Add new SentenceTransformer model

README.md

@@ -9,32 +9,36 @@ tags:
 - loss:MultipleNegativesRankingLoss
 base_model: prajjwal1/bert-small
 widget:
-- source_sentence: How do I polish my English skills?
+- source_sentence: How would it effect our economy if we ban Chinese products in our
+    country.?
   sentences:
-  - How can we polish English skills?
-  - Why should I move to Israel as a Jew?
-  - What are vitamins responsible for?
-- source_sentence: Can I use the Kozuka Gothic Pro font as a font-face on my web site?
+  - How would it effect our economy if we ban Chinese products in our country.?
+  - Which cities in India is suitable for part time teaching job where one can prepare
+    for civil services exam?
+  - What are the best one-liners?
+- source_sentence: Why do we need Java programming?
   sentences:
-  - Can I use the Kozuka Gothic Pro font as a font-face on my web site?
-  - Why are Google, Facebook, YouTube and other social networking sites banned in
-    China?
-  - What font is used in Bloomberg Terminal?
-- source_sentence: Is Quora the best Q&A site?
+  - What is Java? What do I need it for?
+  - Why are Instagram filters free?
+  - Can I still get funding to study in UCL (Computer Vision Masters), if I graduated
+    with a BSc in Computer Science with a 2:1 and I have 3 years or web developent
+    expirience?
+- source_sentence: How can I get a job in Dubai if I am living in U.S?
   sentences:
-  - What was the best Quora question ever?
-  - Is Quora the best inquiry site?
-  - Where do I buy Oway hair products online?
-- source_sentence: How can I customize my walking speed on Google Maps?
+  - My Xiaomi Redmi 2 all of a sudden got heated up and then turned off. Now its not
+    charging. What should I do?
+  - How can one get a job in Dubai?
+  - Which is the best series to watch after FRIENDS ?
+- source_sentence: What is the myth behind Mona Lisa smile?
   sentences:
-  - How do I bring back Google maps icon in my home screen?
-  - How many pages are there in all the Harry Potter books combined?
-  - How can I customize my walking speed on Google Maps?
-- source_sentence: DId something exist before the Big Bang?
+  - Do people understand the message in Mona Lisa Smile?
+  - How does the $9 computer work?
+  - What is the myth behind Mona Lisa smile?
+- source_sentence: What is the font used in the desktop version of Instagram for comments?
   sentences:
-  - How can I improve my memory problem?
-  - Where can I buy Fairy Tail Manga?
-  - Is there a scientific name for what existed before the Big Bang?
+  - What is capital of china?
+  - What font is used for “Turon”?
+  - What is the font used in the desktop version of Instagram for comments?
 pipeline_tag: sentence-similarity
 library_name: sentence-transformers
 ---
@@ -85,12 +89,12 @@ Then you can load this model and run inference.
 from sentence_transformers import SentenceTransformer
 
