Hugging Face's logo

embedl
/
paraphrase-multilingual-MiniLM-L12-v2-quantized-trt

Sentence Similarity
TensorRT
ONNX
quantization
edge
embedl
Model card Files
xet
 Community
dann-od commited on
Commit
dd4c9da
·
verified ·
1 Parent(s): a08d0c6

Shorter Embedl accuracy column title

Browse files
Files changed (1) hide show
  1. README.md +1 -1
README.md CHANGED
@@ -125,7 +125,7 @@ Latency measured with TensorRT + `trtexec`, GPU compute time only
125
  Evaluated on the sts17 validation split. The quantized model
126
  retains nearly all of the FP32 accuracy with a small tolerance.
127
 
128
- | Metric | FP32 (ours) | **Embedl Paraphrase Multilingual Minilm L12 V2 INT8** | Δ |
129
  |---|---|---|---|
130
  | Spearman ρ | 0.8130 | **0.8008** | -0.0122 |
131
  | ρ (ar-ar) | 0.7915 | **0.7906** | -0.0010 |
 
125
  Evaluated on the sts17 validation split. The quantized model
126
  retains nearly all of the FP32 accuracy with a small tolerance.
127
 
128
+ | Metric | FP32 (ours) | **Embedl INT8** | Δ |
129
  |---|---|---|---|
130
  | Spearman ρ | 0.8130 | **0.8008** | -0.0122 |
131
  | ρ (ar-ar) | 0.7915 | **0.7906** | -0.0010 |