Add new SentenceTransformer model

README.md

@@ -5,38 +5,111 @@ tags:
 - feature-extraction
 - dense
 - generated_from_trainer
-- dataset_size:100000
+- dataset_size:359997
 - loss:MultipleNegativesRankingLoss
 base_model: prajjwal1/bert-small
 widget:
-- source_sentence: How do I polish my English skills?
+- source_sentence: When do you use Ms. or Mrs.? Is one for a married woman and one
+    for one that's not married? Which one is for what?
   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?
+  - When do you use Ms. or Mrs.? Is one for a married woman and one for one that's
+    not married? Which one is for what?
+  - Nations that do/does otherwise? Which one do I use?
+  - Why don't bikes have a gear indicator?
+- source_sentence: Which ointment is applied to the face of UFC fighters at the commencement
+    of a bout? What does it do?
   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?
+  - How can I save a Snapchat video that others posted?
+  - Which ointment is applied to the face of UFC fighters at the commencement of a
+    bout? What does it do?
+  - How do I get the body of a UFC Fighter?
+- source_sentence: Do you love the life you live?
   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?
+  - Can I do shoulder and triceps workout on same day? What other combinations like
+    this can I do?
+  - Do you love the life you're living?
+  - Where can you find an online TI-84 calculator?
+- source_sentence: Ordered food on Swiggy 3 days ago.After accepting my money, said
+    no more on Menu! When if ever will I atleast get refund in cr card a/c?
   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?
+  - Is getting to the Tel Aviv airport to catch a 5:30 AM flight very expensive?
+  - How do I die and make it look like an accident?
+  - Ordered food on Swiggy 3 days ago.After accepting my money, said no more on Menu!
+    When if ever will I atleast get refund in cr card a/c?
+- source_sentence: How do you earn money on Quora?
   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 a cheap healthy diet I can keep the same and eat every day?
+  - What are some things new employees should know going into their first day at Maximus?
+  - What is the best way to make money on Quora?
 pipeline_tag: sentence-similarity
 library_name: sentence-transformers
+metrics:
+- cosine_accuracy@1
+- cosine_accuracy@3
+- cosine_accuracy@5
+- cosine_precision@1
+- cosine_precision@3
+- cosine_precision@5
+- cosine_recall@1
+- cosine_recall@3
+- cosine_recall@5
+- cosine_ndcg@10
+- cosine_mrr@1
+- cosine_mrr@5
+- cosine_mrr@10
+- cosine_map@100
+model-index:
+- name: SentenceTransformer based on prajjwal1/bert-small
+  results:
+  - task:
+      type: information-retrieval
+      name: Information Retrieval
+    dataset:
+      name: val
+      type: val
+    metrics:
+    - type: cosine_accuracy@1
+      value: 0.802275
+      name: Cosine Accuracy@1
+    - type: cosine_accuracy@3
+      value: 0.8676
+      name: Cosine Accuracy@3
+    - type: cosine_accuracy@5
+      value: 0.89495
+      name: Cosine Accuracy@5
+    - type: cosine_precision@1
+      value: 0.802275
+      name: Cosine Precision@1
+    - type: cosine_precision@3
+      value: 0.28919999999999996
+      name: Cosine Precision@3
+    - type: cosine_precision@5
+      value: 0.17899
+      name: Cosine Precision@5
+    - type: cosine_recall@1
+      value: 0.802275
+      name: Cosine Recall@1
+    - type: cosine_recall@3
+      value: 0.8676
+      name: Cosine Recall@3
+    - type: cosine_recall@5
+      value: 0.89495
+      name: Cosine Recall@5
+    - type: cosine_ndcg@10
+      value: 0.8626142815266079
+      name: Cosine Ndcg@10
+    - type: cosine_mrr@1
+      value: 0.802275
+      name: Cosine Mrr@1
+    - type: cosine_mrr@5
+      value: 0.8373504166666614
+      name: Cosine Mrr@5
+    - type: cosine_mrr@10
+      value: 0.8418555952380918
+      name: Cosine Mrr@10
+    - type: cosine_map@100
+      value: 0.8448604581747029
+      name: Cosine Map@100
 ---
 
