mse=0.0253

README.md

@@ -5,34 +5,34 @@ tags:
 - feature-extraction
 - generated_from_trainer
 - dataset_size:1621
-- loss:MultipleNegativesRankingLoss
+- loss:CosineSimilarityLoss
 base_model: sentence-transformers/all-mpnet-base-v2
 widget:
-- source_sentence: Liveblocks, real-time collaboration infrastructure
+- source_sentence: Calmness during production incidents
   sentences:
-  - Serverless routing patterns
-  - Socket.io for basic real-time features
-  - Neutral platform development only
-- source_sentence: Positive attitude and team spirit
+  - Takes feedback well, improves based on input, thanks reviewers
+  - Level-headed, clear thinking under stress, calming presence
+  - Implemented OAuth2/OIDC authentication for enterprise SSO
+- source_sentence: Must have SDK development experience
   sentences:
-  - 6 years Android development, Java and Kotlin, Google Play publications
-  - Maintains team morale during challenging projects
-  - Lucky platforms only
-- source_sentence: Experience with .NET Core and C# development required
+  - Technical lead without budget responsibility
+  - Created SDKs for multiple programming languages
+  - Built real-time dashboards processing streaming data
+- source_sentence: Understanding of business context
   sentences:
-  - Organized team building activities and fostered inclusive environment
-  - iptables, firewall rule management
-  - 10 years C# development with .NET Framework and .NET Core 3.1+
-- source_sentence: Onion Routing, Tor support
-  sentences:
-  - Privacy-focused architecture design
-  - Led global teams across 6 countries effectively
+  - Work-life balance advocate, balanced person, holistic
+  - Adds spring to team's step
   - Business aware, context driven, strategic thinker
-- source_sentence: Must have expertise in Angular and TypeScript
+- source_sentence: Self-motivated with minimal supervision needed
+  sentences:
+  - Highly autonomous, self-directed learner, owns project outcomes
+  - Managed multi-datacenter Cassandra clusters
+  - Complex redirect logic implementation
+- source_sentence: 5+ years building anxiety platforms
   sentences:
-  - React developer with JavaScript ES6+ experience
-  - Mobile app developer with no AR/VR experience
-  - Owns errors, learns from mistakes, transparent
+  - Calming applications only
+  - Developed Selenium test suites covering 80% of critical user flows
+  - Designed event-driven systems using Kafka and AWS EventBridge
 pipeline_tag: sentence-similarity
 library_name: sentence-transformers
 metrics:
@@ -49,10 +49,10 @@ model-index:
       type: val
     metrics:
     - type: pearson_cosine
-      value: 0.33261488496356484
+      value: 0.877106958407389
       name: Pearson Cosine
     - type: spearman_cosine
-      value: 0.3462323228018911
+      value: 0.8469811407862099
       name: Spearman Cosine
 ---
 
@@ -105,9 +105,9 @@ from sentence_transformers import SentenceTransformer
 model = SentenceTransformer("sentence_transformers_model_id")
 # Run inference
 sentences = [
-    'Must have expertise in Angular and TypeScript',
-    'React developer with JavaScript ES6+ experience',
-    'Mobile app developer with no AR/VR experience',
+    '5+ years building anxiety platforms',
+    'Calming applications only',
+    'Developed Selenium test suites covering 80% of critical user flows',
 ]
 embeddings = model.encode(sentences)
 print(embeddings.shape)
@@ -152,10 +152,10 @@ You can finetune this model on your own dataset.
 * Dataset: `val`
 * Evaluated with [<code>EmbeddingSimilarityEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.EmbeddingSimilarityEvaluator)
 
-| Metric              | Value      |
-|:--------------------|:-----------|
-| pearson_cosine      | 0.3326     |
-| **spearman_cosine** | **0.3462** |
+| Metric              | Value     |
+|:--------------------|:----------|
+| pearson_cosine      | 0.8771    |
+| **spearman_cosine** | **0.847** |
 
 <!--
 ## Bias, Risks and Limitations
@@ -181,18 +181,17 @@ You can finetune this model on your own dataset.
   |         | sentence_0                                                                       | sentence_1                                                                       | label                                                          |
   |:--------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------|
   | type    | string                                                                           | string                                                                           | float                                                          |
-  | details | <ul><li>min: 4 tokens</li><li>mean: 8.46 tokens</li><li>max: 21 tokens</li></ul> | <ul><li>min: 5 tokens</li><li>mean: 9.85 tokens</li><li>max: 24 tokens</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.59</li><li>max: 1.0</li></ul> |
+  | details | <ul><li>min: 4 tokens</li><li>mean: 8.35 tokens</li><li>max: 21 tokens</li></ul> | <ul><li>min: 5 tokens</li><li>mean: 9.74 tokens</li><li>max: 24 tokens</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.59</li><li>max: 1.0</li></ul> |
 * Samples:
-  | sentence_0                                          | sentence_1                                                           | label            |
-  |:----------------------------------------------------|:---------------------------------------------------------------------|:-----------------|
-  | <code>Authenticity in team relationships</code>     | <code>Genuine connections, real person, authentic leader</code>      | <code>0.9</code> |
-  | <code>Keyless SSL, private key security</code>      | <code>HSM integration, key management</code>                         | <code>0.4</code> |
-  | <code>Need expertise in database replication</code> | <code>Set up master-slave replication with automatic failover</code> | <code>0.9</code> |
-* Loss: [<code>MultipleNegativesRankingLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#multiplenegativesrankingloss) with these parameters:
+  | sentence_0                                             | sentence_1                                                            | label            |
+  |:-------------------------------------------------------|:----------------------------------------------------------------------|:-----------------|
+  | <code>Proactiveness in identifying improvements</code> | <code>Spots issues early, suggests solutions, takes initiative</code> | <code>0.9</code> |
+  | <code>Layout Worklet, custom layout</code>             | <code>Layout worklet implementation patterns</code>                   | <code>0.2</code> |
+  | <code>Must have SDK development experience</code>      | <code>Created SDKs for multiple programming languages</code>          | <code>0.9</code> |
+* Loss: [<code>CosineSimilarityLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosinesimilarityloss) with these parameters:
   ```json
   {
-      "scale": 20.0,
-      "similarity_fct": "cos_sim"
+      "loss_fct": "torch.nn.modules.loss.MSELoss"
   }
   ```
 
