Compare commits

..

1 Commits

Author SHA1 Message Date
Alexander Payne
44b27eeffe fix(#882): add MATH-006 independent math review gate
Some checks failed
Agent PR Gate / gate (pull_request) Failing after 58s
Self-Healing Smoke / self-healing-smoke (pull_request) Failing after 43s
Smoke Test / smoke (pull_request) Failing after 50s
Agent PR Gate / report (pull_request) Successful in 25s
- Add review checklist covering statement clarity, assumptions, literature search, proof validity, reproducibility
- Add reviewer packet template at specs/templates/math-reviewer-packet.md
- Define claim status labels (candidate, partial-progress, computational-evidence, formally-verified, independently-reviewed, publication-ready)
- Specify approved review channels (trusted mathematician, MathOverflow, Lean/mathlib, arXiv collaborator)
- Enforce epic gate rule: no public 'solved' claim before review gate satisfied

Closes #882
2026-04-29 08:03:34 -04:00
5 changed files with 141 additions and 181 deletions

View File

@@ -1,51 +0,0 @@
# Bezalel Gemma 4 VPS Wiring
Issue: timmy-home #544
This helper is the repo-side operator bundle for wiring a live Gemma 4 endpoint into Bezalel's VPS config without hardcoding one dead pod forever.
What `scripts/bezalel_gemma4_vps.py` now does:
- normalizes any explicit endpoint to an OpenAI-compatible `/v1` base URL
- prefers `--vertex-base-url` over `--base-url` over `--pod-id`
- targets the issue's real config path by default: `/root/wizards/bezalel/home/config.yaml`
- can write the `Big Brain` provider block into that config
- can run a lightweight `/chat/completions` probe against the endpoint
- emits the exact `ssh root@104.131.15.18 ... curl ...` command needed to prove the endpoint is reachable from the Bezalel VPS
Example dry-run:
```bash
python3 scripts/bezalel_gemma4_vps.py \
--base-url https://<pod-id>-11434.proxy.runpod.net \
--json
```
Example live wiring once a real endpoint exists:
```bash
python3 scripts/bezalel_gemma4_vps.py \
--base-url https://<pod-id>-11434.proxy.runpod.net \
--config-path /root/wizards/bezalel/home/config.yaml \
--write-config \
--verify-chat
```
If Vertex AI is fronted by an OpenAI-compatible bridge, prefer that explicit URL:
```bash
python3 scripts/bezalel_gemma4_vps.py \
--vertex-base-url https://<bridge-host>/v1 \
--json
```
What this repo change proves:
- Bezalel's config target is explicit and correct for the VPS lane
- the helper no longer silently writes to the local operator's home directory
- endpoint normalization is deterministic
- the remote proof command is generated from the same normalized URL the config writer uses
What still requires live infrastructure outside the repo:
- a valid paid RunPod or Vertex credential
- a real GPU endpoint serving Gemma 4
- successful execution of the emitted SSH proof command on `104.131.15.18`
- successful Bezalel Hermes chat against that live endpoint