 # Download from the 🤗 Hub
-model = SentenceTransformer("sentence_transformers_model_id")
+model = SentenceTransformer("redis/model-a-baseline")
 # Run inference
 sentences = [
-    'DId something exist before the Big Bang?',
-    'Is there a scientific name for what existed before the Big Bang?',
-    'Where can I buy Fairy Tail Manga?',
+    'What is the font used in the desktop version of Instagram for comments?',
+    'What is the font used in the desktop version of Instagram for comments?',
+    'What font is used for “Turon”?',
 ]
 embeddings = model.encode(sentences)
 print(embeddings.shape)
@@ -99,9 +103,9 @@ print(embeddings.shape)
 # Get the similarity scores for the embeddings
 similarities = model.similarity(embeddings, embeddings)
 print(similarities)
-# tensor([[ 1.0000,  0.7596, -0.0398],
-#         [ 0.7596,  1.0000, -0.0308],
-#         [-0.0398, -0.0308,  1.0000]])
+# tensor([[1.0000, 1.0000, 0.3513],
+#         [1.0000, 1.0000, 0.3513],
+#         [0.3513, 0.3513, 1.0000]])
 ```
 
 <!--
@@ -147,18 +151,18 @@ You can finetune this model on your own dataset.
 #### Unnamed Dataset
 
 * Size: 100,000 training samples
-* Columns: <code>sentence_0</code>, <code>sentence_1</code>, and <code>sentence_2</code>
+* Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
 * Approximate statistics based on the first 1000 samples:
-  |         | sentence_0                                                                        | sentence_1                                                                       | sentence_2                                                                         |
-  |:--------|:----------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
-  | type    | string                                                                            | string                                                                           | string                                                                             |
-  | details | <ul><li>min: 3 tokens</li><li>mean: 15.53 tokens</li><li>max: 59 tokens</li></ul> | <ul><li>min: 3 tokens</li><li>mean: 15.5 tokens</li><li>max: 59 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 16.87 tokens</li><li>max: 128 tokens</li></ul> |
+  |         | anchor                                                                            | positive                                                                          | negative                                                                           |
+  |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
+  | type    | string                                                                            | string                                                                            | string                                                                             |
+  | details | <ul><li>min: 6 tokens</li><li>mean: 15.51 tokens</li><li>max: 97 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 15.34 tokens</li><li>max: 97 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 17.07 tokens</li><li>max: 128 tokens</li></ul> |
 * Samples:
-  | sentence_0                                                                              | sentence_1                                                                              | sentence_2                                                             |
-  |:----------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------|:-----------------------------------------------------------------------|
-  | <code>Is there visitor entry facility in Jaipur airport. How much is the ticket?</code> | <code>Is there visitor entry facility in Jaipur airport. How much is the ticket?</code> | <code>How much is the airport tax in bogota?</code>                    |
-  | <code>Which concept is more important: good planning or hard work?</code>               | <code>Which concept is more important: good planning or hard work?</code>               | <code>What is important in life: luck or hard work?</code>             |
-  | <code>What is the most efficient way to make money?</code>                              | <code>How can I make my money make money?</code>                                        | <code>What can one learn about Quantum Mechanics in 10 minutes?</code> |
+  | anchor                                                   | positive                                               | negative                                                                                        |
+  |:---------------------------------------------------------|:-------------------------------------------------------|:------------------------------------------------------------------------------------------------|
+  | <code>How do you trace a fake phone number?</code>       | <code>How do you trace a fake phone number?</code>     | <code>How do I trace an internet connection phone number from middle East?</code>               |
+  | <code>How do I draw cartoon monsters?</code>             | <code>How do I draw cartoon monsters?</code>           | <code>How do you draw cartoons?</code>                                                          |
+  | <code>Do you believe in an afterlife? If so, why?</code> | <code>Do you believe that there's an afterlife?</code> | <code>When a guy looks at you with dreamy, hooded eyes, does he realize he's doing this?</code> |
 * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
   ```json
   {
@@ -171,10 +175,20 @@ You can finetune this model on your own dataset.
 ### Training Hyperparameters
 #### Non-Default Hyperparameters
 
-- `per_device_train_batch_size`: 64
-- `per_device_eval_batch_size`: 64
+- `per_device_train_batch_size`: 256
+- `per_device_eval_batch_size`: 256
+- `learning_rate`: 2e-05
+- `weight_decay`: 0.001
+- `max_steps`: 1170
+- `warmup_ratio`: 0.1
 - `fp16`: True
-- `multi_dataset_batch_sampler`: round_robin
+- `dataloader_drop_last`: True
+- `dataloader_num_workers`: 1
+- `dataloader_prefetch_factor`: 1
+- `optim`: adamw_torch
+- `ddp_find_unused_parameters`: False
+- `push_to_hub`: True
+- `hub_model_id`: redis/model-a-baseline
 
