ducanhdinh / jepa_proof_vicreg_replace

huggingface.co
Total runs: 0
24-hour runs: 0
7-day runs: 0
30-day runs: 0
Model's Last Updated: June 15 2026

Introduction of jepa_proof_vicreg_replace

Model Details of jepa_proof_vicreg_replace

ducanhdinh/jepa_proof_vicreg_replace

BERT encoder pretrained from scratch với VICReg + Lexical Substitution augmentation .

Augmentation strategy

Thay vì span masking, model này dùng lexical substitution để tạo view 2:

Mô tả
View 1 Câu gốc (không thay đổi)
View 2 15–20% token ngẫu nhiên được thay bằng token ngữ nghĩa gần, dự đoán bởi BERT MLM (top-5, loại trừ token gốc)

Quá trình tạo view 2 được thực hiện offline 1 lần trước khi train, không gọi BERT thêm lần nào trong vòng lặp training.

Kiến trúc VICReg
Text → BERT (mean-pool) → z ∈ R^768 → Expander MLP → z' ∈ R^3072
                                                       ↑ VICReg loss áp dụng tại đây
Loss term Hệ số Mô tả
Invariance 25.0 MSE giữa z1 và z2 (căn chỉnh hai views)
Variance 25.0 Giữ std của mỗi chiều ≥ 1 (chống collapse)
Covariance 1.0 Decorrelate các chiều embedding

Expander gồm 3 lớp Linear-BatchNorm-ReLU (dim = 3072 ).

Thông số huấn luyện
Tham số Giá trị
Max sequence length 256
Batch size 256
Epochs 10
Learning rate 0.0001
Expander dim 3072
Mask ratio (lexsubst) 0.15–0.2
Top-k candidates 5
sim_coeff 25.0
std_coeff 25.0
cov_coeff 1.0
Cách dùng — BERT encoder (feature extraction)
from transformers import BertModel, BertTokenizerFast
import torch

tokenizer = BertTokenizerFast.from_pretrained("ducanhdinh/jepa_proof_vicreg_replace")
bert      = BertModel.from_pretrained("ducanhdinh/jepa_proof_vicreg_replace/encoder")

encoded = tokenizer(
    ["Hello world!", "VICReg with lexical substitution."],
    return_tensors="pt",
    padding=True,
    truncation=True,
)
with torch.no_grad():
    out    = bert(**encoded)
    hidden = out.last_hidden_state          # (B, T, 768)
    mask   = encoded["attention_mask"].unsqueeze(-1).float()
    emb    = (hidden * mask).sum(1) / mask.sum(1).clamp(min=1)  # mean-pool → (B, 768)
Cách dùng — Full model (encoder + expander)
import torch
from text_vicreg_replace import TextVICRegReplace, VICRegReplacePretrainConfig

cfg   = VICRegReplacePretrainConfig()
model = TextVICRegReplace(cfg)
state = torch.load(
    hf_hub_download("ducanhdinh/jepa_proof_vicreg_replace", "pytorch_model.bin"),
    map_location="cpu",
)
model.load_state_dict(state)
model.eval()

Runs of ducanhdinh jepa_proof_vicreg_replace on huggingface.co

0
Total runs
0
24-hour runs
0
3-day runs
0
7-day runs
0
30-day runs

More Information About jepa_proof_vicreg_replace huggingface.co Model

More jepa_proof_vicreg_replace license Visit here:

https://choosealicense.com/licenses/apache-2.0

jepa_proof_vicreg_replace huggingface.co

jepa_proof_vicreg_replace huggingface.co is an AI model on huggingface.co that provides jepa_proof_vicreg_replace's model effect (), which can be used instantly with this ducanhdinh jepa_proof_vicreg_replace model. huggingface.co supports a free trial of the jepa_proof_vicreg_replace model, and also provides paid use of the jepa_proof_vicreg_replace. Support call jepa_proof_vicreg_replace model through api, including Node.js, Python, http.

jepa_proof_vicreg_replace huggingface.co Url

https://huggingface.co/ducanhdinh/jepa_proof_vicreg_replace

ducanhdinh jepa_proof_vicreg_replace online free

jepa_proof_vicreg_replace huggingface.co is an online trial and call api platform, which integrates jepa_proof_vicreg_replace's modeling effects, including api services, and provides a free online trial of jepa_proof_vicreg_replace, you can try jepa_proof_vicreg_replace online for free by clicking the link below.

ducanhdinh jepa_proof_vicreg_replace online free url in huggingface.co:

https://huggingface.co/ducanhdinh/jepa_proof_vicreg_replace

jepa_proof_vicreg_replace install

jepa_proof_vicreg_replace is an open source model from GitHub that offers a free installation service, and any user can find jepa_proof_vicreg_replace on GitHub to install. At the same time, huggingface.co provides the effect of jepa_proof_vicreg_replace install, users can directly use jepa_proof_vicreg_replace installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

jepa_proof_vicreg_replace install url in huggingface.co:

https://huggingface.co/ducanhdinh/jepa_proof_vicreg_replace

Url of jepa_proof_vicreg_replace

jepa_proof_vicreg_replace huggingface.co Url

Provider of jepa_proof_vicreg_replace huggingface.co

ducanhdinh
ORGANIZATIONS

Other API from ducanhdinh

huggingface.co

Total runs: 718
Run Growth: 534
Growth Rate: 74.37%
Updated:November 24 2025