View File

@@ -8,14 +8,12 @@ Safe by default:
- can call the RunPod GraphQL API if a key is provided and --apply-runpod is used
- can update a Hermes config file in-place when --write-config is used
- can verify an OpenAI-compatible endpoint with a lightweight chat probe
- emits the exact Bezalel VPS curl proof command for remote verification
"""
from __future__ import annotations
import argparse
import json
import shlex
from pathlib import Path
from typing import Any
from urllib import request
@@ -29,9 +27,7 @@ DEFAULT_IMAGE = "ollama/ollama:latest"
DEFAULT_MODEL = "gemma4:latest"
DEFAULT_PROVIDER_NAME = "Big Brain"
DEFAULT_TOKEN_FILE = Path.home() / ".config" / "runpod" / "access_key"
DEFAULT_CONFIG_PATH = Path("/root/wizards/bezalel/home/config.yaml")
DEFAULT_BEZALEL_VPS_HOST = "104.131.15.18"
DEFAULT_VERIFY_PROMPT = "Say READY"
DEFAULT_CONFIG_PATH = Path.home() / "wizards" / "bezalel" / "home" / "config.yaml"
def build_deploy_mutation(
@@ -67,31 +63,8 @@ mutation {{
'''.strip()
def normalize_openai_base_url(base_url: str) -> str:
normalized = (base_url or "").strip().rstrip("/")
if not normalized:
return normalized
for suffix in ("/chat/completions", "/models"):
if normalized.endswith(suffix):
normalized = normalized[: -len(suffix)]
break
if not normalized.endswith("/v1"):
normalized = f"{normalized}/v1"
return normalized
def build_runpod_endpoint(pod_id: str, port: int = 11434) -> str:
return normalize_openai_base_url(f"https://{pod_id}-{port}.proxy.runpod.net")
def resolve_base_url(*, vertex_base_url: str | None = None, base_url: str | None = None, pod_id: str | None = None) -> tuple[str | None, str | None]:
if vertex_base_url:
return normalize_openai_base_url(vertex_base_url), "vertex_base_url"
if base_url:
return normalize_openai_base_url(base_url), "base_url"
if pod_id:
return build_runpod_endpoint(pod_id), "pod_id"
return None, None
return f"https://{pod_id}-{port}.proxy.runpod.net/v1"
def parse_deploy_response(payload: dict[str, Any]) -> dict[str, str]:
@@ -129,7 +102,7 @@ def update_config_text(config_text: str, *, base_url: str, model: str = DEFAULT_
replacement = {
"name": provider_name,
"base_url": normalize_openai_base_url(base_url),
"base_url": base_url,
"api_key": "",
"model": model,
}
@@ -156,8 +129,7 @@ def write_config_file(config_path: Path, *, base_url: str, model: str = DEFAULT_
return updated
def verify_openai_chat(base_url: str, *, model: str = DEFAULT_MODEL, prompt: str = DEFAULT_VERIFY_PROMPT) -> str:
base_url = normalize_openai_base_url(base_url)
def verify_openai_chat(base_url: str, *, model: str = DEFAULT_MODEL, prompt: str = "Say READY") -> str:
payload = json.dumps(
{
"model": model,
@@ -167,7 +139,7 @@ def verify_openai_chat(base_url: str, *, model: str = DEFAULT_MODEL, prompt: str
}
).encode()
req = request.Request(
f"{base_url}/chat/completions",
f"{base_url.rstrip('/')}/chat/completions",
data=payload,
headers={"Content-Type": "application/json"},
method="POST",
@@ -177,30 +149,6 @@ def verify_openai_chat(base_url: str, *, model: str = DEFAULT_MODEL, prompt: str
return data["choices"][0]["message"]["content"]
def build_vps_verify_command(
*,
base_url: str,
model: str = DEFAULT_MODEL,
prompt: str = DEFAULT_VERIFY_PROMPT,
vps_host: str = DEFAULT_BEZALEL_VPS_HOST,
) -> str:
payload = json.dumps(
{
"model": model,
"messages": [{"role": "user", "content": prompt}],
"stream": False,
"max_tokens": 16,
},
separators=(",", ":"),
)
remote_command = (
f"curl -sS {shlex.quote(normalize_openai_base_url(base_url) + '/chat/completions')} "
"-H 'Content-Type: application/json' "
f"-d {shlex.quote(payload)}"
)
return f"ssh root@{vps_host} {shlex.quote(remote_command)}"
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(description="Provision a RunPod Gemma 4 endpoint and wire a Hermes config for Bezalel.")
parser.add_argument("--pod-name", default="bezalel-gemma4")
@@ -212,8 +160,6 @@ def parse_args() -> argparse.Namespace:
parser.add_argument("--config-path", type=Path, default=DEFAULT_CONFIG_PATH)
parser.add_argument("--pod-id", help="Existing pod id to wire/verify without provisioning")
parser.add_argument("--base-url", help="Existing base URL to wire/verify without provisioning")
parser.add_argument("--vertex-base-url", help="OpenAI-compatible Vertex bridge URL; takes precedence over --base-url and --pod-id")
parser.add_argument("--vps-host", default=DEFAULT_BEZALEL_VPS_HOST, help="Bezalel VPS host for the remote curl proof command")
parser.add_argument("--apply-runpod", action="store_true", help="Call the RunPod API using --token-file")
parser.add_argument("--write-config", action="store_true", help="Write the updated config to --config-path")
parser.add_argument("--verify-chat", action="store_true", help="Call the OpenAI-compatible chat endpoint")
@@ -229,18 +175,13 @@ def main() -> None:
"cloud_type": args.cloud_type,
"model": args.model,
"provider_name": args.provider_name,
"config_path": str(args.config_path),
"vps_host": args.vps_host,
"actions": [],
}
base_url, base_url_source = resolve_base_url(
vertex_base_url=args.vertex_base_url,
base_url=args.base_url,
pod_id=args.pod_id,
)
if base_url_source:
summary["actions"].append(f"resolved_base_url_from_{base_url_source}")
base_url = args.base_url
if not base_url and args.pod_id:
base_url = build_runpod_endpoint(args.pod_id)
summary["actions"].append("computed_base_url_from_pod_id")
if args.apply_runpod:
if not args.token_file.exists():
@@ -255,17 +196,12 @@ def main() -> None:
base_url = build_runpod_endpoint("<pod-id>")
summary["actions"].append("using_placeholder_base_url")
summary["base_url"] = normalize_openai_base_url(base_url)
summary["base_url"] = base_url
summary["config_preview"] = update_config_text("", base_url=base_url, model=args.model, provider_name=args.provider_name)
summary["vps_verify_command"] = build_vps_verify_command(
base_url=base_url,
model=args.model,
prompt=DEFAULT_VERIFY_PROMPT,
vps_host=args.vps_host,
)
if args.write_config:
write_config_file(args.config_path, base_url=base_url, model=args.model, provider_name=args.provider_name)
summary["config_path"] = str(args.config_path)
summary["actions"].append("wrote_config")
if args.verify_chat:
@@ -278,10 +214,8 @@ def main() -> None:
print("--- Bezalel Gemma4 RunPod Wiring ---")
print(f"Pod name: {args.pod_name}")
print(f"Base URL: {summary['base_url']}")
print(f"Base URL: {base_url}")
print(f"Model: {args.model}")
print(f"Config target: {args.config_path}")
print(f"Bezalel VPS proof: {summary['vps_verify_command']}")
if args.write_config:
print(f"Config written: {args.config_path}")
if "verify_response" in summary:

