AI/firecrawl
1
0
Fork 0
mirror of https://git.mirrors.martin98.com/https://github.com/mendableai/firecrawl synced 2025-08-03 06:30:37 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix(llmExtract): cost calculation

Browse Source
This commit is contained in:
Gergő Móricz 2025-04-16 01:04:48 -07:00
parent 0d813b628b
commit 129b10e478
1 changed files with 11 additions and 28 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
apps/api/src/scraper/scrapeURL/transformers/llmExtract.ts
View File

@ -259,23 +259,6 @@ export async function generateCompletions({
throw new Error("document.markdown is undefined -- this is unexpected"); throw new Error("document.markdown is undefined -- this is unexpected");
} }
const { maxInputTokens, maxOutputTokens } = getModelLimits(
currentModel.modelId,
);
// Calculate 80% of max input tokens (for content)
const maxTokensSafe = Math.floor(maxInputTokens * 0.8);
// Use the new trimming function
const {
text: trimmedMarkdown,
numTokens,
warning: trimWarning,
} = trimToTokenLimit(markdown, maxTokensSafe, model.modelId, previousWarning);
// WE USE BIG MODELS NOW
// markdown = trimmedMarkdown;
// warning = trimWarning;
try { try {
const prompt = const prompt =
options.prompt !== undefined options.prompt !== undefined
@ -300,16 +283,16 @@ export async function generateCompletions({
return { return {
extract, extract,
warning, warning,
numTokens, numTokens: result.usage?.promptTokens ?? 0,
totalUsage: { totalUsage: {
promptTokens: numTokens, promptTokens: result.usage?.promptTokens ?? 0,
completionTokens: result.usage?.completionTokens ?? 0, completionTokens: result.usage?.completionTokens ?? 0,
totalTokens: numTokens + (result.usage?.completionTokens ?? 0), totalTokens: result.usage?.promptTokens ?? 0 + (result.usage?.completionTokens ?? 0),
}, },
model: currentModel.modelId, model: currentModel.modelId,
cost: calculateCost( cost: calculateCost(
currentModel.modelId, currentModel.modelId,
numTokens, result.usage?.promptTokens ?? 0,
result.usage?.completionTokens ?? 0, result.usage?.completionTokens ?? 0,
), ),
}; };
@ -341,16 +324,16 @@ export async function generateCompletions({
return { return {
extract, extract,
warning, warning,
numTokens, numTokens: result.usage?.promptTokens ?? 0,
totalUsage: { totalUsage: {
promptTokens: numTokens, promptTokens: result.usage?.promptTokens ?? 0,
completionTokens: result.usage?.completionTokens ?? 0, completionTokens: result.usage?.completionTokens ?? 0,
totalTokens: numTokens + (result.usage?.completionTokens ?? 0), totalTokens: result.usage?.promptTokens ?? 0 + (result.usage?.completionTokens ?? 0),
}, },
model: currentModel.modelId, model: currentModel.modelId,
cost: calculateCost( cost: calculateCost(
currentModel.modelId, currentModel.modelId,
numTokens, result.usage?.promptTokens ?? 0,
result.usage?.completionTokens ?? 0, result.usage?.completionTokens ?? 0,
), ),
}; };
@ -541,13 +524,13 @@ export async function generateCompletions({
} }
// Since generateObject doesn't provide token usage, we'll estimate it // Since generateObject doesn't provide token usage, we'll estimate it
const promptTokens = numTokens; const promptTokens = result.usage?.promptTokens ?? 0;
const completionTokens = result?.usage?.completionTokens ?? 0; const completionTokens = result.usage?.completionTokens ?? 0;
return { return {
extract, extract,
warning, warning,
numTokens, numTokens: promptTokens,
totalUsage: { totalUsage: {
promptTokens, promptTokens,
completionTokens, completionTokens,