 # SentenceTransformer based on prajjwal1/bert-small
@@ -85,12 +158,12 @@ 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?',
+    'How do you earn money on Quora?',
+    'What is the best way to make money on Quora?',
+    'What are some things new employees should know going into their first day at Maximus?',
 ]
 embeddings = model.encode(sentences)
 print(embeddings.shape)
@@ -99,9 +172,9 @@ 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([[ 1.0001,  0.9960, -0.0211],
+#         [ 0.9960,  1.0000, -0.0245],
+#         [-0.0211, -0.0245,  1.0000]])
 ```
 
 <!--
@@ -128,6 +201,32 @@ You can finetune this model on your own dataset.
 *List how the model may foreseeably be misused and address what users ought not to do with the model.*
 -->
 
+## Evaluation
+
+### Metrics
+
+#### Information Retrieval
+
+* Dataset: `val`
+* Evaluated with [<code>InformationRetrievalEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.InformationRetrievalEvaluator)
+
+| Metric             | Value      |
+|:-------------------|:-----------|
+| cosine_accuracy@1  | 0.8023     |
+| cosine_accuracy@3  | 0.8676     |
+| cosine_accuracy@5  | 0.895      |
+| cosine_precision@1 | 0.8023     |
+| cosine_precision@3 | 0.2892     |
+| cosine_precision@5 | 0.179      |
+| cosine_recall@1    | 0.8023     |
+| cosine_recall@3    | 0.8676     |
+| cosine_recall@5    | 0.895      |
+| **cosine_ndcg@10** | **0.8626** |
+| cosine_mrr@1       | 0.8023     |
+| cosine_mrr@5       | 0.8374     |
+| cosine_mrr@10      | 0.8419     |
+| cosine_map@100     | 0.8449     |
+
 <!--
 ## Bias, Risks and Limitations
 
@@ -146,23 +245,49 @@ 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>
+* Size: 359,997 training samples
+* Columns: <code>anchor</code>, <code>positive</code>, and <code>negative</code>
+* Approximate statistics based on the first 1000 samples:
+  |         | anchor                                                                            | positive                                                                          | negative                                                                           |
+  |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------|
+  | type    | string                                                                            | string                                                                            | string                                                                             |
+  | details | <ul><li>min: 4 tokens</li><li>mean: 15.46 tokens</li><li>max: 49 tokens</li></ul> | <ul><li>min: 4 tokens</li><li>mean: 15.52 tokens</li><li>max: 49 tokens</li></ul> | <ul><li>min: 5 tokens</li><li>mean: 16.99 tokens</li><li>max: 128 tokens</li></ul> |
+* Samples:
+  | anchor                                                                          | positive                                                                                             | negative                                                                                                                 |
+  |:--------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------|:-------------------------------------------------------------------------------------------------------------------------|
+  | <code>Shall I upgrade my iPhone 5s to iOS 10 final version?</code>              | <code>Should I upgrade an iPhone 5s to iOS 10?</code>                                                | <code>Whether extension of CA-articleship is to be served at same firm/company?</code>                                   |
+  | <code>Is Donald Trump really going to be the president of United States?</code> | <code>Do you think Donald Trump could conceivably be the next President of the United States?</code> | <code>Since solid carbon dioxide is dry ice and incredibly cold, why doesn't it have an effect on global warming?</code> |
+  | <code>What are real tips to improve work life balance?</code>                   | <code>What are the best ways to create a work life balance?</code>                                   | <code>How do you open a briefcase combination lock without the combination?</code>                                       |
+* Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
+  ```json
+  {
+      "scale": 5.0,
+      "similarity_fct": "cos_sim",
+      "gather_across_devices": false
+  }
+  ```
+
+### Evaluation Dataset
+
+#### Unnamed Dataset
+
+* Size: 40,000 evaluation samples
+* 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: 15.71 tokens</li><li>max: 65 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 15.79 tokens</li><li>max: 65 tokens</li></ul> | <ul><li>min: 6 tokens</li><li>mean: 16.97 tokens</li><li>max: 78 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>Why were feathered dinosaur fossils only found in the last 20 years?</code>                                 | <code>Why were feathered dinosaur fossils only found in the last 20 years?</code>                                 | <code>Why are only few people aware that many dinosaurs had feathers?</code>                                                              |
+  | <code>If FOX News is the conservative news station, which cable news network is for liberals/progressives?</code> | <code>If FOX News is the conservative news station, which cable news network is for liberals/progressives?</code> | <code>How much did Fox News and conservative leaning media networks stoke the anger that contributed to Donald Trump's popularity?</code> |
+  | <code>How can guys last longer during sex?</code>                                                                 | <code>How do I last longer in sex?</code>                                                                         | <code>What is a permanent solution for rough and puffy hair?</code>                                                                       |
 * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
   ```json
   {
-      "scale": 20.0,
+      "scale": 5.0,
       "similarity_fct": "cos_sim",
       "gather_across_devices": false
   }
@@ -171,36 +296,49 @@ 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
+- `eval_strategy`: steps
+- `per_device_train_batch_size`: 1024
+- `per_device_eval_batch_size`: 1024
+- `learning_rate`: 2e-05
+- `weight_decay`: 0.001
+- `max_steps`: 5000
+- `warmup_ratio`: 0.1
 - `fp16`: True
-- `multi_dataset_batch_sampler`: round_robin
+- `dataloader_drop_last`: True
+- `dataloader_num_workers`: 1
+- `dataloader_prefetch_factor`: 1
+- `load_best_model_at_end`: True
+- `optim`: adamw_torch
+- `ddp_find_unused_parameters`: False
+- `push_to_hub`: True
+- `hub_model_id`: redis/model-a-baseline
+- `eval_on_start`: True
 