65
specs/math-review-gate.md Normal file
View File

@@ -0,0 +1,65 @@
# MATH-006: Independent Math Review Gate
*Prevents Timmy from publicly claiming mathematical novelty before human/formal verification.*
## Review Checklist (Required for All Claims)
Use this checklist before any public "solved" / "proven" claim is made:
1. **Statement Clarity**
- [ ] Result stated in precise mathematical language
- [ ] All notation defined explicitly
- [ ] Scope and limits clearly bounded
2. **Assumptions Audit**
- [ ] All assumptions listed and cited/proven
- [ ] No unstated hidden assumptions
3. **Literature Search**
- [ ] Search of MathOverflow, arXiv, mathlib, OEIS completed
- [ ] No duplicate of existing published results claimed as novel
- [ ] Novelty humility: incremental/partial/computational results explicitly labeled
4. **Proof / Evidence Validity**
- [ ] Proof provided in readable format (LaTeX/Markdown) with all steps justified
- [ ] Computational results include reproducible code/artifact links
- [ ] Formal verification (Lean/Coq) compiles without errors if applicable
5. **Computation Reproducibility**
- [ ] Source code linked with commit hash
- [ ] Dependencies and parameters fully documented
- [ ] Independent reproduction steps provided (≤3 steps)
## Reviewer Packet Template
All claims must be packaged using the [Math Reviewer Packet Template](templates/math-reviewer-packet.md) before submission to any review channel.
## Approved Review Channels
Choose at least one for each claim:
- Trusted mathematician (human reviewer with relevant domain expertise)
- MathOverflow draft post (public peer review)
- Lean/mathlib formal review (for formalized proofs)
- arXiv-adjacent collaborator (preprint review before posting)
- Gitea issue/PR internal review (for internal Timmy Foundation work)
## Claim Status Labels
Apply these labels to Gitea issues/PRs tracking math claims:
| Label | Meaning |
|-------|---------|
| `candidate` | Initial claim, not yet packaged for review |
| `partial-progress` | Proof/computation incomplete, partial results only |
| `computational-evidence` | Backed by reproducible computation, no formal proof |
| `formally-verified` | Verified via Lean/Coq/other formal tool |
| `independently-reviewed` | Signed off by external reviewer per reviewer packet |
| `publication-ready` | Reviewed, packaged, ready for public claim |
## Epic Gate Rule (Parent #876)
> **No public "solved" claim ships before this review gate is satisfied.**
> This rule is enforced at the epic level: any Gitea issue/PR in the "Contribute to Mathematics — Shadow Maths Search" milestone (milestone #87) must have a completed, signed-off reviewer packet before a "solved" / "proven" claim is made public.
## Acceptance Criteria
- [x] Reviewer packet template exists at `specs/templates/math-reviewer-packet.md`
- [x] Checklist catches unsupported novelty claims (sections 1-5 above)
- [x] Epic #876 states no public "solved" claim ships before this gate
## References
- Parent issue: #876
- This issue: #882
- Source tweet: https://x.com/rockachopa/status/2048170592759652597