 #### All Hyperparameters
 <details><summary>Click to expand</summary>
@@ -183,24 +197,24 @@ You can finetune this model on your own dataset.
 - `do_predict`: False
 - `eval_strategy`: no
 - `prediction_loss_only`: True
-- `per_device_train_batch_size`: 64
-- `per_device_eval_batch_size`: 64
+- `per_device_train_batch_size`: 256
+- `per_device_eval_batch_size`: 256
 - `per_gpu_train_batch_size`: None
 - `per_gpu_eval_batch_size`: None
 - `gradient_accumulation_steps`: 1
 - `eval_accumulation_steps`: None
 - `torch_empty_cache_steps`: None
-- `learning_rate`: 5e-05
-- `weight_decay`: 0.0
+- `learning_rate`: 2e-05
+- `weight_decay`: 0.001
 - `adam_beta1`: 0.9
 - `adam_beta2`: 0.999
 - `adam_epsilon`: 1e-08
-- `max_grad_norm`: 1
-- `num_train_epochs`: 3
-- `max_steps`: -1
+- `max_grad_norm`: 1.0
+- `num_train_epochs`: 3.0
+- `max_steps`: 1170
 - `lr_scheduler_type`: linear
 - `lr_scheduler_kwargs`: {}
-- `warmup_ratio`: 0.0
+- `warmup_ratio`: 0.1
 - `warmup_steps`: 0
 - `log_level`: passive
 - `log_level_replica`: warning
@@ -228,9 +242,9 @@ You can finetune this model on your own dataset.
 - `tpu_num_cores`: None
 - `tpu_metrics_debug`: False
 - `debug`: []
-- `dataloader_drop_last`: False
-- `dataloader_num_workers`: 0
-- `dataloader_prefetch_factor`: None
+- `dataloader_drop_last`: True
+- `dataloader_num_workers`: 1
+- `dataloader_prefetch_factor`: 1
 - `past_index`: -1
 - `disable_tqdm`: False
 - `remove_unused_columns`: True
@@ -245,23 +259,23 @@ You can finetune this model on your own dataset.
 - `parallelism_config`: None
 - `deepspeed`: None
 - `label_smoothing_factor`: 0.0
-- `optim`: adamw_torch_fused
+- `optim`: adamw_torch
 - `optim_args`: None
 - `adafactor`: False
 - `group_by_length`: False
 - `length_column_name`: length
 - `project`: huggingface
 - `trackio_space_id`: trackio
-- `ddp_find_unused_parameters`: None
+- `ddp_find_unused_parameters`: False
 - `ddp_bucket_cap_mb`: None
 - `ddp_broadcast_buffers`: False
 - `dataloader_pin_memory`: True
 - `dataloader_persistent_workers`: False
 - `skip_memory_metrics`: True
 - `use_legacy_prediction_loop`: False
-- `push_to_hub`: False
+- `push_to_hub`: True
 - `resume_from_checkpoint`: None
-- `hub_model_id`: None
+- `hub_model_id`: redis/model-a-baseline
 - `hub_strategy`: every_save
 - `hub_private_repo`: None
 - `hub_always_push`: False
@@ -295,7 +309,7 @@ You can finetune this model on your own dataset.
 - `average_tokens_across_devices`: True
 - `prompts`: None
 - `batch_sampler`: batch_sampler
-- `multi_dataset_batch_sampler`: round_robin
+- `multi_dataset_batch_sampler`: proportional
 - `router_mapping`: {}
 - `learning_rate_mapping`: {}
 
@@ -304,15 +318,17 @@ You can finetune this model on your own dataset.
 ### Training Logs
 | Epoch  | Step | Training Loss |
 |:------:|:----:|:-------------:|
-| 0.3199 | 500  | 0.2284        |
-| 0.6398 | 1000 | 0.0571        |
-| 0.9597 | 1500 | 0.0486        |
-| 1.2796 | 2000 | 0.0378        |
-| 1.5995 | 2500 | 0.0367        |
-| 1.9194 | 3000 | 0.0338        |
-| 2.2393 | 3500 | 0.0327        |
-| 2.5592 | 4000 | 0.0285        |
-| 2.8791 | 4500 | 0.0285        |
+| 0.2564 | 100  | 0.8076        |
+| 0.5128 | 200  | 0.1403        |
+| 0.7692 | 300  | 0.1169        |
+| 1.0256 | 400  | 0.1085        |
+| 1.2821 | 500  | 0.0938        |
+| 1.5385 | 600  | 0.09          |
+| 1.7949 | 700  | 0.0898        |
+| 2.0513 | 800  | 0.0826        |
+| 2.3077 | 900  | 0.0797        |
+| 2.5641 | 1000 | 0.081         |
+| 2.8205 | 1100 | 0.0803        |
 
 
 ### Framework Versions