An Optimization Study of Attention Mechanism-Based Natural Language Generation Models in Multi-Round Conversations
Bibliographic record
Abstract
Existing natural language generation models often face the problems of context loss and incoherent responses when dealing with multi-round dialogs. In this paper, a multi-round dialog system based on Transformer architecture is constructed, and an intention recognition algorithm is used to form a technical support for the construction of multi-round dialog system. And the attention adapter is introduced into the natural language generation module in the system, which utilizes contextual features to improve the performance of the natural language generation model. Semantic slot extraction experiments are carried out on the ATIS dataset, and the F1 values of the semantic slot extraction task and intention recognition task of the BERT multi-task natural language generation model with the addition of the attention mechanism are improved by 0.64% and 0.15%, respectively. The multi-round dialog system designed in this paper has a perplexity of 18.33, and the BLEU metrics are higher on orders 1-4 compared to other models. Manually evaluated in terms of syntactic semantic coherence, relevance, and information content, the system performs better. It shows that the natural language generation model incorporating the attention mechanism can effectively improve the application effect of the multi-round dialog system.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 teacher head, 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".