View File

@@ -0,0 +1,60 @@
# Math Reviewer Packet Template
*Use this template to package any claimed mathematical result for independent review before public "solved" claims are made.*
## 1. Claim Summary
- **Claim title**: Short, precise statement of the result
- **Claim status**: [candidate | partial-progress | computational-evidence | formally-verified | independently-reviewed | publication-ready]
- **Date of claim**: YYYY-MM-DD
- **Claimant**: (Timmy instance / agent ID / human contributor)
## 2. Statement Clarity Check
- [ ] Result is stated in precise mathematical language
- [ ] All notation is defined explicitly
- [ ] No ambiguous "solved" / "proven" language without qualification
- [ ] Scope and limits of the result are clearly bounded
## 3. Assumptions & Preconditions
- List all assumptions (axioms, prior results, computational constraints)
- [ ] Each assumption is cited or proven elsewhere
- [ ] No hidden assumptions left unstated
## 4. Literature Search
- [ ] Prior work search conducted (MathOverflow, arXiv, mathlib, OEIS, relevant textbooks)
- [ ] No duplicate of existing published results claimed as novel
- [ ] Novelty humility: acknowledges if result is incremental, partial, or computational
## 5. Proof / Evidence Validity
### For Proof-Based Results
- [ ] Full proof provided in machine-readable format (LaTeX / Markdown)
- [ ] Each step is logically justified
- [ ] No gaps longer than 2 sentences without explicit citation or lemma
### For Computational Results
- [ ] Code/artifact link provided (reproducible environment)
- [ ] Random seeds / parameters fully documented
- [ ] Output verified by independent script (if applicable)
### For Formal Verification
- [ ] Lean / Coq / other formal proof assistant file linked
- [ ] Compiles without errors on standard toolchain
## 6. Reproducibility Package
- [ ] All source code used is linked (repo commit hash / Gitea issue/PR reference)
- [ ] Dependencies listed with versions
- [ ] Minimal reproduction steps provided (3 steps or fewer)
## 7. Review Channel & Sign-off
- **Selected review channel**: (trusted mathematician / MathOverflow draft / Lean/mathlib review / arXiv-adjacent collaborator / other)
- **Reviewer identity**: (handle / name / affiliation)
- **Review date**: YYYY-MM-DD
- **Review outcome**: [APPROVED | REVISION REQUIRED | REJECTED]
- **Reviewer notes**: (free text)
## 8. Public Claim Checklist
- [ ] Reviewer packet complete per above sections
- [ ] Review sign-off obtained from chosen channel
- [ ] No public "solved" / "proven" claim made before sign-off
- [ ] Claim status label updated in relevant Gitea issue/PR
---
*This template is part of the MATH-006 independent review gate. No public novelty claim ships without a completed, signed-off packet.*

