mse=0.0234

README.md

@@ -4,35 +4,35 @@ tags:
 - sentence-similarity
 - feature-extraction
 - generated_from_trainer
-- dataset_size:1621
+- dataset_size:3072
 - loss:CosineSimilarityLoss
 base_model: sentence-transformers/all-mpnet-base-v2
 widget:
-- source_sentence: Calmness during production incidents
+- source_sentence: Build type system for programming language from scratch
   sentences:
-  - 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
+  - Uses TypeScript for type-safe JavaScript
+  - Led architecture decision meetings resulting in consensus
+  - Integrated Stripe, PayPal, and custom payment solutions
+- source_sentence: Privacy engineering skills
   sentences:
-  - Technical lead without budget responsibility
-  - Created SDKs for multiple programming languages
-  - Built real-time dashboards processing streaming data
-- source_sentence: Understanding of business context
+  - Implemented differential privacy
+  - Technical implementation without vendor management
+  - Created developer-friendly APIs with Swagger docs
+- source_sentence: Privacy Pass, privacy protocol
   sentences:
-  - Work-life balance advocate, balanced person, holistic
-  - Adds spring to team's step
-  - Business aware, context driven, strategic thinker
-- source_sentence: Self-motivated with minimal supervision needed
+  - Modern development tools only
+  - Excellent at breaking down complex topics for junior developers
+  - Privacy-preserving authentication methods
+- source_sentence: JVM tuning and profiling
   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
+  - Performance monitoring patterns
+  - Optimized GC settings reducing pause times
+  - Senior developer with proven track record debugging distributed system race conditions
+- source_sentence: Knowledge sharing enthusiasm
   sentences:
-  - Calming applications only
-  - Developed Selenium test suites covering 80% of critical user flows
-  - Designed event-driven systems using Kafka and AWS EventBridge
+  - Regular meetup speaker and blogger
+  - Optimized Spark jobs processing terabytes of data daily
+  - Configured database partitioning
 pipeline_tag: sentence-similarity
 library_name: sentence-transformers
 metrics:
@@ -49,10 +49,10 @@ model-index:
       type: val
     metrics:
     - type: pearson_cosine
-      value: 0.877106958407389
+      value: 0.8977247913414342
       name: Pearson Cosine
     - type: spearman_cosine
-      value: 0.8469811407862099
+      value: 0.8052388814564073
       name: Spearman Cosine
 ---
 
@@ -105,9 +105,9 @@ from sentence_transformers import SentenceTransformer
 model = SentenceTransformer("sentence_transformers_model_id")
 # Run inference
 sentences = [
-    '5+ years building anxiety platforms',
-    'Calming applications only',
-    'Developed Selenium test suites covering 80% of critical user flows',
+    'Knowledge sharing enthusiasm',
+    'Regular meetup speaker and blogger',
+    'Configured database partitioning',
 ]
 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      | 0.8771    |
-| **spearman_cosine** | **0.847** |
+| Metric              | Value      |
+|:--------------------|:-----------|
+| pearson_cosine      | 0.8977     |
+| **spearman_cosine** | **0.8052** |
 
 <!--
 ## Bias, Risks and Limitations
@@ -175,19 +175,19 @@ You can finetune this model on your own dataset.
 
 #### Unnamed Dataset
 
-* Size: 1,621 training samples
+* Size: 3,072 training samples
 * Columns: <code>sentence_0</code>, <code>sentence_1</code>, and <code>label</code>
 * Approximate statistics based on the first 1000 samples:
-  |         | sentence_0                                                                       | sentence_1                                                                       | label                                                          |
-  |:--------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------------------------|:---------------------------------------------------------------|
-  | type    | string                                                                           | string                                                                           | float                                                          |
-  | 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> |
+  |         | sentence_0                                                                       | sentence_1                                                                        | label                                                          |
+  |:--------|:---------------------------------------------------------------------------------|:----------------------------------------------------------------------------------|:---------------------------------------------------------------|
+  | type    | string                                                                           | string                                                                            | float                                                          |
+  | details | <ul><li>min: 4 tokens</li><li>mean: 9.74 tokens</li><li>max: 27 tokens</li></ul> | <ul><li>min: 5 tokens</li><li>mean: 11.06 tokens</li><li>max: 24 tokens</li></ul> | <ul><li>min: 0.0</li><li>mean: 0.67</li><li>max: 1.0</li></ul> |
 * Samples:
-  | 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> |
+  | sentence_0                                                                 | sentence_1                                                                 | label            |
+  |:---------------------------------------------------------------------------|:---------------------------------------------------------------------------|:-----------------|
+  | <code>Boundary-value testing and equivalence partitioning expertise</code> | <code>QA engineer designing test cases with boundary value analysis</code> | <code>0.9</code> |
+  | <code>Must have strong decision-making skills</code>                       | <code>Makes timely decisions based on available information</code>         | <code>0.7</code> |
+  | <code>8+ years building real-time collaboration tools</code>               | <code>Traditional request-response application development</code>          | <code>0.2</code> |
 * Loss: [<code>CosineSimilarityLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#cosinesimilarityloss) with these parameters:
   ```json
   {
@@ -326,20 +326,24 @@ You can finetune this model on your own dataset.
 ### Training Logs
 | Epoch  | Step | val_spearman_cosine |
 |:------:|:----:|:-------------------:|
-| 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              |
+| 0.5208 | 50   | 0.6737              |
+| 1.0    | 96   | 0.7384              |
+| 1.0417 | 100  | 0.7431              |
+| 1.5625 | 150  | 0.7703              |
+| 2.0    | 192  | 0.7790              |
+| 2.0833 | 200  | 0.7817              |
+| 2.6042 | 250  | 0.8011              |
+| 3.0    | 288  | 0.7967              |
+| 3.125  | 300  | 0.7963              |
+| 3.6458 | 350  | 0.8046              |
+| 4.0    | 384  | 0.8052              |
 
 
 ### Framework Versions
-- Python: 3.12.9
+- Python: 3.12.10
 - Sentence Transformers: 4.1.0
 - Transformers: 4.52.4
-- PyTorch: 2.7.1
+- PyTorch: 2.7.1+cu126
 - Accelerate: 1.7.0
 - Datasets: 3.6.0
 - Tokenizers: 0.21.1

config_sentence_transformers.json

@@ -2,7 +2,7 @@
   "__version__": {
     "sentence_transformers": "4.1.0",
     "transformers": "4.52.4",
-    "pytorch": "2.7.1"
+    "pytorch": "2.7.1+cu126"
   },
   "prompts": {},
   "default_prompt_name": null,

eval/similarity_evaluation_val_results.csv

@@ -1,5 +1,5 @@
 epoch,steps,cosine_pearson,cosine_spearman
-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
+1.0,96,0.8272483418053012,0.7384040919120075
+2.0,192,0.8806144722889805,0.7789630856263889
+3.0,288,0.8940053252264049,0.7967165513263559
+4.0,384,0.8977247913414342,0.8052388814564073

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1618565b9451f2af29c2284506ba4dabcc0133025fa60b4d4119c356d737eb24
+oid sha256:628af632e016d61b250d100cdf4a3b0b13f3c1b2802767ceea7fd31e83f3ebfa
 size 437967672