wraithklu/codesentinel-clone-detector

SentenceTransformer based on microsoft/codebert-base

This is a sentence-transformers model finetuned from microsoft/codebert-base. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: microsoft/codebert-base
• Maximum Sequence Length: 256 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Supported Modality: Text

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'RobertaModel'})
  (1): Pooling({'embedding_dimension': 768, 'pooling_mode': 'mean', 'include_prompt': True})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("wraithklu/codesentinel-clone-detector")
# Run inference
sentences = [
    'protected void serveStaticContent(HttpServletRequest request, HttpServletResponse response, String pathInfo) throws ServletException {\n        InputStream is = servletConfig.getServletContext().getResourceAsStream(pathInfo);\n        if (is == null) {\n            throw new ServletException("Static resource " + pathInfo + " is not available");\n        }\n        try {\n            int ind = pathInfo.lastIndexOf(".");\n            if (ind != -1 && ind < pathInfo.length()) {\n                String type = STATIC_CONTENT_TYPES.get(pathInfo.substring(ind + 1));\n                if (type != null) {\n                    response.setContentType(type);\n                }\n            }\n            ServletOutputStream os = response.getOutputStream();\n            IOUtils.copy(is, os);\n            os.flush();\n        } catch (IOException ex) {\n            throw new ServletException("Static resource " + pathInfo + " can not be written to the output stream");\n        }\n    }',
    'private static String makeLoginRequest(String authenticationUrl, String accountEmail, String accountPassword) throws IOException {\n        URL url = new URL(authenticationUrl);\n        HttpURLConnection urlConnection = (HttpURLConnection) url.openConnection();\n        urlConnection.setRequestMethod("POST");\n        urlConnection.setDoInput(true);\n        urlConnection.setDoOutput(true);\n        urlConnection.setUseCaches(false);\n        urlConnection.setRequestProperty("Content-Type", "application/x-www-form-urlencoded");\n        StringBuffer content = new StringBuffer();\n        content.append("Email=").append(URLEncoder.encode(accountEmail, "UTF-8"));\n        content.append("&Passwd=").append(URLEncoder.encode(accountPassword, "UTF-8"));\n        content.append("&source=").append(URLEncoder.encode("Google Base data API for OFBiz", "UTF-8"));\n        content.append("&service=").append(URLEncoder.encode("gbase", "UTF-8"));\n        OutputStream outputStream = urlConnection.getOutputStream();\n        outputStream.write(content.toString().getBytes("UTF-8"));\n        outputStream.close();\n        int responseCode = urlConnection.getResponseCode();\n        InputStream inputStream;\n        if (responseCode == HttpURLConnection.HTTP_OK) {\n            inputStream = urlConnection.getInputStream();\n        } else {\n            inputStream = urlConnection.getErrorStream();\n        }\n        return toString(inputStream);\n    }',
    'public static void main(String[] args) throws Exception {\n        if (args.length != 2) {\n            System.out.println("arguments: sourcefile destfile");\n            System.exit(1);\n        }\n        FileChannel in = new FileInputStream(args[0]).getChannel(), out = new FileOutputStream(args[1]).getChannel();\n        in.transferTo(0, in.size(), out);\n    }',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[ 1.0000, -0.0114,  0.9605],
#         [-0.0114,  1.0000,  0.0068],
#         [ 0.9605,  0.0068,  1.0000]])

Evaluation

Metrics

Semantic Similarity

• Dataset: bigclonebench-validation
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.905
spearman_cosine	0.8343

Training Details

Training Dataset

Unnamed Dataset

• Size: 10,000 training samples
• Columns: sentence_0, sentence_1, and label
• Approximate statistics based on the first 100 samples:

  	sentence_0	sentence_1	label
  type	string	string	float
  modality	text	text
  details	min: 134 tokens mean: 240.02 tokens max: 256 tokens	min: 104 tokens mean: 237.68 tokens max: 256 tokens	min: 0.0 mean: 0.49 max: 1.0

• Samples:

