Skip to content
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.

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

Add postprocessor::Sequence #1005

Closed
wants to merge 19 commits into from
Closed
Show file tree
Hide file tree
Changes from all commits
Commits
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
6 changes: 6 additions & 0 deletions bindings/node/lib/bindings/decoders.d.ts
Original file line number Diff line number Diff line change
Expand Up @@ -49,3 +49,9 @@ export function ctcDecoder(
word_delimiter_token?: string,
cleanup?: boolean
): Decoder;

/**
* Instantiate a new Sequence Decoder
* @param [decoders] The decoders to chain
*/
export function sequenceDecoder(decoders: Decoder[]): Decoder;
1 change: 1 addition & 0 deletions bindings/node/lib/bindings/decoders.js
Original file line number Diff line number Diff line change
Expand Up @@ -6,4 +6,5 @@ module.exports = {
metaspaceDecoder: native.decoders_Metaspace,
bpeDecoder: native.decoders_BPEDecoder,
ctcDecoder: native.decoders_CTC,
sequenceDecoder: native.decoders_Sequence,
};
30 changes: 29 additions & 1 deletion bindings/node/lib/bindings/decoders.test.ts
Original file line number Diff line number Diff line change
@@ -1,4 +1,10 @@
import { bpeDecoder, ctcDecoder, metaspaceDecoder, wordPieceDecoder } from "./decoders";
import {
bpeDecoder,
ctcDecoder,
metaspaceDecoder,
sequenceDecoder,
wordPieceDecoder,
} from "./decoders";

describe("wordPieceDecoder", () => {
it("accepts `undefined` as first parameter", () => {
Expand Down Expand Up @@ -42,3 +48,25 @@ describe("ctcDecoder", () => {
).toEqual("hello");
});
});

describe("sequenceDecoder", () => {
it("accepts `empty list` as parameter", () => {
expect(sequenceDecoder([])).toBeDefined();
});
it("encodes correctly", () => {
expect(
sequenceDecoder([ctcDecoder(), metaspaceDecoder()]).decode([
"▁",
"▁",
"H",
"H",
"i",
"i",
"▁",
"y",
"o",
"u",
])
).toEqual("Hi you");
});
});
102 changes: 23 additions & 79 deletions bindings/node/native/Cargo.lock

Some generated files are not rendered by default. Learn more about how customized files appear on GitHub.

33 changes: 31 additions & 2 deletions bindings/node/native/src/decoders.rs
Original file line number Diff line number Diff line change
Expand Up @@ -14,11 +14,11 @@ pub struct Decoder {
}

impl tk::Decoder for Decoder {
fn decode(&self, tokens: Vec<String>) -> tk::Result<String> {
fn decode_chain(&self, tokens: Vec<String>) -> tk::Result<Vec<String>> {
self.decoder
.as_ref()
.ok_or("Uninitialized Decoder")?
.decode(tokens)
.decode_chain(tokens)
}
}

Expand All @@ -42,6 +42,7 @@ declare_types! {
.map_err(|e| Error(format!("{}", e)))?;

Ok(cx.string(output).upcast())

}
}
}
Expand Down Expand Up @@ -115,12 +116,40 @@ fn ctc_decoder(mut cx: FunctionContext) -> JsResult<JsDecoder> {
Ok(decoder)
}

/// sequence()
fn sequence(mut cx: FunctionContext) -> JsResult<JsDecoder> {
let decoders = cx.argument::<JsArray>(0)?.to_vec(&mut cx)?;
let mut sequence = Vec::with_capacity(decoders.len());

decoders.into_iter().try_for_each(|decoder| {
match decoder.downcast::<JsDecoder>().or_throw(&mut cx) {
Ok(decoder) => {
let guard = cx.lock();
if let Some(decoder_arc) = &decoder.borrow(&guard).decoder {
let decoder: DecoderWrapper = (**decoder_arc).clone();
sequence.push(decoder);
}
Ok(())
}
Err(e) => Err(e),
}
})?;

let mut pretok = JsDecoder::new::<_, JsDecoder, _>(&mut cx, vec![])?;
let guard = cx.lock();
pretok.borrow_mut(&guard).decoder = Some(Arc::new(tk::DecoderWrapper::Sequence(
tk::decoders::sequence::Sequence::new(sequence),
)));
Ok(pretok)
}

/// Register everything here
pub fn register(m: &mut ModuleContext, prefix: &str) -> NeonResult<()> {
m.export_function(&format!("{}_ByteLevel", prefix), byte_level)?;
m.export_function(&format!("{}_WordPiece", prefix), wordpiece)?;
m.export_function(&format!("{}_Metaspace", prefix), metaspace)?;
m.export_function(&format!("{}_BPEDecoder", prefix), bpe_decoder)?;
m.export_function(&format!("{}_CTC", prefix), ctc_decoder)?;
m.export_function(&format!("{}_Sequence", prefix), sequence)?;
Ok(())
}
2 changes: 1 addition & 1 deletion bindings/node/native/src/pre_tokenizers.rs
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,7 @@ impl FromJsValue for JsSplitDelimiterBehavior {
}
}

impl<'s> From<JsSplitDelimiterBehavior> for SplitDelimiterBehavior {
impl From<JsSplitDelimiterBehavior> for SplitDelimiterBehavior {
fn from(v: JsSplitDelimiterBehavior) -> Self {
v.0
}
Expand Down
11 changes: 11 additions & 0 deletions bindings/node/native/src/processors.rs
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,17 @@ impl tk::PostProcessor for Processor {
.ok_or("Uninitialized PostProcessor")?
.process(encoding, pair_encoding, add_special_tokens)
}

fn process_chain(
&self,
encodings: Vec<Encoding>,
add_special_tokens: bool,
) -> tk::Result<Vec<Encoding>> {
self.processor
.as_ref()
.ok_or("Uninitialized PostProcessor")?
.process_chain(encodings, add_special_tokens)
}
}

declare_types! {
Expand Down
Loading