FastLearn: A Rapid Learning Agent for Chat Models to Acquire Latest Knowledge
Bibliographic record
Abstract
Most large language models (LLMs), especially chat models, do not undergo updates post-deployment, which results in a lack of awareness of constantly changing new real-time information. External knowledge retrieval methods exist to enhance these models, but for high-traffic chat models, the extensive use of search engine API resources can be costly. In our work, we study the performance of LLMs in answering tasks involving real-time factual knowledge. Due to the lack of datasets for Q&A involving the latest real-time information, we first carefully constructed NewlyQA, a new dynamic QA benchmark including questions about world knowledge whose answers evolve over time. Inspired by human learning methods, we propose FastLearn, a simple rapid learning method that enhances model responses by merging relevant, up-to-date information retrieved from search engine APIs (filtered through our Time-Prompt mechanism) into prompts, followed by Lora fine-tuning to “inject” new knowledge into LLMs via external parameters, significantly improving performance on FreshQA, NewlyQA and OpenDataEval. Our experiments show that FastLearn outperforms most other search engine augmented prompt methods, like FreshPrompt, Self-Ask, and the commercial PERPLEXITY.AI. Additionally, we further analyze the two main mechanisms of FastLearn, Time-Prompt and Self-Correction, finding that both play key roles in enhancing model answer performance. To facilitate future research, we will open-source NewlyQA and commit to regularly updating its answers.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.005 | 0.023 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.006 |
| Open science | 0.005 | 0.004 |
| Research integrity | 0.002 | 0.004 |
| Insufficient payload (model declined to judge) | 0.012 | 0.006 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".