 #### All Hyperparameters
 <details><summary>Click to expand</summary>
 
 - `overwrite_output_dir`: False
 - `do_predict`: False
-- `eval_strategy`: no
+- `eval_strategy`: steps
 - `prediction_loss_only`: True
-- `per_device_train_batch_size`: 64
-- `per_device_eval_batch_size`: 64
+- `per_device_train_batch_size`: 1024
+- `per_device_eval_batch_size`: 1024
 - `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`: 5000
 - `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
@@ -228,14 +366,14 @@ 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
 - `label_names`: None
-- `load_best_model_at_end`: False
+- `load_best_model_at_end`: True
 - `ignore_data_skip`: False
 - `fsdp`: []
 - `fsdp_min_num_params`: 0
@@ -245,23 +383,23 @@ 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
@@ -288,31 +426,73 @@ You can finetune this model on your own dataset.
 - `neftune_noise_alpha`: None
 - `optim_target_modules`: None
 - `batch_eval_metrics`: False
-- `eval_on_start`: False
+- `eval_on_start`: True
 - `use_liger_kernel`: False
 - `liger_kernel_config`: None
 - `eval_use_gather_object`: False
 - `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`: {}
 
 </details>
 
 ### 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        |
+| Epoch   | Step | Training Loss | Validation Loss | val_cosine_ndcg@10 |
+|:-------:|:----:|:-------------:|:---------------:|:------------------:|
+| 0       | 0    | -             | 6.2276          | 0.8046             |
+| 0.2849  | 100  | 5.882         | 3.9651          | 0.8557             |
+| 0.5698  | 200  | 4.3197        | 3.2712          | 0.8551             |
+| 0.8547  | 300  | 3.93          | 3.2258          | 0.8543             |
+| 1.1396  | 400  | 3.7914        | 3.1976          | 0.8544             |
+| 1.4245  | 500  | 3.7016        | 3.1692          | 0.8550             |
+| 1.7094  | 600  | 3.6345        | 3.1525          | 0.8556             |
+| 1.9943  | 700  | 3.5897        | 3.1350          | 0.8563             |
+| 2.2792  | 800  | 3.5494        | 3.1225          | 0.8566             |
+| 2.5641  | 900  | 3.5241        | 3.1107          | 0.8570             |
+| 2.8490  | 1000 | 3.4975        | 3.1020          | 0.8572             |
+| 3.1339  | 1100 | 3.4786        | 3.0944          | 0.8574             |
+| 3.4188  | 1200 | 3.4606        | 3.0875          | 0.8579             |
+| 3.7037  | 1300 | 3.4466        | 3.0821          | 0.8582             |
+| 3.9886  | 1400 | 3.4327        | 3.0765          | 0.8586             |
+| 4.2735  | 1500 | 3.4196        | 3.0725          | 0.8589             |
