Training in progress, step 5000

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text

1_Pooling/config.json

@@ -1,7 +1,7 @@
 {
-    "word_embedding_dimension": 384,
-    "pooling_mode_cls_token": false,
-    "pooling_mode_mean_tokens": true,
+    "word_embedding_dimension": 512,
+    "pooling_mode_cls_token": true,
+    "pooling_mode_mean_tokens": false,
     "pooling_mode_max_tokens": false,
     "pooling_mode_mean_sqrt_len_tokens": false,
     "pooling_mode_weightedmean_tokens": false,

Information-Retrieval_evaluation_val_results.csv

@@ -9,3 +9,4 @@ epoch,steps,cosine-Accuracy@1,cosine-Accuracy@3,cosine-Accuracy@5,cosine-Precisi
 -1,-1,0.82975,0.903025,0.9308,0.82975,0.82975,0.3010083333333333,0.903025,0.18616000000000002,0.9308,0.82975,0.8688179166666645,0.8729221527777756,0.894185079953941,0.8751251735048098
 -1,-1,0.8288,0.899775,0.925775,0.8288,0.8288,0.29992499999999994,0.899775,0.185155,0.925775,0.8288,0.8661879166666627,0.8703450396825356,0.8910978019383597,0.8726020537429935
 -1,-1,0.826575,0.900725,0.92805,0.826575,0.826575,0.30024166666666663,0.900725,0.18561000000000002,0.92805,0.826575,0.8658308333333287,0.8701137103174557,0.891705546917102,0.8723575730144177
+-1,-1,0.82585,0.902175,0.930075,0.82585,0.82585,0.30072499999999996,0.902175,0.186015,0.930075,0.82585,0.8661279166666617,0.8703281448412645,0.8922105025555344,0.8724788643099791

README.md

@@ -5,124 +5,51 @@ tags:
 - feature-extraction
 - dense
 - generated_from_trainer
-- dataset_size:359997
+- dataset_size:100000
 - loss:MultipleNegativesRankingLoss
-base_model: thenlper/gte-small
+base_model: prajjwal1/bert-small
 widget:
-- 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?
+- source_sentence: How do I polish my English skills?
   sentences:
-  - 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?
+  - 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?
   sentences:
-  - 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?
+  - 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?
   sentences:
-  - 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?
+  - 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?
   sentences:
-  - 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?
+  - 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?
   sentences:
-  - 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?
+  - 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?
 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 thenlper/gte-small
-  results:
-  - task:
-      type: information-retrieval
-      name: Information Retrieval
-    dataset:
-      name: val
-      type: val
-    metrics:
-    - type: cosine_accuracy@1
-      value: 0.82585
-      name: Cosine Accuracy@1
-    - type: cosine_accuracy@3
-      value: 0.902175
-      name: Cosine Accuracy@3
-    - type: cosine_accuracy@5
-      value: 0.930075
-      name: Cosine Accuracy@5
-    - type: cosine_precision@1
-      value: 0.82585
-      name: Cosine Precision@1
-    - type: cosine_precision@3
-      value: 0.30072499999999996
-      name: Cosine Precision@3
-    - type: cosine_precision@5
-      value: 0.186015
-      name: Cosine Precision@5
-    - type: cosine_recall@1
-      value: 0.82585
-      name: Cosine Recall@1
-    - type: cosine_recall@3
-      value: 0.902175
-      name: Cosine Recall@3
-    - type: cosine_recall@5
-      value: 0.930075
-      name: Cosine Recall@5
-    - type: cosine_ndcg@10
-      value: 0.8922105025555344
-      name: Cosine Ndcg@10
-    - type: cosine_mrr@1
-      value: 0.82585
-      name: Cosine Mrr@1
-    - type: cosine_mrr@5
-      value: 0.8661279166666617
-      name: Cosine Mrr@5
-    - type: cosine_mrr@10
-      value: 0.8703281448412645
-      name: Cosine Mrr@10
-    - type: cosine_map@100
-      value: 0.8724788643099791
-      name: Cosine Map@100
 ---
 
-# SentenceTransformer based on thenlper/gte-small
+# SentenceTransformer based on prajjwal1/bert-small
 