  sentence_0	sentence_1	label
  public InputLattice(String file, int type, boolean convertJoinMeet) throws FileNotFoundException, IOException { InputStream latStream = null; labels = new Vector(); upperCoversVector = new Vector(); Vector upperCovers = new Vector(); edgeColors = new Hashtable(); Vector upperCovering = new Vector(); int level = 0; String str; String currentElt = null; String cover = null; boolean first = true; StreamTokenizer in; switch(type) { case FILE: latStream = new FileInputStream(file); break; case STRING: latStream = new StringBufferInputStream(file); break; case URL: URL url = new URL("http", Config.HOST, Config.PORT, "/" + Config.HASSE + "/" + Config.URL_STREAMER + file); System.out.println("url is " + url); URLConnection connection = url.openCo...	private String calculateHash(String s) { if (s == null) { return null; } MessageDigest messageDigest; try { messageDigest = MessageDigest.getInstance("SHA"); } catch (NoSuchAlgorithmException e) { logger.error("Could not find a message digest algorithm."); return null; } messageDigest.update(s.getBytes()); byte[] hash = messageDigest.digest(); StringBuilder sb = new StringBuilder(); for (byte b : hash) { sb.append(String.format("%02x", b)); } return sb.toString(); }	0.0
  void load(URL url) throws IOException { BufferedReader r = new BufferedReader(new InputStreamReader(url.openStream())); Vector3f scale = new Vector3f(1, 1, 1); Group currentGroup = new Group(); currentGroup.name = "default"; groups.add(currentGroup); String line; while ((line = r.readLine()) != null) { String[] params = line.split(" +"); if (params.length == 0) continue; String command = params[0]; if (params[0].equals("v")) { Vector3f vertex = new Vector3f(Float.parseFloat(params[1]) * scale.x, Float.parseFloat(params[2]) * scale.y, Float.parseFloat(params[3]) * scale.z); verticies.add(vertex); radius = Math.max(radius, vertex.length()); } if (command.equals("center")) { epicenter = new Vector3f(Float.parseFloat(params[1]), Float.parseFloat(params[2]), Float.parseFloat(params[3])); } ...	public static boolean canWeConnectToInternet() { String s = "http://www.google.com/"; URL url = null; boolean can = false; URLConnection conection = null; try { url = new URL(s); } catch (MalformedURLException e) { System.out.println("This should never happend. Error in URL name. URL specified was:" + s + "."); } try { conection = url.openConnection(); conection.connect(); can = true; } catch (IOException e) { can = false; } if (can) { } return can; }	0.0
  public static java.io.ByteArrayOutputStream getFileByteStream(URL _url) { java.io.ByteArrayOutputStream buffer = new java.io.ByteArrayOutputStream(); try { InputStream input = _url.openStream(); IOUtils.copy(input, buffer); IOUtils.closeQuietly(input); } catch (Exception err) { throw new RuntimeException(err); } return buffer; }	public BufferedImage processUsingTemp(InputStream input, DjatokaDecodeParam params) throws DjatokaException { File in; try { in = File.createTempFile("tmp", ".jp2"); FileOutputStream fos = new FileOutputStream(in); in.deleteOnExit(); IOUtils.copyStream(input, fos); } catch (IOException e) { logger.error(e, e); throw new DjatokaException(e); } BufferedImage bi = process(in.getAbsolutePath(), params); if (in != null) in.delete(); return bi; }	1.0

• Loss: CosineSimilarityLoss with these parameters:

  {
      "loss_fct": "torch.nn.modules.loss.MSELoss",
      "cos_score_transformation": "torch.nn.modules.linear.Identity"
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• per_device_train_batch_size: 32
• num_train_epochs: 2
• fp16: True
• per_device_eval_batch_size: 32
• multi_dataset_batch_sampler: round_robin

All Hyperparameters

Click to expand

• per_device_train_batch_size: 32
• num_train_epochs: 2
• max_steps: -1
• learning_rate: 5e-05
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0
• optim: adamw_torch_fused
• optim_args: None
• weight_decay: 0.0
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• optim_target_modules: None
• gradient_accumulation_steps: 1
• average_tokens_across_devices: True
• max_grad_norm: 1
• label_smoothing_factor: 0.0
• bf16: False
• fp16: True
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• use_liger_kernel: False
• liger_kernel_config: None
• use_cache: False
• neftune_noise_alpha: None
• torch_empty_cache_steps: None
• auto_find_batch_size: False
• log_on_each_node: True
• logging_nan_inf_filter: True
• include_num_input_tokens_seen: no
• log_level: passive
• log_level_replica: warning
• disable_tqdm: False
• project: huggingface
• trackio_space_id: None
• trackio_bucket_id: None
• trackio_static_space_id: None
• per_device_eval_batch_size: 32
• prediction_loss_only: True
• eval_on_start: False
• eval_do_concat_batches: True
• eval_use_gather_object: False
• eval_accumulation_steps: None
• include_for_metrics: []
• batch_eval_metrics: False
• save_only_model: False
• save_on_each_node: False
• enable_jit_checkpoint: False
• push_to_hub: False
• hub_private_repo: None
• hub_model_id: None
• hub_strategy: every_save
• hub_always_push: False
• hub_revision: None
• load_best_model_at_end: False
• ignore_data_skip: False
• restore_callback_states_from_checkpoint: False
• full_determinism: False
• seed: 42
• data_seed: None
• use_cpu: False
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• parallelism_config: None
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• dataloader_prefetch_factor: None
• dataloader_multiprocessing_context: None
• dataloader_in_order: True
• remove_unused_columns: True
• label_names: None
• train_sampling_strategy: random
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• ddp_static_graph: None
• ddp_backend: None
• ddp_timeout: 1800
• fsdp: None
• fsdp_config: None
• deepspeed: None
• debug: []
• skip_memory_metrics: True
• do_predict: False
• resume_from_checkpoint: None
• local_rank: -1
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: round_robin
• router_mapping: {}
• learning_rate_mapping: {}
• warmup_ratio: None

Training Logs

Epoch	Step	Training Loss	bigclonebench-validation_spearman_cosine
0.4984	156	-	0.8142
0.9968	312	-	0.8325
1.0	313	-	0.8325
1.4952	468	-	0.8346
1.5974	500	0.0901	-
1.9936	624	-	0.8343
2.0	626	-	0.8343

Training Time

• Training: 13.3 minutes

Framework Versions

• Python: 3.13.16
• Sentence Transformers: 5.7.0
• Transformers: 5.22.1
• PyTorch: 2.14.1+cu130
• Accelerate: 1.15.0
• Datasets: 4.8.5
• Tokenizers: 0.23.3

Additional Resources

• Training and Finetuning Embedding Models with Sentence Transformers: the end-to-end guide for training or finetuning Sentence Transformer models.
• Introduction to Matryoshka Embedding Models: variable-size embeddings that can be truncated with minimal quality loss.
• Binary and Scalar Embedding Quantization for Significantly Faster & Cheaper Retrieval: post-training compression of embedding vectors.
• Multimodal Embedding & Reranker Models with Sentence Transformers: use text, image, audio, and video models through the same API.
• Training and Finetuning Multimodal Embedding & Reranker Models with Sentence Transformers: train multimodal embedding models, with a Visual Document Retrieval walkthrough.

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}