feat(ai): switch to qwen2.5-coder-8k model
- Use larger coder model (32.8B params) for better reasoning - Update client default model - Update all AI routes to use qwen2.5-coder-8k:latest - Already installed and available on server Ollama Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
This commit is contained in:
@@ -107,7 +107,7 @@ export async function POST() {
|
||||
let res: Awaited<ReturnType<typeof client.chat.completions.create>>
|
||||
try {
|
||||
res = await client.chat.completions.create({
|
||||
model: 'qwen3:14b',
|
||||
model: 'qwen2.5-coder-8k:latest',
|
||||
max_tokens: 2048,
|
||||
messages: [
|
||||
{
|
||||
@@ -166,7 +166,7 @@ Flag only zones with rising or elevated risk (at most 5 flags; do not flag healt
|
||||
|
||||
await withUser(session.sub, async tx =>
|
||||
writeAuditLog(tx, 'incidents', session.sub, 'ai_risk_flags',
|
||||
{ flags, summary: input.summary, model: 'qwen3:14b' })
|
||||
{ flags, summary: input.summary, model: 'qwen2.5-coder-8k:latest' })
|
||||
)
|
||||
|
||||
return NextResponse.json({ flags, summary: input.summary })
|
||||
|
||||
@@ -57,7 +57,7 @@ export async function POST(
|
||||
let res: Awaited<ReturnType<typeof client.chat.completions.create>>
|
||||
try {
|
||||
res = await client.chat.completions.create({
|
||||
model: 'qwen3:14b',
|
||||
model: 'qwen2.5-coder-8k:latest',
|
||||
max_tokens: 2048,
|
||||
messages: [
|
||||
{
|
||||
@@ -105,7 +105,7 @@ Provide 3–5 Why steps drilling from immediate cause to root cause. Frame CAPA
|
||||
await withUser(session.sub, async tx => {
|
||||
await writeAuditLog(tx, 'incidents', id, 'ai_rca_draft', {
|
||||
root_cause_summary: draft.root_cause_summary as string,
|
||||
model: 'qwen3:14b',
|
||||
model: 'qwen2.5-coder-8k:latest',
|
||||
})
|
||||
})
|
||||
|
||||
|
||||
@@ -48,7 +48,7 @@ export async function POST(
|
||||
let res: Awaited<ReturnType<typeof client.chat.completions.create>>
|
||||
try {
|
||||
res = await client.chat.completions.create({
|
||||
model: 'qwen3:14b',
|
||||
model: 'qwen2.5-coder-8k:latest',
|
||||
max_tokens: 1024,
|
||||
messages: [
|
||||
{
|
||||
@@ -103,7 +103,7 @@ Output exactly this JSON structure:
|
||||
}
|
||||
|
||||
await withUser(session.sub, async tx => {
|
||||
await writeAuditLog(tx, 'incidents', id, 'ai_triage_suggest', { suggestion: input, model: 'qwen3:14b' })
|
||||
await writeAuditLog(tx, 'incidents', id, 'ai_triage_suggest', { suggestion: input, model: 'qwen2.5-coder-8k:latest' })
|
||||
})
|
||||
|
||||
return NextResponse.json(input)
|
||||
|
||||
@@ -40,7 +40,7 @@ export async function POST(request: NextRequest) {
|
||||
let res: Awaited<ReturnType<typeof client.chat.completions.create>>
|
||||
try {
|
||||
res = await client.chat.completions.create({
|
||||
model: 'qwen3:14b',
|
||||
model: 'qwen2.5-coder-8k:latest',
|
||||
max_tokens: 1024,
|
||||
messages: [
|
||||
{
|
||||
@@ -87,7 +87,7 @@ Scoring criteria: specificity (location, time, persons involved), completeness (
|
||||
|
||||
await withUser(session.sub, async tx => {
|
||||
await writeAuditLog(tx, 'incidents', session.sub, 'ai_quality_check', {
|
||||
score: input.score, passes: input.passes, model: 'qwen3:14b',
|
||||
score: input.score, passes: input.passes, model: 'qwen2.5-coder-8k:latest',
|
||||
})
|
||||
})
|
||||
|
||||
|
||||
@@ -5,5 +5,5 @@ export function createDeepSeekClient(apiKey: string): OpenAI {
|
||||
}
|
||||
|
||||
export function createLocalLLMClient(): OpenAI {
|
||||
return new OpenAI({ apiKey: 'local', baseURL: 'http://localhost:11434/v1' })
|
||||
return new OpenAI({ apiKey: 'local', baseURL: 'http://localhost:11434/v1', defaultQuery: { model: 'qwen2.5-coder-8k:latest' } })
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user