Llm Tuning Patterns
- 457 installs
- 3.9k repo stars
- Updated January 26, 2026
- parcadei/continuous-claude-v3
llm-tuning-patterns is a Claude Code skill that supplies evidence-based LLM parameter and prompt presets for developers who need reliable model behavior across theorem proving, codegen, and reasoning tasks.
About
llm-tuning-patterns is a configuration skill from parcadei/continuous-claude-v3 that documents evidence-based LLM settings drawn from APOLLO parity analysis and Godel-Prover research. It prescribes concrete values such as max_tokens 4096, temperature 0.6, and top_p 0.95 for formal reasoning, plus task-specific tables for other workloads. Developers reach for llm-tuning-patterns when agent outputs are too terse, too random, or inconsistent across proof planning, tactic exploration, and implementation prompts. The skill is user-invocable false, so agents load it automatically when tuning generation behavior rather than when browsing docs manually.
- llm-tuning-patterns
Llm Tuning Patterns by the numbers
- 457 all-time installs (skills.sh)
- +2 installs in the week ending Aug 4, 2026 (Skillselion tracking)
- Ranked #910 of 4,347 Backend & APIs skills by installs in the Skillselion catalog
- Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/parcadei/continuous-claude-v3 --skill llm-tuning-patternsAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 457 |
|---|---|
| repo stars | ★ 3.9k |
| Last updated | January 26, 2026 |
| Repository | parcadei/continuous-claude-v3 ↗ |
What LLM temperature and token limits work for formal proofs?
Use llm-tuning-patterns for development tasks
Who is it for?
Developers running coding agents on formal reasoning, long chain-of-thought tasks, or mixed codegen workloads who want research-backed defaults instead of guessing hyperparameters.
Skip if: Teams that only need one fixed model profile with no task-specific tuning, or projects with no LLM-backed agent workflows.
When should I use this skill?
An agent is choosing temperature, max_tokens, top_p, or structured proof-plan prompts for a specialized reasoning or codegen task.
What you get
Task-specific parameter tables, proof-plan prompt templates, and rationale notes for agent LLM calls
- Parameter tables
- Proof-plan prompt pattern
By the numbers
- Recommends max_tokens 4096 for proof chain-of-thought
- Sets temperature 0.6 and top_p 0.95 for formal reasoning tasks
Files
LLM Tuning Patterns
Evidence-based patterns for configuring LLM parameters, based on APOLLO and Godel-Prover research.
Pattern
Different tasks require different LLM configurations. Use these evidence-based settings.
Theorem Proving / Formal Reasoning
Based on APOLLO parity analysis:
| Parameter | Value | Rationale |
|---|---|---|
| max_tokens | 4096 | Proofs need space for chain-of-thought |
| temperature | 0.6 | Higher creativity for tactic exploration |
| top_p | 0.95 | Allow diverse proof paths |
Proof Plan Prompt
Always request a proof plan before tactics:
Given the theorem to prove:
[theorem statement]
First, write a high-level proof plan explaining your approach.
Then, suggest Lean 4 tactics to implement each step.The proof plan (chain-of-thought) significantly improves tactic quality.
Parallel Sampling
For hard proofs, use parallel sampling:
- Generate N=8-32 candidate proof attempts
- Use best-of-N selection
- Each sample at temperature 0.6-0.8
Code Generation
| Parameter | Value | Rationale |
|---|---|---|
| max_tokens | 2048 | Sufficient for most functions |
| temperature | 0.2-0.4 | Prefer deterministic output |
Creative / Exploration Tasks
| Parameter | Value | Rationale |
|---|---|---|
| max_tokens | 4096 | Space for exploration |
| temperature | 0.8-1.0 | Maximum creativity |
Anti-Patterns
- Too low tokens for proofs: 512 tokens truncates chain-of-thought
- Too low temperature for proofs: 0.2 misses creative tactic paths
- No proof plan: Jumping to tactics without planning reduces success rate
Source Sessions
- This session: APOLLO parity - increased max_tokens 512->4096, temp 0.2->0.6
- This session: Added proof plan prompt for chain-of-thought before tactics
Related skills
Lark Openapi ExplorerInstantly explore, test, and generate calls against the full Lark (Feishu) OpenAPI surface without leaving their agent workflow.493k
Lark EventConsume real-time events from Lark/Feishu as structured NDJSON streams inside AI agent workflows.387k16kLark Openapi ExplorerWhen an existing Lark/Feishu skill or CLI command cannot fulfill a specific requirement and they need to discover and invoke the exact native OpenAPI endpoint.386k16k
Just ScrapeQuickly search, crawl, extract structured JSON, or monitor web pages without writing custom scraping code.245k41Lark AppsQuery the current visibility and permission scope of a Lark (Feishu) app without writing HTTP client code.235k16k
Design An InterfaceGenerate multiple radically different interface or API designs in parallel before committing to one.188k198kHow it compares
Pick llm-tuning-patterns over generic prompt tips when you need task-specific numeric LLM presets backed by formal-reasoning research rather than one-size-fits-all defaults.
FAQ
What temperature does llm-tuning-patterns recommend for formal proofs?
llm-tuning-patterns recommends temperature 0.6 for theorem proving and formal reasoning, based on APOLLO parity analysis, paired with max_tokens 4096 and top_p 0.95 to leave room for chain-of-thought and diverse tactic paths.
When should developers load llm-tuning-patterns?
Developers should load llm-tuning-patterns when configuring coding agents whose outputs vary by task type, especially formal reasoning, long proofs, or codegen that needs evidence-backed max_tokens, temperature, and top_p defaults.