-This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [thenlper/gte-small](https://huggingface.co/thenlper/gte-small). It maps sentences & paragraphs to a 384-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
+This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [prajjwal1/bert-small](https://huggingface.co/prajjwal1/bert-small). It maps sentences & paragraphs to a 512-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
 
 ## Model Details
 
 ### Model Description
 - **Model Type:** Sentence Transformer
-- **Base model:** [thenlper/gte-small](https://huggingface.co/thenlper/gte-small) <!-- at revision 17e1f347d17fe144873b1201da91788898c639cd -->
+- **Base model:** [prajjwal1/bert-small](https://huggingface.co/prajjwal1/bert-small) <!-- at revision 0ec5f86f27c1a77d704439db5e01c307ea11b9d4 -->
 - **Maximum Sequence Length:** 128 tokens
-- **Output Dimensionality:** 384 dimensions
+- **Output Dimensionality:** 512 dimensions
 - **Similarity Function:** Cosine Similarity
 <!-- - **Training Dataset:** Unknown -->
 <!-- - **Language:** Unknown -->
@@ -139,8 +66,7 @@ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [t
 ```
 SentenceTransformer(
   (0): Transformer({'max_seq_length': 128, 'do_lower_case': False, 'architecture': 'BertModel'})
-  (1): Pooling({'word_embedding_dimension': 384, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
-  (2): Normalize()
+  (1): Pooling({'word_embedding_dimension': 512, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
 )
 ```
 
@@ -159,23 +85,23 @@ Then you can load this model and run inference.
 from sentence_transformers import SentenceTransformer
 
 # Download from the 🤗 Hub
-model = SentenceTransformer("redis/model-a-baseline")
+model = SentenceTransformer("sentence_transformers_model_id")
 # Run inference
 sentences = [
-    '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?',
+    '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?',
 ]
 embeddings = model.encode(sentences)
 print(embeddings.shape)
-# [3, 384]
+# [3, 512]
 
 # Get the similarity scores for the embeddings
 similarities = model.similarity(embeddings, embeddings)
 print(similarities)
-# tensor([[1.0000, 0.9984, 0.0403],
-#         [0.9984, 1.0000, 0.0422],
-#         [0.0403, 0.0422, 1.0000]])
+# tensor([[ 1.0000,  0.7596, -0.0398],
+#         [ 0.7596,  1.0000, -0.0308],
+#         [-0.0398, -0.0308,  1.0000]])
 ```
 
 <!--
@@ -202,32 +128,6 @@ 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  | 0.8258     |
-| cosine_accuracy@3  | 0.9022     |
-| cosine_accuracy@5  | 0.9301     |
-| cosine_precision@1 | 0.8258     |
-| cosine_precision@3 | 0.3007     |
-| cosine_precision@5 | 0.186      |
-| cosine_recall@1    | 0.8258     |
-| cosine_recall@3    | 0.9022     |
-| cosine_recall@5    | 0.9301     |
-| **cosine_ndcg@10** | **0.8922** |
-| cosine_mrr@1       | 0.8258     |
-| cosine_mrr@5       | 0.8661     |
-| cosine_mrr@10      | 0.8703     |
-| cosine_map@100     | 0.8725     |
-
 <!--
 ## Bias, Risks and Limitations
 
@@ -246,49 +146,23 @@ You can finetune this model on your own dataset.
 
 #### Unnamed Dataset
 
-* 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": 7.0,
-      "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>
+* Size: 100,000 training samples
+* Columns: <code>sentence_0</code>, <code>sentence_1</code>, and <code>sentence_2</code>
 * Approximate statistics based on the first 1000 samples:
-  |         | 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> |
+  |         | 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> |
 * Samples:
-  | 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>                                                                       |
+  | 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> |
 * Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
   ```json
   {
-      "scale": 7.0,
+      "scale": 20.0,
       "similarity_fct": "cos_sim",
       "gather_across_devices": false
   }
@@ -297,49 +171,36 @@ You can finetune this model on your own dataset.
 ### Training Hyperparameters
 #### Non-Default Hyperparameters
 
