这是indexloc提供的服务，不要输入任何密码
Skip to content

Handle special token in TikToken #528

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
merged 2 commits into from
Jan 4, 2024
Merged

Handle special token in TikToken #528

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ac29747
Handle special token in TikToken
timothycarambat Jan 4, 2024
9264d52
remove duplicate method
timothycarambat Jan 4, 2024
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
2 changes: 1 addition & 1 deletion server/utils/helpers/chat/index.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -300,7 +300,7 @@ function cannonball({
// if the delta is the token difference between where our prompt is in size
// and where we ideally need to land.
const delta = initialInputSize - targetTokenSize;
const tokenChunks = tokenManager.tokensFromString(input);
const tokenChunks = tokenManager.countFromString(input);
const middleIdx = Math.floor(tokenChunks.length / 2);

// middle truncate the text going left and right of midpoint
Expand Down
14 changes: 5 additions & 9 deletions server/utils/helpers/tiktoken.js
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,31 +3,27 @@ const { getEncodingNameForModel, getEncoding } = require("js-tiktoken");
class TokenManager {
constructor(model = "gpt-3.5-turbo") {
this.model = model;
this.encoderName = this.getEncodingFromModel(model);
this.encoderName = this.#getEncodingFromModel(model);
this.encoder = getEncoding(this.encoderName);
this.buffer = 50;
}

getEncodingFromModel(model) {
#getEncodingFromModel(model) {
try {
return getEncodingNameForModel(model);
} catch {
return "cl100k_base";
}
}

tokensFromString(input = "") {
const tokens = this.encoder.encode(input);
return tokens;
}

bytesFromTokens(tokens = []) {
const bytes = this.encoder.decode(tokens);
return bytes;
}

// Pass in an empty array of disallowedSpecials to handle all tokens as text and to be tokenized.
// https://github.com/openai/tiktoken/blob/9e79899bc248d5313c7dd73562b5e211d728723d/tiktoken/core.py#L91C20-L91C38
countFromString(input = "") {
const tokens = this.encoder.encode(input);
const tokens = this.encoder.encode(input, undefined, []);
return tokens.length;
}

Expand Down