond it...	{ "dataset": "scitldr", "split": "train", "source_variant": "A", "paper_id": "rylCP6NFDB", "target_index": 0, "num_targets": 1, "ic": false }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: Deep Innovation Protection\n\nPaper content:\nEvolutionary-based optimization approaches have recently shown promising results in domains...	{ "dataset": "scitldr", "split": "train", "source_variant": "AIC", "paper_id": "SygLu0VtPH", "target_index": 0, "num_targets": 1, "ic": true }
[ { "role": "user", "content": [ { "type": "text", "text": "Write a concise one-sentence TLDR summary of the scientific paper content below.\n\nTitle: \n\nPaper content:\nOver the passage of time Unmanned Autonomous Vehicles (UAVs), especially\nAutonomous flying drones grabbed a lot of a...	{ "dataset": "scitldr", "split": "train", "source_variant": "FullText", "paper_id": "Syx9rnRcYm", "target_index": 0, "num_targets": 1, "ic": false }

End of preview. Expand in Data Studio

SciTLDR (Chat-Format Preparation)

This dataset is a chat-format preparation of SciTLDR for summarization SFT.

Format

This format is commonly referred to as:

chat-format SFT data
instruction-tuning conversations
OpenAI-style messages format

Included files

train.jsonl
validation.jsonl
stats.json
prepare_scitldr_unsloth.py

Source

Base dataset: allenai/scitldr
Variants used:
- A
- AIC
- FullText

Original Dataset Highlights

Original dataset: allenai/scitldr
Focus: extreme summarization of scientific papers (TLDR generation).
Reported scale on source card: 5.4K TLDRs over ~3.2K papers.
Multi-target setup: each paper can have multiple valid TLDR summaries.
Paper: TLDR: Extreme Summarization of Scientific Documents

Preparation summary

Task: one-sentence scientific TLDR generation.
User input is built from paper title and source.
Assistant target is drawn from target.
Supports:
- target-policy first: first target only
- target-policy all: one row per target
Final train/validation splits are balanced across A, AIC, and FullText.

Schema

Each JSONL row contains:

messages
- user: instruction + title + paper content
- assistant: TLDR summary sentence
meta: split, source variant, paper_id, target index/count

Reproduction

python prepare_scitldr_unsloth.py --target-policy all

Downloads last month: 28

Paper for ali5341/scitldr-chat-format

TLDR: Extreme Summarization of Scientific Documents

Paper • 2004.15011 • Published Apr 30, 2020 • 2