-- `eval_strategy`: steps
-- `per_device_train_batch_size`: 128
-- `per_device_eval_batch_size`: 128
-- `learning_rate`: 0.0002
-- `weight_decay`: 0.0001
-- `max_steps`: 10000
-- `warmup_ratio`: 0.1
+- `per_device_train_batch_size`: 64
+- `per_device_eval_batch_size`: 64
 - `fp16`: True
-- `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
+- `multi_dataset_batch_sampler`: round_robin
 
 #### All Hyperparameters
 <details><summary>Click to expand</summary>
 
 - `overwrite_output_dir`: False
 - `do_predict`: False
-- `eval_strategy`: steps
+- `eval_strategy`: no
 - `prediction_loss_only`: True
-- `per_device_train_batch_size`: 128
-- `per_device_eval_batch_size`: 128
+- `per_device_train_batch_size`: 64
+- `per_device_eval_batch_size`: 64
 - `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`: 0.0002
-- `weight_decay`: 0.0001
+- `learning_rate`: 5e-05
+- `weight_decay`: 0.0
 - `adam_beta1`: 0.9
 - `adam_beta2`: 0.999
 - `adam_epsilon`: 1e-08
-- `max_grad_norm`: 1.0
-- `num_train_epochs`: 3.0
-- `max_steps`: 10000
+- `max_grad_norm`: 1
+- `num_train_epochs`: 3
+- `max_steps`: -1
 - `lr_scheduler_type`: linear
 - `lr_scheduler_kwargs`: {}
-- `warmup_ratio`: 0.1
+- `warmup_ratio`: 0.0
 - `warmup_steps`: 0
 - `log_level`: passive
 - `log_level_replica`: warning
@@ -367,14 +228,14 @@ You can finetune this model on your own dataset.
 - `tpu_num_cores`: None
 - `tpu_metrics_debug`: False
 - `debug`: []
-- `dataloader_drop_last`: True
-- `dataloader_num_workers`: 1
-- `dataloader_prefetch_factor`: 1
+- `dataloader_drop_last`: False
+- `dataloader_num_workers`: 0
+- `dataloader_prefetch_factor`: None
 - `past_index`: -1
 - `disable_tqdm`: False
 - `remove_unused_columns`: True
 - `label_names`: None
-- `load_best_model_at_end`: True
+- `load_best_model_at_end`: False
 - `ignore_data_skip`: False
 - `fsdp`: []
 - `fsdp_min_num_params`: 0
@@ -384,23 +245,23 @@ You can finetune this model on your own dataset.
 - `parallelism_config`: None
 - `deepspeed`: None
 - `label_smoothing_factor`: 0.0
-- `optim`: adamw_torch
+- `optim`: adamw_torch_fused
 - `optim_args`: None
 - `adafactor`: False
 - `group_by_length`: False
 - `length_column_name`: length
 - `project`: huggingface
 - `trackio_space_id`: trackio
-- `ddp_find_unused_parameters`: False
+- `ddp_find_unused_parameters`: None
 - `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`: True
+- `push_to_hub`: False
 - `resume_from_checkpoint`: None
-- `hub_model_id`: redis/model-a-baseline
+- `hub_model_id`: None
 - `hub_strategy`: every_save
 - `hub_private_repo`: None
 - `hub_always_push`: False
@@ -427,63 +288,31 @@ You can finetune this model on your own dataset.
 - `neftune_noise_alpha`: None
 - `optim_target_modules`: None
 - `batch_eval_metrics`: False
-- `eval_on_start`: True
+- `eval_on_start`: False
 - `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`: proportional
+- `multi_dataset_batch_sampler`: round_robin
 - `router_mapping`: {}
 - `learning_rate_mapping`: {}
 
