mlboydaisuke / bge-large-en-v1.5-ExecuTorch

huggingface.co
Total runs: 52
24-hour runs: 0
7-day runs: -51
30-day runs: -27
Model's Last Updated: September 04 2026
feature-extraction

Introduction of bge-large-en-v1.5-ExecuTorch

Model Details of bge-large-en-v1.5-ExecuTorch

bge-large-en-v1.5 — ExecuTorch

BGE's English model at 24 layers. Text in, one 1024-dimensional vector out, for search and retrieval that never leaves the device.

  • Source : BAAI/bge-large-en-v1.5 — 24 layers, 1024 dimensions, 30,522 vocabulary
  • License : mit
  • Input : input_ids and attention_mask , both [1, 256] int64
  • Output : [1, 1024] , CLS-pooled and L2-normalised inside the graph
The recipe is in the graph, and it was read off this repo

sentence-transformers stores it per model, and the shelf's seven embedding models do not agree. This one pools CLS and normalises , read from 1_Pooling/config.json and modules.json rather than inferred from the family name. Getting it wrong does not throw; it returns vectors that look fine and rank wrong.

Verification
build file size (MB) Mac ms* worst cosine vs eager retrieval budget
fp32 embed_bge_large_en_xnnpack_fp32.pte 1336.7 104.5 1.000000 0%
fp16 embed_bge_large_en_xnnpack_fp16.pte 668.8 223.3 1.000000 2%
Core ML (fp16, iOS) embed_bge_large_en_coreml_all.pte 669.2 26.2 0.999997 9%

*Mac arm64, median of 10, one 256-token sequence — a reference point for relative cost, not a device number. Torch eager fp32 on the same machine is 100.7 ms.

Cosine is measured against the model run in eager through its own pooling, over eight sentences. The last column is the one that decides: rank those eight against each other, and ask whether this build's score error is smaller than the gap between the document a query retrieves and the runner-up. Every shipped build keeps all eight top-1 results.

Not shipped: int8

embed_bge_large_en_xnnpack_int8.pte is 431.7 MB — smaller than fp16's 668.8 MB, because the token embedding table is only 125 MB of the 1336.7 MB model (9%), leaving most of the weight in linears for int8 to shrink.

It is withheld on the number that decides. Ranking the eight test sentences against each other, this build moves a pair score by at most 0.0123 while the closest fp32 decision — the gap between the document a query retrieves and the runner-up — is 0.0090 . That is 137% of the room available, against a bar of 50%.

Correlation reads 0.999490 for this build, which no correlation gate would stop.

torch.export -> to_edge_transform_and_lower(partitioner) -> .pte (conversion scripts: executorch-models )

Runs of mlboydaisuke bge-large-en-v1.5-ExecuTorch on huggingface.co

52
Total runs
0
24-hour runs
-4
3-day runs
-51
7-day runs
-27
30-day runs

More Information About bge-large-en-v1.5-ExecuTorch huggingface.co Model

More bge-large-en-v1.5-ExecuTorch license Visit here:

https://choosealicense.com/licenses/mit

bge-large-en-v1.5-ExecuTorch huggingface.co

bge-large-en-v1.5-ExecuTorch huggingface.co is an AI model on huggingface.co that provides bge-large-en-v1.5-ExecuTorch's model effect (), which can be used instantly with this mlboydaisuke bge-large-en-v1.5-ExecuTorch model. huggingface.co supports a free trial of the bge-large-en-v1.5-ExecuTorch model, and also provides paid use of the bge-large-en-v1.5-ExecuTorch. Support call bge-large-en-v1.5-ExecuTorch model through api, including Node.js, Python, http.

bge-large-en-v1.5-ExecuTorch huggingface.co Url

https://huggingface.co/mlboydaisuke/bge-large-en-v1.5-ExecuTorch

mlboydaisuke bge-large-en-v1.5-ExecuTorch online free

bge-large-en-v1.5-ExecuTorch huggingface.co is an online trial and call api platform, which integrates bge-large-en-v1.5-ExecuTorch's modeling effects, including api services, and provides a free online trial of bge-large-en-v1.5-ExecuTorch, you can try bge-large-en-v1.5-ExecuTorch online for free by clicking the link below.

mlboydaisuke bge-large-en-v1.5-ExecuTorch online free url in huggingface.co:

https://huggingface.co/mlboydaisuke/bge-large-en-v1.5-ExecuTorch

bge-large-en-v1.5-ExecuTorch install

bge-large-en-v1.5-ExecuTorch is an open source model from GitHub that offers a free installation service, and any user can find bge-large-en-v1.5-ExecuTorch on GitHub to install. At the same time, huggingface.co provides the effect of bge-large-en-v1.5-ExecuTorch install, users can directly use bge-large-en-v1.5-ExecuTorch installed effect in huggingface.co for debugging and trial. It also supports api for free installation.

bge-large-en-v1.5-ExecuTorch install url in huggingface.co:

https://huggingface.co/mlboydaisuke/bge-large-en-v1.5-ExecuTorch

Url of bge-large-en-v1.5-ExecuTorch

bge-large-en-v1.5-ExecuTorch huggingface.co Url

Provider of bge-large-en-v1.5-ExecuTorch huggingface.co

mlboydaisuke
ORGANIZATIONS

Other API from mlboydaisuke