fse/paranmt-300-raw-float16

A static embedding model for fse: the mean of the token rows of a text, normalized to unit length.

Model

• rows (vocabulary): 77224
• dimensions of the table: 300 (float16)
• table stored by quantize_rows@1 from the float32 table xxh3-128:d0457b3eb8b4213d26d2056fa52cf0cf
• tokenizer: fse_words@1
• steps per row: normalize@1
• fingerprint: xxh3-128:26b7176aacbdf42b64c8e90791847319

Usage

This model requires fse 2.0 or later.

pip install "fse>=2.0"

import fse

encoder = fse.Encoder.from_pretrained("fse/paranmt-300-raw-float16")
vectors = encoder.encode(["A sentence."])

The encoder normalizes every vector to unit length (the stored step normalize@1). To get the raw weighted mean instead, for example to fold a linear head into the table, load it with normalize=False:

encoder = fse.Encoder.from_pretrained("fse/paranmt-300-raw-float16", normalize=False)

The suffix -raw means without SIF weighting, not without normalization. Both -raw and -sif10 models load normalized by default.

Source and license

Converted from the gensim model fse/paranmt-300 at revision a08e15771891207a22f840a3091a00a1657d888b.

The vectors fse computes are those of the quantized table: on the cpu backend bit for bit as its reference computes them, not those of the model before quantization. How close they come to it is measured, not promised by this file.

License: unknown. The license of the source model applies to its use.

About fse

Made with fse, fast sentence embeddings.

Choosing a method and storage format

On the 1,379 STS-B test pairs, raw float32 scored 79.81644 Pearson x100 and SIF-10 float32 scored 78.11215. Use raw for a plain mean of the source vectors. Use SIF-10 to apply the fixed frequency weighting and training-corpus component removal measured here. Evaluate the choice on your own data.

Use float32 to retain the prepared table without storage rounding. Use float16 or int8 when the measured size reduction and neighbour changes below meet your needs. fse computes all three in float32.

This model	Measured value
Model file bytes	47523995
STS-B Pearson x100	79.81645
Difference from same-method float32, points	+0.00001
Recall@10 against same-method float32	0.999891
Default tokenizer Pearson x100	79.81645

The controlled STS-B comparison uses the 1.x notebook's NLTK preprocessing through an external-tokenizer view of this same table. The default-tokenizer row uses the saved fse_words@1 tokenizer on raw text. Recall@10 compares each of the 2,758 STS-B sentence rows against all other rows, excludes self, and resolves equal similarities by row index. No load-time or speed claim is made.

Source terms and attribution

The authors did not state a license for these vectors in the release materials we inspected. license: unknown does not grant permission. This conversion is kept for preservation. See ParaNMT and the paper. Its training data, CzEng 1.6, is freely available only for non-commercial research. This does not establish a license for the derived vectors. Rights holders can request removal through the repository issues.