Soft HyDE asks an LLM for an outline, bulleted answer, or keyword-expanded pseudo-passage instead of a polished hypothetical document. Because the encoder needs topical fingerprint rather than prose quality, the shorter representation preserves retrieval signal while reducing tokens and cost. section 7.6.5
Deploy Soft HyDE with ordinary query-side and downstream safeguards:
- Rewrite the query before generating the hypothetical. section 7.6.5
- Combine the generated representation with BM25 in a hybrid pipeline, preserving exact identifier matches. section 7.6.5
- Cache generations against normalized queries so repeated questions avoid another LLM call. section 7.6.5
- Use a re-ranker, citation requirements, and answer-level evaluation to catch plausible but incorrectly framed retrieval. section 7.6.5
Support search
For a support corpus with repeated long-tail questions, normalize the query, generate a short Soft HyDE outline on a cache miss, retrieve through hybrid BM25 and semantic search, then rerank the results. section 7.6.5
Table 7.2 Comparison of HyDE variants and query-side techniques
|
Variation
|
Primary benefit
|
Implementation cost/complexity
|
Latency impact
|
Use case
|
|
Standard HyDE
|
Significant zero-shot retrieval improvement
|
Low
|
Medium
|
General-purpose RAG where baseline retrieval is poor.
|
|
Multi-Sample HyDE
|
Improved stability and robustness to noise
|
Medium (requires averaging)
|
High
|
Applications where query robustness is critical.
|
|
Hybrid HyDE (Fusion)
|
Combines semantic and lexical strengths
|
High (requires fusion logic)
|
Medium
|
High-precision search that cannot miss keyword matches.
|
|
HyDE + Re-ranker
|
Maximizes precision for top results
|
High (requires a second model)
|
High
|
Safety-critical or expert systems (e.g., legal, medical).
|
|
Reverse HyDE
|
Eliminates query-time LLM latency
|
Medium (shifts cost to indexing)
|
None at query time
|
Read-heavy corpora that change rarely (docs, catalogs,
KBs).
|
|
RQ-RAG
|
Built-in rewrite / decompose / disambiguate; strong on
multi-hop
|
High (requires supervised fine-tuning of the generator)
|
High (multiple sequential refinements per query)
|
Systems where query patterns are complex and a
custom-trained generator is acceptable.
|
Compare Soft HyDE's compact query representation with standard HyDE, multi-sample, hybrid, reranked, and reverse variants using benefit, complexity, latency, and use case. section 7.6.5
