Skip to content

fix(llm): stop bisecting hollow responses, retry the same chunk instead (#2880) - #2887

Closed
chintondutta wants to merge 2 commits into
Graphify-Labs:v8from
chintondutta:fix/2880-hollow-response-retry
Closed

fix(llm): stop bisecting hollow responses, retry the same chunk instead (#2880)#2887
chintondutta wants to merge 2 commits into
Graphify-Labs:v8from
chintondutta:fix/2880-hollow-response-retry

docs(llm): clarify hollow is a graphify-internal sentinel, not provid…

105c039
Select commit
Loading
Failed to load commit list.
Graphify Labs / Graphify Formal Verification completed Aug 20, 2026 in 0s

Formal verification incomplete

Compared b14b52e94ec3d9840413d81777f4c134eac0a40d (base) against 105c0395de2ab4249e9afb9b794b9eb7b3349f85 (head).

Covered 1/9 changed functions (5 abstained, 3 never run)

equivalent (proved) distinguished may-equivalent (sampled) unsupported error
0 0 1 2 3

0 function(s) beyond the cap and 3 skipped when the budget ran out.

  • graphify/llm\.py · _call_azure: error — harness produced no verdict (rc=124): timeout after 30s
  • graphify/llm\.py · _call_bedrock: unsupported — not verifiable: all 200 sampled inputs raised on both versions — the function never executed, so 'no divergence' would be vacuous (mostly ParamValidationError — names the real obstacle, not a sampling gap)
  • graphify/llm\.py · _call_claude: error — harness produced no verdict (rc=124): timeout after 30s
  • graphify/llm\.py · _call_claude_cli: unsupported — not verifiable: all 200 sampled inputs raised on both versions — the function never executed, so 'no divergence' would be vacuous (mostly RuntimeError — names the real obstacle, not a sampling gap)
  • graphify/llm\.py · _call_openai_compat: error — harness produced no verdict (rc=124): timeout after 30s

Formal verification is advisory: equivalent is a proof over a bounded sound subset; may_equivalent is sampled, not proven; unsupported and error are honest abstentions.