SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: sentence-transformers/all-MiniLM-L6-v2
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 384 dimensions
• Similarity Function: Cosine Similarity

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 256, '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()
)

Usage

Direct Usage (Sentence Transformers)

Training Details

Training Dataset

Unnamed Dataset

• Size: 19,604 training samples
• Columns: anchor, positive, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	positive	negative
  type	string	string	string
  details	min: 6 tokens mean: 9.57 tokens max: 21 tokens	min: 36 tokens mean: 83.77 tokens max: 256 tokens	min: 39 tokens mean: 86.33 tokens max: 256 tokens

• Samples:

  anchor	positive	negative
  python code to write bool value 1	def writeBoolean(self, n):[object Object] """[object Object] Writes a Boolean to the stream.[object Object] """[object Object] t = TYPE_BOOL_TRUE[object Object][object Object] if n is False:[object Object] t = TYPE_BOOL_FALSE[object Object][object Object] self.stream.write(t)	def cmd_dot(conf: Config):[object Object] """Print out a neat targets dependency tree based on requested targets.[object Object][object Object] Use graphviz to render the dot file, e.g.:[object Object][object Object] > ybt dot :foo :bar | dot -Tpng -o graph.png[object Object] """[object Object] build_context = BuildContext(conf)[object Object] populate_targets_graph(build_context, conf)[object Object] if conf.output_dot_file is None:[object Object] write_dot(build_context, conf, sys.stdout)[object Object] else:[object Object] with open(conf.output_dot_file, 'w') as out_file:[object Object] write_dot(build_context, conf, out_file)
  "python how to manipulate clipboard"	def paste(xsel=False):[object Object] """Returns system clipboard contents."""[object Object] selection = "primary" if xsel else "clipboard"[object Object] try:[object Object] return subprocess.Popen(["xclip", "-selection", selection, "-o"], stdout=subprocess.PIPE).communicate()[0].decode("utf-8")[object Object] except OSError as why:[object Object] raise XclipNotFound	def log_request(self, code='-', size='-'):[object Object] """Selectively log an accepted request."""[object Object][object Object] if self.server.logRequests:[object Object] BaseHTTPServer.BaseHTTPRequestHandler.log_request(self, code, size)
  python colored output to html	def _format_json(data, theme):[object Object] """Pretty print a dict as a JSON, with colors if pygments is present."""[object Object] output = json.dumps(data, indent=2, sort_keys=True)[object Object][object Object] if pygments and sys.stdout.isatty():[object Object] style = get_style_by_name(theme)[object Object] formatter = Terminal256Formatter(style=style)[object Object] return pygments.highlight(output, JsonLexer(), formatter)[object Object][object Object] return output	def average(arr):[object Object] """average of the values, must have more than 0 entries.[object Object][object Object] :param arr: list of numbers[object Object] :type arr: number[] a number array[object Object] :return: average[object Object] :rtype: float[object Object][object Object] """[object Object] if len(arr) == 0:[object Object] sys.stderr.write("ERROR: no content in array to take average\n")[object Object] sys.exit()[object Object] if len(arr) == 1: return arr[0][object Object] return float(sum(arr))/float(len(arr))

• Loss: MultipleNegativesRankingLoss with these parameters:
  json

  1{ 2 "scale": 20.0, 3 "similarity_fct": "cos_sim", 4 "gather_across_devices": false 5}

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 16
• learning_rate: 1e-05
• num_train_epochs: 2
• warmup_steps: 1000
• router_mapping: {'texts': {'0': 'anchor', '1': 'positive', '2': 'negative'}}

All Hyperparameters

Training Logs

Framework Versions

• Python: 3.12.3
• Sentence Transformers: 5.1.1
• Transformers: 4.57.1
• PyTorch: 2.9.0+cu128
• Accelerate: 1.10.1
• Datasets: 3.6.0
• Tokenizers: 0.22.1

Citation

BibTeX

Sentence Transformers

MultipleNegativesRankingLoss