Skip to content

Remove final raw Groq inference path #65

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
stackbilt-admin merged 1 commit into from
Jun 2, 2026
Merged

Remove final raw Groq inference path #65

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
63 changes: 9 additions & 54 deletions web/src/groq.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,10 +44,10 @@ export async function askGroq(
}
}

// ─── Logprobs-enabled classification ─────────────────────────
// Requests logprobs from Groq's OpenAI-compatible API and computes
// geometric mean of token probabilities for the classification value.
// Returns both self-reported confidence and token-level confidence.
// ─── Classification confidence helper ────────────────────────
// Provider-backed replacement for the former raw Groq logprobs call.
// llm-providers does not expose provider logprob request options yet, so
// tokenConfidence mirrors self-reported confidence until that contract lands.

export interface LogprobClassification {
pattern: string;
Expand All @@ -64,65 +64,20 @@ export async function askGroqWithLogprobs(
userPrompt: string,
baseUrl = 'https://api.groq.com',
): Promise<LogprobClassification> {
const response = await fetch(`${baseUrl}/openai/v1/chat/completions`, {
method: 'POST',
headers: {
'Content-Type': 'application/json',
'Authorization': `Bearer ${apiKey}`,
},
body: JSON.stringify({
model,
temperature: 0.1,
max_tokens: 200,
response_format: { type: 'json_object' },
logprobs: true,
top_logprobs: 3,
messages: [
{ role: 'system', content: systemPrompt },
{ role: 'user', content: userPrompt },
],
}),
});

if (!response.ok) {
const errText = await response.text();
throw new Error(`Groq logprobs API error ${response.status}: ${errText}`);
}

interface LogprobToken {
token: string;
logprob: number;
}

const data = await response.json<{
choices: {
message: { content: string };
logprobs?: { content?: LogprobToken[] };
}[];
}>();

const raw = data.choices[0]?.message?.content ?? '{}';
const parsed = JSON.parse(raw) as {
const { parsed } = await askGroqJson<{
pattern?: string;
complexity?: number;
needs_tools?: boolean;
confidence?: number;
};

// Compute geometric mean of token logprobs
let tokenConfidence = 0.5; // fallback if no logprobs
const logprobTokens = data.choices[0]?.logprobs?.content;
if (logprobTokens && logprobTokens.length > 0) {
const sumLogprobs = logprobTokens.reduce((sum, t) => sum + t.logprob, 0);
tokenConfidence = Math.exp(sumLogprobs / logprobTokens.length);
}
}>(apiKey, model, systemPrompt, userPrompt, baseUrl, { maxTokens: 200, temperature: 0.1 });
const confidence = parsed.confidence ?? 0.5;

return {
pattern: (parsed.pattern ?? 'general_knowledge').toLowerCase().replace(/[^a-z_]/g, ''),
complexity: parsed.complexity ?? 2,
needs_tools: parsed.needs_tools ?? false,
selfReportedConfidence: parsed.confidence ?? 0.5,
tokenConfidence,
selfReportedConfidence: confidence,
tokenConfidence: confidence,
};
}

Expand Down
35 changes: 7 additions & 28 deletions web/tests/groq.test.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
// Groq helper tests — askGroq, askGroqJson, askGroqWithLogprobs, probeConsistency
// Mocks provider factory and fetch() to test API interaction without real calls
// Mocks provider factory to test API interaction without real calls

import { describe, it, expect, vi, beforeEach } from 'vitest';

Expand All @@ -22,10 +22,6 @@ vi.mock('../src/kernel/memory/index.js', () => ({
},
}));

// Mock global fetch
const mockFetch = vi.fn();
vi.stubGlobal('fetch', mockFetch);

const { askGroq, askGroqJson, askGroqWithLogprobs, probeConsistency } = await import('../src/groq.js');

function providerResponse(content: unknown, usage = { inputTokens: 100, outputTokens: 50, totalTokens: 150, cost: 0.001 }) {
Expand All @@ -38,15 +34,6 @@ function providerResponse(content: unknown, usage = { inputTokens: 100, outputTo
};
}

function groqLogprobResponse(content: string, logprobs: Array<{ token: string; logprob: number }>) {
return new Response(JSON.stringify({
choices: [{
message: { content },
logprobs: { content: logprobs },
}],
}), { status: 200, headers: { 'Content-Type': 'application/json' } });
}

describe('askGroq', () => {
beforeEach(() => {
vi.clearAllMocks();
Expand Down Expand Up @@ -144,39 +131,31 @@ describe('askGroqWithLogprobs', () => {
providerMocks.createLLMProviderFactory.mockReturnValue({ generateResponse: providerMocks.generateResponse });
});

it('parses classification with token confidence', async () => {
mockFetch.mockResolvedValue(groqLogprobResponse(
it('parses classification with provider-backed confidence', async () => {
providerMocks.generateResponse.mockResolvedValue(providerResponse(
'{"pattern":"greeting","complexity":0,"needs_tools":false,"confidence":0.99}',
[
{ token: '{"', logprob: -0.01 },
{ token: 'pattern', logprob: -0.02 },
{ token: '":"', logprob: -0.01 },
{ token: 'greeting', logprob: -0.05 },
],
));
const result = await askGroqWithLogprobs('key', 'model', 'sys', 'user');
expect(result.pattern).toBe('greeting');
expect(result.complexity).toBe(0);
expect(result.needs_tools).toBe(false);
expect(result.selfReportedConfidence).toBe(0.99);
expect(result.tokenConfidence).toBeGreaterThan(0);
expect(result.tokenConfidence).toBeLessThanOrEqual(1);
expect(result.tokenConfidence).toBe(0.99);
});

it('defaults missing fields', async () => {
mockFetch.mockResolvedValue(groqLogprobResponse('{}', []));
providerMocks.generateResponse.mockResolvedValue(providerResponse('{}'));
const result = await askGroqWithLogprobs('key', 'model', 'sys', 'user');
expect(result.pattern).toBe('general_knowledge');
expect(result.complexity).toBe(2);
expect(result.needs_tools).toBe(false);
expect(result.selfReportedConfidence).toBe(0.5);
expect(result.tokenConfidence).toBe(0.5); // fallback when no logprobs
expect(result.tokenConfidence).toBe(0.5);
});

it('sanitizes pattern string', async () => {
mockFetch.mockResolvedValue(groqLogprobResponse(
providerMocks.generateResponse.mockResolvedValue(providerResponse(
'{"pattern":"Greeting!123"}',
[{ token: 'x', logprob: -0.1 }],
));
const result = await askGroqWithLogprobs('key', 'model', 'sys', 'user');
expect(result.pattern).toBe('greeting');
Expand Down