 </details>
 
 ### Training Logs
-| Epoch  | Step  | Training Loss | Validation Loss | val_cosine_ndcg@10 |
-|:------:|:-----:|:-------------:|:---------------:|:------------------:|
-| 0      | 0     | -             | 3.6640          | 0.8927             |
-| 0.0889 | 250   | 1.5177        | 0.4409          | 0.8904             |
-| 0.1778 | 500   | 0.5386        | 0.4150          | 0.8907             |
-| 0.2667 | 750   | 0.5035        | 0.4014          | 0.8899             |
-| 0.3556 | 1000  | 0.4858        | 0.4000          | 0.8887             |
-| 0.4445 | 1250  | 0.4715        | 0.3927          | 0.8883             |
-| 0.5334 | 1500  | 0.4644        | 0.3875          | 0.8889             |
-| 0.6223 | 1750  | 0.4555        | 0.3821          | 0.8898             |
-| 0.7112 | 2000  | 0.4476        | 0.3795          | 0.8890             |
-| 0.8001 | 2250  | 0.4403        | 0.3779          | 0.8894             |
-| 0.8890 | 2500  | 0.4375        | 0.3750          | 0.8903             |
-| 0.9780 | 2750  | 0.4323        | 0.3718          | 0.8893             |
-| 1.0669 | 3000  | 0.4132        | 0.3665          | 0.8895             |
-| 1.1558 | 3250  | 0.4081        | 0.3660          | 0.8883             |
-| 1.2447 | 3500  | 0.4047        | 0.3650          | 0.8894             |
-| 1.3336 | 3750  | 0.403         | 0.3624          | 0.8905             |
-| 1.4225 | 4000  | 0.4003        | 0.3608          | 0.8901             |
-| 1.5114 | 4250  | 0.3986        | 0.3595          | 0.8903             |
-| 1.6003 | 4500  | 0.3982        | 0.3580          | 0.8911             |
-| 1.6892 | 4750  | 0.3951        | 0.3572          | 0.8911             |
-| 1.7781 | 5000  | 0.3963        | 0.3560          | 0.8915             |
-| 1.8670 | 5250  | 0.3925        | 0.3549          | 0.8913             |
-| 1.9559 | 5500  | 0.3922        | 0.3535          | 0.8920             |
-| 2.0448 | 5750  | 0.3794        | 0.3512          | 0.8913             |
-| 2.1337 | 6000  | 0.37          | 0.3501          | 0.8911             |
-| 2.2226 | 6250  | 0.3702        | 0.3504          | 0.8913             |
-| 2.3115 | 6500  | 0.3696        | 0.3491          | 0.8915             |
-| 2.4004 | 6750  | 0.3685        | 0.3482          | 0.8922             |
-| 2.4893 | 7000  | 0.3675        | 0.3470          | 0.8920             |
-| 2.5782 | 7250  | 0.3659        | 0.3460          | 0.8916             |
-| 2.6671 | 7500  | 0.3634        | 0.3459          | 0.8915             |
-| 2.7560 | 7750  | 0.3657        | 0.3448          | 0.8918             |
-| 2.8450 | 8000  | 0.3639        | 0.3442          | 0.8919             |
-| 2.9339 | 8250  | 0.3623        | 0.3430          | 0.8923             |
-| 3.0228 | 8500  | 0.3603        | 0.3425          | 0.8920             |
-| 3.1117 | 8750  | 0.3504        | 0.3424          | 0.8917             |
-| 3.2006 | 9000  | 0.3501        | 0.3419          | 0.8920             |
-| 3.2895 | 9250  | 0.3505        | 0.3418          | 0.8920             |
-| 3.3784 | 9500  | 0.3483        | 0.3413          | 0.8922             |
-| 3.4673 | 9750  | 0.3478        | 0.3410          | 0.8920             |
-| 3.5562 | 10000 | 0.3492        | 0.3408          | 0.8922             |
+| 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        |
 