+| 4.5584  | 1600 | 3.4075        | 3.0682          | 0.8592             |
+| 4.8433  | 1700 | 3.4001        | 3.0639          | 0.8593             |
+| 5.1282  | 1800 | 3.3895        | 3.0598          | 0.8597             |
+| 5.4131  | 1900 | 3.381         | 3.0564          | 0.8599             |
+| 5.6980  | 2000 | 3.3751        | 3.0531          | 0.8601             |
+| 5.9829  | 2100 | 3.3672        | 3.0507          | 0.8603             |
+| 6.2678  | 2200 | 3.3571        | 3.0486          | 0.8605             |
+| 6.5527  | 2300 | 3.3549        | 3.0447          | 0.8608             |
+| 6.8376  | 2400 | 3.3471        | 3.0432          | 0.8608             |
+| 7.1225  | 2500 | 3.3428        | 3.0411          | 0.8609             |
+| 7.4074  | 2600 | 3.3358        | 3.0390          | 0.8613             |
+| 7.6923  | 2700 | 3.3304        | 3.0377          | 0.8612             |
+| 7.9772  | 2800 | 3.3283        | 3.0358          | 0.8614             |
+| 8.2621  | 2900 | 3.3217        | 3.0334          | 0.8616             |
+| 8.5470  | 3000 | 3.319         | 3.0324          | 0.8618             |
+| 8.8319  | 3100 | 3.3166        | 3.0305          | 0.8620             |
+| 9.1168  | 3200 | 3.3115        | 3.0295          | 0.8620             |
+| 9.4017  | 3300 | 3.3077        | 3.0279          | 0.8619             |
+| 9.6866  | 3400 | 3.3044        | 3.0270          | 0.8620             |
+| 9.9715  | 3500 | 3.3031        | 3.0256          | 0.8623             |
+| 10.2564 | 3600 | 3.2988        | 3.0253          | 0.8623             |
+| 10.5413 | 3700 | 3.2968        | 3.0239          | 0.8623             |
+| 10.8262 | 3800 | 3.295         | 3.0233          | 0.8624             |
+| 11.1111 | 3900 | 3.2933        | 3.0227          | 0.8624             |
+| 11.3960 | 4000 | 3.2913        | 3.0217          | 0.8626             |
+| 11.6809 | 4100 | 3.2884        | 3.0211          | 0.8624             |
+| 11.9658 | 4200 | 3.2875        | 3.0207          | 0.8625             |
+| 12.2507 | 4300 | 3.2859        | 3.0202          | 0.8626             |
+| 12.5356 | 4400 | 3.2841        | 3.0196          | 0.8626             |
+| 12.8205 | 4500 | 3.2839        | 3.0193          | 0.8625             |
+| 13.1054 | 4600 | 3.2839        | 3.0188          | 0.8627             |
+| 13.3903 | 4700 | 3.2818        | 3.0188          | 0.8626             |
+| 13.6752 | 4800 | 3.2799        | 3.0185          | 0.8627             |
+| 13.9601 | 4900 | 3.2812        | 3.0183          | 0.8627             |
+| 14.2450 | 5000 | 3.2817        | 3.0183          | 0.8626             |
 
 
 ### Framework Versions