Upload intfloat/e5-large-v2 ctranslate2 weights

README.md

@@ -4,6 +4,9 @@ tags:
 - int8
 - float16
 - mteb
+- Sentence Transformers
+- sentence-similarity
+- sentence-transformers
 model-index:
 - name: e5-large-v2
   results:
@@ -2608,7 +2611,7 @@ Speedup inference while reducing memory by 2x-4x using int8 inference in C++ on
 
 quantized version of [intfloat/e5-large-v2](https://huggingface.co/intfloat/e5-large-v2)
 ```bash
-pip install hf-hub-ctranslate2>=2.12.0 ctranslate2>=3.16.0
+pip install hf-hub-ctranslate2>=2.12.0 ctranslate2>=3.17.1
 ```
 
 ```python
@@ -2648,16 +2651,20 @@ embeddings = model.encode(
 print(embeddings.shape, embeddings)
 scores = (embeddings @ embeddings.T) * 100
 
+# Hint: you can also host this code via REST API and
+# via github.com/michaelfeil/infinity  
+
+
 ```
 
-Checkpoint compatible to [ctranslate2>=3.16.0](https://github.com/OpenNMT/CTranslate2)
+Checkpoint compatible to [ctranslate2>=3.17.1](https://github.com/OpenNMT/CTranslate2)
 and [hf-hub-ctranslate2>=2.12.0](https://github.com/michaelfeil/hf-hub-ctranslate2)
 - `compute_type=int8_float16` for `device="cuda"`
 - `compute_type=int8`  for `device="cpu"`
 
-Converted on 2023-06-19 using
+Converted on 2023-10-13 using
 ```
-ct2-transformers-converter --model intfloat/e5-large-v2 --output_dir ~/tmp-ct2fast-e5-large-v2 --force --copy_files tokenizer.json modules.json README.md tokenizer_config.json sentence_bert_config.json vocab.txt special_tokens_map.json .gitattributes --trust_remote_code
+LLama-2 -> removed <pad> token.
 ```
 
 # Licence and other remarks:
@@ -2706,7 +2713,7 @@ batch_dict = tokenizer(input_texts, max_length=512, padding=True, truncation=Tru
 outputs = model(**batch_dict)
 embeddings = average_pool(outputs.last_hidden_state, batch_dict['attention_mask'])
 
-# (Optionally) normalize embeddings
+# normalize embeddings
 embeddings = F.normalize(embeddings, p=2, dim=1)
 scores = (embeddings[:2] @ embeddings[2:].T) * 100
 print(scores.tolist())
@@ -2721,6 +2728,52 @@ Please refer to our paper at [https://arxiv.org/pdf/2212.03533.pdf](https://arxi
 Check out [unilm/e5](https://github.com/microsoft/unilm/tree/master/e5) to reproduce evaluation results 
 on the [BEIR](https://arxiv.org/abs/2104.08663) and [MTEB benchmark](https://arxiv.org/abs/2210.07316).
 
+## Support for Sentence Transformers
+
+Below is an example for usage with sentence_transformers.
+```python
+from sentence_transformers import SentenceTransformer
+model = SentenceTransformer('intfloat/e5-large-v2')
+input_texts = [
+    'query: how much protein should a female eat',
+    'query: summit define',
+    "passage: As a general guideline, the CDC's average requirement of protein for women ages 19 to 70 is 46 grams per day. But, as you can see from this chart, you'll need to increase that if you're expecting or training for a marathon. Check out the chart below to see how much protein you should be eating each day.",
+    "passage: Definition of summit for English Language Learners. : 1  the highest point of a mountain : the top of a mountain. : 2  the highest level. : 3  a meeting or series of meetings between the leaders of two or more governments."
+]
+embeddings = model.encode(input_texts, normalize_embeddings=True)
+```
+
+Package requirements
+
+`pip install sentence_transformers~=2.2.2`
+
+Contributors: [michaelfeil](https://huggingface.co/michaelfeil)
+
+## FAQ
+
+**1. Do I need to add the prefix "query: " and "passage: " to input texts?**
+
+Yes, this is how the model is trained, otherwise you will see a performance degradation.
+
+Here are some rules of thumb:
+- Use "query: " and "passage: " correspondingly for asymmetric tasks such as passage retrieval in open QA, ad-hoc information retrieval.
+
+- Use "query: " prefix for symmetric tasks such as semantic similarity, paraphrase retrieval.
+
+- Use "query: " prefix if you want to use embeddings as features, such as linear probing classification, clustering.
+
+**2. Why are my reproduced results slightly different from reported in the model card?**
+
+Different versions of `transformers` and `pytorch` could cause negligible but non-zero performance differences.
+
+**3. Why does the cosine similarity scores distribute around 0.7 to 1.0?**
+
+This is a known and expected behavior as we use a low temperature 0.01 for InfoNCE contrastive loss. 
+
+For text embedding tasks like text retrieval or semantic similarity, 
+what matters is the relative order of the scores instead of the absolute values, 
+so this should not be an issue.
+
 ## Citation
 
 If you find our paper or models helpful, please consider cite as follows:
@@ -2736,4 +2789,4 @@ If you find our paper or models helpful, please consider cite as follows:
 
 ## Limitations
 
-This model only works for English texts. Long texts will be truncated to at most 512 tokens.
+This model only works for English texts. Long texts will be truncated to at most 512 tokens.

model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:6a2e972c674871a0be45c33e92a898e4b04256a882fdf6e6a72a2629facaea59
-size 1340583884
+oid sha256:074f41ce6dbf6564f709b9bfa09f88894c3b05c85f0eeb515bea9c2b72a1c67f
+size 670300108