 
 ### Framework Versions

config.json

@@ -1,24 +1,60 @@
 {
+  "_sliding_window_pattern": 6,
   "architectures": [
-    "BertModel"
+    "Gemma3TextModel"
   ],
-  "attention_probs_dropout_prob": 0.1,
-  "classifier_dropout": null,
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "attn_logit_softcapping": null,
+  "bos_token_id": 2,
   "dtype": "float32",
-  "hidden_act": "gelu",
-  "hidden_dropout_prob": 0.1,
-  "hidden_size": 384,
+  "eos_token_id": 1,
+  "final_logit_softcapping": null,
+  "head_dim": 256,
+  "hidden_activation": "gelu_pytorch_tanh",
+  "hidden_size": 768,
   "initializer_range": 0.02,
-  "intermediate_size": 1536,
-  "layer_norm_eps": 1e-12,
-  "max_position_embeddings": 512,
-  "model_type": "bert",
-  "num_attention_heads": 12,
-  "num_hidden_layers": 12,
+  "intermediate_size": 1152,
+  "layer_types": [
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "max_position_embeddings": 2048,
+  "model_type": "gemma3_text",
+  "num_attention_heads": 3,
+  "num_hidden_layers": 24,
+  "num_key_value_heads": 1,
   "pad_token_id": 0,
-  "position_embedding_type": "absolute",
+  "query_pre_attn_scalar": 256,
+  "rms_norm_eps": 1e-06,
+  "rope_local_base_freq": 10000.0,
+  "rope_scaling": null,
+  "rope_theta": 1000000.0,
+  "sliding_window": 257,
   "transformers_version": "4.57.3",
-  "type_vocab_size": 2,
+  "use_bidirectional_attention": true,
   "use_cache": true,
-  "vocab_size": 30522
+  "vocab_size": 262144
 }

eval/Information-Retrieval_evaluation_val_results.csv

@@ -796,3 +796,25 @@ epoch,steps,cosine-Accuracy@1,cosine-Accuracy@3,cosine-Accuracy@5,cosine-Precisi
 3.3783783783783785,9500,0.825675,0.9017,0.930125,0.825675,0.825675,0.30056666666666665,0.9017,0.186025,0.930125,0.825675,0.8659704166666623,0.8701954761904717,0.8921878189564996,0.87231294268874
 3.4672830725462305,9750,0.82565,0.902125,0.930125,0.82565,0.82565,0.3007083333333333,0.902125,0.186025,0.930125,0.82565,0.865978333333329,0.8701194940476146,0.8919567604167592,0.8723131653149703
 3.5561877667140824,10000,0.82585,0.902175,0.930075,0.82585,0.82585,0.30072499999999996,0.902175,0.186015,0.930075,0.82585,0.8661279166666617,0.8703281448412645,0.8922105025555344,0.8724788643099791
+0,0,0.826075,0.9019,0.926375,0.826075,0.826075,0.3006333333333333,0.9019,0.18527500000000002,0.926375,0.826075,0.8656154166666614,0.869528363095233,0.8903094666814794,0.8717042027739657
+0,0,0.8261,0.90195,0.926325,0.8261,0.8261,0.3006499999999999,0.90195,0.185265,0.926325,0.8261,0.8656149999999949,0.8695360813492012,0.8903149406977251,0.8717113603841892
+0.04445234708392603,250,0.821025,0.89705,0.923075,0.821025,0.821025,0.29901666666666665,0.89705,0.18461500000000006,0.923075,0.821025,0.8610470833333292,0.8650748412698367,0.8863330153490923,0.8674308798239341
+0.08890469416785206,500,0.823475,0.8979,0.92295,0.823475,0.823475,0.29929999999999995,0.8979,0.18459000000000003,0.92295,0.823475,0.8622916666666625,0.8664229067460257,0.8874503348960284,0.8687474801005728
+0.13335704125177808,750,0.011325,0.0256,0.03865,0.011325,0.011325,0.008533333333333332,0.0256,0.007730000000000001,0.03865,0.011325,0.02088541666666677,0.022595833333333384,0.029270205325053842,0.035727285015858624
+0.17780938833570412,1000,0.823625,0.8967,0.923175,0.823625,0.823625,0.2989,0.8967,0.18463500000000005,0.923175,0.823625,0.8622445833333304,0.866245079365076,0.8871180256539423,0.8685892558516832
+0.22226173541963015,1250,0.0006,0.7982,0.88095,0.0006,0.0006,0.2660666666666667,0.7982,0.17619,0.88095,0.0006,0.2884245833334066,0.2949442757937322,0.4513885856647716,0.2977867903240451
+0.26671408250355616,1500,0.0006,0.7985,0.881125,0.0006,0.0006,0.26616666666666666,0.7985,0.176225,0.881125,0.0006,0.28853791666673956,0.2950232043651601,0.4514484069261291,0.29788707270526127
+0.3111664295874822,1750,0.000775,0.7987,0.880525,0.000775,0.000775,0.2662333333333333,0.7987,0.176105,0.880525,0.000775,0.288679583333407,0.29516037698420866,0.45141007661888277,0.2980129241232368
+0.35561877667140823,2000,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.40007112375533427,2250,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.4445234708392603,2500,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.48897581792318634,2750,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.5334281650071123,3000,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.5778805120910384,3250,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.6223328591749644,3500,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.6667852062588905,3750,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.7112375533428165,4000,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.7556899004267426,4250,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.8001422475106685,4500,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.8445945945945946,4750,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344
+0.8890469416785206,5000,0.00065,0.7986,0.880825,0.00065,0.00065,0.26619999999999994,0.7986,0.17616500000000002,0.880825,0.00065,0.288667083333407,0.2951483234127803,0.45147470340355694,0.2980051496600344