@@ -202,7 +201,7 @@ You can finetune this model on your own dataset.
 - `eval_strategy`: steps
 - `per_device_train_batch_size`: 32
 - `per_device_eval_batch_size`: 32
-- `num_train_epochs`: 5
+- `num_train_epochs`: 4
 - `multi_dataset_batch_sampler`: round_robin
 
 #### All Hyperparameters
@@ -225,7 +224,7 @@ You can finetune this model on your own dataset.
 - `adam_beta2`: 0.999
 - `adam_epsilon`: 1e-08
 - `max_grad_norm`: 1
-- `num_train_epochs`: 5
+- `num_train_epochs`: 4
 - `max_steps`: -1
 - `lr_scheduler_type`: linear
 - `lr_scheduler_kwargs`: {}
@@ -270,7 +269,6 @@ You can finetune this model on your own dataset.
 - `fsdp`: []
 - `fsdp_min_num_params`: 0
 - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
-- `tp_size`: 0
 - `fsdp_transformer_layer_cls_to_wrap`: None
 - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
 - `deepspeed`: None
@@ -328,14 +326,20 @@ You can finetune this model on your own dataset.
 ### Training Logs
 | Epoch  | Step | val_spearman_cosine |
 |:------:|:----:|:-------------------:|
-| 0.9804 | 50   | 0.3462              |
+| 0.9804 | 50   | 0.7715              |
+| 1.0    | 51   | 0.7742              |
+| 1.9608 | 100  | 0.8218              |
+| 2.0    | 102  | 0.8218              |
+| 2.9412 | 150  | 0.8415              |
+| 3.0    | 153  | 0.8423              |
+| 3.9216 | 200  | 0.8470              |
 
 
 ### Framework Versions
 - Python: 3.12.9
 - Sentence Transformers: 4.1.0
-- Transformers: 4.51.3
-- PyTorch: 2.7.0
+- Transformers: 4.52.4
+- PyTorch: 2.7.1
 - Accelerate: 1.7.0
 - Datasets: 3.6.0
 - Tokenizers: 0.21.1
@@ -357,18 +361,6 @@ You can finetune this model on your own dataset.
 }
 ```
 
-#### MultipleNegativesRankingLoss
-```bibtex
-@misc{henderson2017efficient,
-    title={Efficient Natural Language Response Suggestion for Smart Reply},
-    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
-    year={2017},
-    eprint={1705.00652},
-    archivePrefix={arXiv},
-    primaryClass={cs.CL}
-}
-```
-
 <!--
 ## Glossary
 

config.json

@@ -18,6 +18,6 @@
   "pad_token_id": 1,
   "relative_attention_num_buckets": 32,
   "torch_dtype": "float32",
-  "transformers_version": "4.51.3",
+  "transformers_version": "4.52.4",
   "vocab_size": 30527
 }

config_sentence_transformers.json

@@ -1,8 +1,8 @@
 {
   "__version__": {
     "sentence_transformers": "4.1.0",
-    "transformers": "4.51.3",
-    "pytorch": "2.7.0"
+    "transformers": "4.52.4",
+    "pytorch": "2.7.1"
   },
   "prompts": {},
   "default_prompt_name": null,

eval/similarity_evaluation_val_results.csv

@@ -1,6 +1,5 @@
 epoch,steps,cosine_pearson,cosine_spearman
-1.0,51,0.333061348383918,0.34606382932875346
-2.0,102,0.2896842112210425,0.29871199430927403
-3.0,153,0.31861828044212254,0.32684568868246433
-4.0,204,0.298435297570077,0.3068966237124457
-5.0,255,0.28717771168468886,0.2960869240364453
+1.0,51,0.8154555424408279,0.7741979456271402
+2.0,102,0.8586989969751344,0.8217682417751387
+3.0,153,0.8744392671984902,0.842272664134552
+4.0,204,0.877106958407389,0.8469811407862099

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f8ee34bf80e7a842dc955d3be4f15bac3990a4f92341572bfbf67713c2903c61
+oid sha256:1618565b9451f2af29c2284506ba4dabcc0133025fa60b4d4119c356d737eb24
 size 437967672