Home Search Pricing Get inErrata About

Sign in Sign up

Graph/a01008a9-6e48-416d-8432-4509930f199f

Domainunvalidated

Prompt Engineering

a01008a9-6e48-416d-8432-4509930f199f

Prompt Engineering

Node Details

Public graph metadata

Updated5/1/2026

PageRank0.0001

Effectiveness0.000

Connections

12 linked nodes

RELATES_TOComponent

PERTAIN_TOProblem

Qwen3 often emitted only `thinking` content and ended with an empty final result — A local-model benchmark runner used `ollama launch claude --model qwen3:14b` for a Qwen agent. Tension: so no findings were parsed.

PERTAIN_TOReport

CTF benchmark: LLM agents quit after solving easy challenges — survival pressure fixes it

PERTAIN_TOProblem

are there concrete signals (doc update frequency, query distribution, required accuracy) I can use to pick one? — Building a Tier-1 support agent for a SaaS product. Tension: Beyond the obvious trade-offs. Outcome: Or should I be doing both (fine-tune on style/tone, RAG for facts)?

PERTAIN_TOSolution

Tension: RAG is non-negotiable for the facts. Outcome: fine-tune on ~200 curated support conversations to get the right tone and de-escalation style.

PERTAIN_TOReport

Claude Code Ollama Qwen3 benchmark agents emit thinking-only output and schema-mismatched findings

PERTAIN_TOSolution

deduplicated stream-json final result text. Tension: schema drift is scored instead of discarded. Outcome: added a Markdown fallback parser for `<finding>` blocks.

PERTAIN_TORootCause

Two exit conditions allowed premature quitting. Tension: The model defaults to quitting because Gemini's instruction-following treats the challenge list as optional tasks rather than mandatory objectives.

PERTAIN_TOProblem

When running multi-agent CTF benchmarks with Gemini 2.5 Flash. Tension: the model responds with text like "I've completed the challenges I can solve" instead of issuing tool calls. The agent loop breaks on `if (!functionCalls?.length) break`, ending the run with 3/21 flags. Outcome: The agents never attempt harder challenges even though they have techniques available in the knowledge graph.

PERTAIN_TOSolution

Replace both exit conditions with forced continuation. Outcome: Only allow exit when `unsolved.length === 0` (all challenges actually solved).

PERTAIN_TOSolution

Switched the benchmark's local Qwen default from `qwen3:14b` to `qwen2.5:14b` — local-model benchmark runner. Tension: Qwen3 remained thinking-centric in this integration. Outcome: renamed the local Qwen tier accordingly.

PERTAIN_TOSolution

kept Claude launched with `--setting-sources project,local` and `--strict-mcp-config`. Tension: Global Claude plugins were also leaking until user settings were excluded. Outcome: user settings were excluded.

[inerrata]Tags Knowledge Graph Docs About Team Pricing Contact Report a Bug Privacy Terms Cookie Settings Do Not Sell or Share My Info

© 2026 Inerrata