final_metrics.json

@@ -1,16 +1,16 @@
 {
-  "val_cosine_accuracy@1": 0.826575,
-  "val_cosine_accuracy@3": 0.900725,
-  "val_cosine_accuracy@5": 0.92805,
-  "val_cosine_precision@1": 0.826575,
-  "val_cosine_precision@3": 0.30024166666666663,
-  "val_cosine_precision@5": 0.18561000000000002,
-  "val_cosine_recall@1": 0.826575,
-  "val_cosine_recall@3": 0.900725,
-  "val_cosine_recall@5": 0.92805,
-  "val_cosine_ndcg@10": 0.891705546917102,
-  "val_cosine_mrr@1": 0.826575,
-  "val_cosine_mrr@5": 0.8658308333333287,
-  "val_cosine_mrr@10": 0.8701137103174557,
-  "val_cosine_map@100": 0.8723575730144177
+  "val_cosine_accuracy@1": 0.82585,
+  "val_cosine_accuracy@3": 0.902175,
+  "val_cosine_accuracy@5": 0.930075,
+  "val_cosine_precision@1": 0.82585,
+  "val_cosine_precision@3": 0.30072499999999996,
+  "val_cosine_precision@5": 0.186015,
+  "val_cosine_recall@1": 0.82585,
+  "val_cosine_recall@3": 0.902175,
+  "val_cosine_recall@5": 0.930075,
+  "val_cosine_ndcg@10": 0.8922105025555344,
+  "val_cosine_mrr@1": 0.82585,
+  "val_cosine_mrr@5": 0.8661279166666617,
+  "val_cosine_mrr@10": 0.8703281448412645,
+  "val_cosine_map@100": 0.8724788643099791
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:368ea36921477c1e5c05b855edd91867c031b131e8d9b5c31b5b7d66e904530f
-size 133462128
+oid sha256:9f9a47024db2f08e7f0c48f2a85ed40a9741f18dc45828790d6b88202b309171
+size 1211486072

modules.json

@@ -10,11 +10,5 @@
     "name": "1",
     "path": "1_Pooling",
     "type": "sentence_transformers.models.Pooling"
-  },
-  {
-    "idx": 2,
-    "name": "2",
-    "path": "2_Normalize",
-    "type": "sentence_transformers.models.Normalize"
   }
 ]

special_tokens_map.json

@@ -1,34 +1,30 @@
 {
-  "cls_token": {
-    "content": "[CLS]",
+  "boi_token": "<start_of_image>",
+  "bos_token": {
+    "content": "<bos>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
-  "mask_token": {
-    "content": "[MASK]",
+  "eoi_token": "<end_of_image>",
+  "eos_token": {
+    "content": "<eos>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
+  "image_token": "<image_soft_token>",
   "pad_token": {
-    "content": "[PAD]",
-    "lstrip": false,
-    "normalized": false,
-    "rstrip": false,
-    "single_word": false
-  },
-  "sep_token": {
-    "content": "[SEP]",
+    "content": "<pad>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
   "unk_token": {
-    "content": "[UNK]",
+    "content": "<unk>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c4b7fb9f1bde49d5d0ac89d63b3b97256d730788bbc479278759fa6f4c5faa36
+oid sha256:eda78a9adeb8ee61251aa1b4dd9dd8932131463e584245b900135bee7256aae0
 size 6161