View File

@@ -1,20 +1,14 @@
from __future__ import annotations
import json
from pathlib import Path
from unittest.mock import patch
import yaml
from scripts.bezalel_gemma4_vps import (
DEFAULT_CONFIG_PATH,
DEFAULT_BEZALEL_VPS_HOST,
build_deploy_mutation,
build_runpod_endpoint,
build_vps_verify_command,
normalize_openai_base_url,
parse_deploy_response,
resolve_base_url,
update_config_text,
verify_openai_chat,
)
@@ -34,10 +28,6 @@ class _FakeResponse:
return False
def test_default_config_path_targets_bezalel_vps_root_config() -> None:
assert DEFAULT_CONFIG_PATH == Path("/root/wizards/bezalel/home/config.yaml")
def test_build_deploy_mutation_uses_ollama_image_and_openai_port() -> None:
query = build_deploy_mutation(name="bezalel-gemma4", gpu_type="NVIDIA L40S", model_tag="gemma4:latest")
@@ -47,30 +37,6 @@ def test_build_deploy_mutation_uses_ollama_image_and_openai_port() -> None:
assert 'volumeMountPath: "/root/.ollama"' in query
def test_normalize_openai_base_url_adds_v1_suffix() -> None:
assert normalize_openai_base_url("https://pod-11434.proxy.runpod.net") == "https://pod-11434.proxy.runpod.net/v1"
def test_normalize_openai_base_url_trims_chat_completions_suffix() -> None:
assert normalize_openai_base_url("https://pod-11434.proxy.runpod.net/v1/chat/completions") == "https://pod-11434.proxy.runpod.net/v1"
def test_resolve_base_url_prefers_vertex_over_base_and_pod_id() -> None:
base_url, source = resolve_base_url(
vertex_base_url="https://vertex.example.com/openai",
base_url="https://plain.example.com",
pod_id="abc123",
)
assert source == "vertex_base_url"
assert base_url == "https://vertex.example.com/openai/v1"
def test_resolve_base_url_falls_back_to_base_url_before_pod_id() -> None:
base_url, source = resolve_base_url(base_url="https://plain.example.com", pod_id="abc123")
assert source == "base_url"
assert base_url == "https://plain.example.com/v1"
def test_build_runpod_endpoint_appends_v1_suffix() -> None:
assert build_runpod_endpoint("abc123") == "https://abc123-11434.proxy.runpod.net/v1"
@@ -94,7 +60,7 @@ def test_parse_deploy_response_extracts_pod_id_and_endpoint() -> None:
}
def test_update_config_text_upserts_big_brain_provider_and_normalizes_base_url() -> None:
def test_update_config_text_upserts_big_brain_provider() -> None:
original = """
model:
default: kimi-k2.5
@@ -106,7 +72,7 @@ custom_providers:
model: gemma3:27b
"""
updated = update_config_text(original, base_url="https://new-pod-11434.proxy.runpod.net", model="gemma4:latest")
updated = update_config_text(original, base_url="https://new-pod-11434.proxy.runpod.net/v1", model="gemma4:latest")
parsed = yaml.safe_load(updated)
assert parsed["model"] == {"default": "kimi-k2.5", "provider": "kimi-coding"}
@@ -120,14 +86,7 @@ custom_providers:
]
def test_build_vps_verify_command_targets_bezalel_host_and_chat_completions() -> None:
command = build_vps_verify_command(base_url="https://pod-11434.proxy.runpod.net", model="gemma4:latest")
assert command.startswith(f"ssh root@{DEFAULT_BEZALEL_VPS_HOST} ")
assert "/v1/chat/completions" in command
assert "gemma4:latest" in command
def test_verify_openai_chat_calls_chat_completions_with_normalized_base_url() -> None:
def test_verify_openai_chat_calls_chat_completions() -> None:
response_payload = {
"choices": [
{
@@ -142,7 +101,7 @@ def test_verify_openai_chat_calls_chat_completions_with_normalized_base_url() ->
"scripts.bezalel_gemma4_vps.request.urlopen",
return_value=_FakeResponse(response_payload),
) as mocked:
result = verify_openai_chat("https://pod-11434.proxy.runpod.net", model="gemma4:latest", prompt="say READY")
result = verify_openai_chat("https://pod-11434.proxy.runpod.net/v1", model="gemma4:latest", prompt="say READY")
assert result == "READY"
req = mocked.call_args.args[0]
@@ -150,10 +109,3 @@ def test_verify_openai_chat_calls_chat_completions_with_normalized_base_url() ->
payload = json.loads(req.data.decode())
assert payload["model"] == "gemma4:latest"
assert payload["messages"][0]["content"] == "say READY"
def test_readme_documents_root_config_path_and_vps_proof_command() -> None:
readme = Path("scripts/README_bezalel_gemma4_vps.md").read_text()
assert "/root/wizards/bezalel/home/config.yaml" in readme
assert "ssh root@104.131.15.18" in readme
assert "--vertex-base-url" in readme