Update README.md

README.md

@@ -48,12 +48,14 @@ as well as the English [MNLI dataset](https://huggingface.co/datasets/multi_nli)
 The main advantage of distilled models is that they are smaller (faster inference, lower memory requirements) than their teachers (XLM-RoBERTa-large).
 The disadvantage is that they lose some of the performance of their larger teachers. 
 
+For highest inference speed, I recommend using this 6-layer model. For higher performance I recommend 
+[mDeBERTa-v3-base-mnli-xnli](https://huggingface.co/MoritzLaurer/mDeBERTa-v3-base-mnli-xnli) (as of 14.02.2023).
 
 ### How to use the model
 #### Simple zero-shot classification pipeline
 ```python
 from transformers import pipeline
-classifier = pipeline("zero-shot-classification", model="MoritzLaurer/xlm-v-base-mnli-xnli")
+classifier = pipeline("zero-shot-classification", model="MoritzLaurer/multilingual-MiniLMv2-L6-mnli-xnli")
 
 sequence_to_classify = "Angela Merkel ist eine Politikerin in Deutschland und Vorsitzende der CDU"
 candidate_labels = ["politics", "economy", "entertainment", "environment"]
@@ -66,7 +68,7 @@ from transformers import AutoTokenizer, AutoModelForSequenceClassification
 import torch
 device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
 
-model_name = "MoritzLaurer/xlm-v-base-mnli-xnli"
+model_name = "MoritzLaurer/multilingual-MiniLMv2-L6-mnli-xnli"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 model = AutoModelForSequenceClassification.from_pretrained(model_name)
 
@@ -93,7 +95,8 @@ avoids catastrophic forgetting of the other languages it was pre-trained on;
 and significantly reduces training costs. 
 
 ### Training procedure
-The model was trained using the Hugging Face trainer with the following hyperparameters.
+The model was trained using the Hugging Face trainer with the following hyperparameters. 
+The exact underlying model is [mMiniLMv2-L6-H384-distilled-from-XLMR-Large](https://huggingface.co/nreimers/mMiniLMv2-L6-H384-distilled-from-XLMR-Large).
 ```
 training_args = TrainingArguments(
     num_train_epochs=3,              # total number of training epochs
@@ -119,13 +122,13 @@ XLM-RoBERTa-large instead of -base (multilingual-MiniLM-L6-v2).
 |Datasets|avg_xnli|ar|bg|de|el|en|es|fr|hi|ru|sw|th|tr|ur|vi|zh|
 | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
 |Accuracy|0.713|0.687|0.742|0.719|0.723|0.789|0.748|0.741|0.691|0.714|0.642|0.699|0.696|0.664|0.723|0.721|
-|Speed (text/sec)|6093.0|6210.0|6003.0|6053.0|5409.0|6531.0|6205.0|5615.0|5734.0|5970.0|6219.0|6289.0|6533.0|5851.0|5970.0|6798.0|
+|Speed text/sec (A100 GPU, eval_batch=120)|6093.0|6210.0|6003.0|6053.0|5409.0|6531.0|6205.0|5615.0|5734.0|5970.0|6219.0|6289.0|6533.0|5851.0|5970.0|6798.0|
 
 
 |Datasets|mnli_m|mnli_mm|
 | :---: | :---: | :---: |
 |Accuracy|0.782|0.8|
-|Speed (text/sec)|4430.0|4395.0|
+|Speed text/sec (A100 GPU, eval_batch=120)|4430.0|4395.0|