mec-is commited on
Commit
fbbe070
·
verified ·
1 Parent(s): a5ecfc8

Add domain-adapted CVE embedding model

Browse files
cve/domain_adapted_model/1_Pooling/config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 384,
3
+ "pooling_mode_cls_token": false,
4
+ "pooling_mode_mean_tokens": true,
5
+ "pooling_mode_max_tokens": false,
6
+ "pooling_mode_mean_sqrt_len_tokens": false,
7
+ "pooling_mode_weightedmean_tokens": false,
8
+ "pooling_mode_lasttoken": false,
9
+ "include_prompt": true
10
+ }
cve/domain_adapted_model/README.md ADDED
@@ -0,0 +1,463 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - sentence-transformers
4
+ - sentence-similarity
5
+ - feature-extraction
6
+ - dense
7
+ - generated_from_trainer
8
+ - dataset_size:70262
9
+ - loss:DenoisingAutoEncoderLoss
10
+ base_model: sentence-transformers/all-MiniLM-L6-v2
11
+ widget:
12
+ - source_sentence: 'no'
13
+ sentences:
14
+ - '(no title)
15
+
16
+ CVE-2015-2087 | Unrestricted file upload vulnerability in the Avatar Uploader
17
+ module before 6.x-1.3 for Drupal allows remote authenticated users to execute
18
+ arbitrary PHP code by uploading a file with a PHP extension, then accessing it
19
+ via unspecified vectors. | n/a n/a'
20
+ - "McAfee Agent (MA) for Linux Privilege Escalation vulnerability \nCVE-2018-6705\
21
+ \ | McAfee Agent (MA) for Linux Privilege Escalation vulnerability | Privilege\
22
+ \ escalation vulnerability in McAfee Agent (MA) for Linux 5.0.0 through 5.0.6,\
23
+ \ 5.5.0, and 5.5.1 allows local users to perform arbitrary command execution via\
24
+ \ specific conditions. | CWE-377 | McAfee McAfee Agent (MA) for Linux"
25
+ - '(no title)
26
+
27
+ CVE-2016-8128'
28
+ - source_sentence: ') CVE-2017-10347 Vulnerability Java SE component Java SE (subcomponent
29
+ that are Java SE 6u161 7u151 and Java Embedded 8u144 via, JRockit Successful of
30
+ this unauthorized service of SE Note This to, sandboxed or sandboxed Java applets
31
+ that load and run, code the Java for Java typically servers load and trusted code
32
+ e.g., installed by an administrator Base Score (impacts .::: L/PR:: U/C) | Corporation'
33
+ sentences:
34
+ - '(no title)
35
+
36
+ CVE-2017-5678'
37
+ - '(no title)
38
+
39
+ CVE-2018-15126 | LibVNC before commit 73cb96fec028a576a5a24417b57723b55854ad7b
40
+ contains heap use-after-free vulnerability in server code of file transfer extension
41
+ that can result remote code execution | n/a LibVNC'
42
+ - '(no title)
43
+
44
+ CVE-2017-10347 | Vulnerability in the Java SE, JRockit component of Oracle Java
45
+ SE (subcomponent: Serialization). Supported versions that are affected are Java
46
+ SE: 6u161, 7u151, 8u144 and 9; Java SE Embedded: 8u144. Easily exploitable vulnerability
47
+ allows unauthenticated attacker with network access via multiple protocols to
48
+ compromise Java SE, JRockit. Successful attacks of this vulnerability can result
49
+ in unauthorized ability to cause a partial denial of service (partial DOS) of
50
+ Java SE, JRockit. Note: This vulnerability applies to Java deployments, typically
51
+ in clients running sandboxed Java Web Start applications or sandboxed Java applets,
52
+ that load and run untrusted code (e.g., code that comes from the internet) and
53
+ rely on the Java sandbox for security. This vulnerability does not apply to Java
54
+ deployments, typically in servers, that load and run only trusted code (e.g.,
55
+ code installed by an administrator). CVSS 3.0 Base Score 5.3 (Availability impacts).
56
+ CVSS Vector: (CVSS:3.0/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L). | Oracle Corporation
57
+ Java'
58
+ - source_sentence: title) discovered in iOS before . Safari is before on . before
59
+ on Windows is tvOS 11 affected . The "" remote attackers execute or cause a site
60
+ | n/a
61
+ sentences:
62
+ - '(no title)
63
+
64
+ CVE-2017-7104 | An issue was discovered in certain Apple products. iOS before
65
+ 11 is affected. Safari before 11 is affected. iCloud before 7.0 on Windows is
66
+ affected. iTunes before 12.7 on Windows is affected. tvOS before 11 is affected.
67
+ The issue involves the "WebKit" component. It allows remote attackers to execute
68
+ arbitrary code or cause a denial of service (memory corruption and application
69
+ crash) via a crafted web site. | n/a n/a'
70
+ - '(no title)
71
+
72
+ CVE-2014-3572 | The ssl3_get_key_exchange function in s3_clnt.c in OpenSSL before
73
+ 0.9.8zd, 1.0.0 before 1.0.0p, and 1.0.1 before 1.0.1k allows remote SSL servers
74
+ to conduct ECDHE-to-ECDH downgrade attacks and trigger a loss of forward secrecy
75
+ by omitting the ServerKeyExchange message. | n/a n/a'
76
+ - '(no title)
77
+
78
+ CVE-2017-8067 | drivers/char/virtio_console.c in the Linux kernel 4.9.x and 4.10.x
79
+ before 4.10.12 interacts incorrectly with the CONFIG_VMAP_STACK option, which
80
+ allows local users to cause a denial of service (system crash or memory corruption)
81
+ or possibly have unspecified other impact by leveraging use of more than one virtual
82
+ page for a DMA scatterlist. | n/a n/a'
83
+ - source_sentence: no) CVE-2016-2346 | Allround Developer on unverified data, allows
84
+ man-in-the-middle arbitrary code by fields in data | n/a
85
+ sentences:
86
+ - '(no title)
87
+
88
+ CVE-2018-7125 | A remote code execution vulnerability was identified in HPE Intelligent
89
+ Management Center (IMC) PLAT earlier than version 7.3 E0506P09. | n/a HPE Intelligent
90
+ Management Center (IMC) PLAT'
91
+ - '(no title)
92
+
93
+ CVE-2017-14246 | An out of bounds read in the function d2ulaw_array() in ulaw.c
94
+ of libsndfile 1.0.28 may lead to a remote DoS attack or information disclosure,
95
+ related to mishandling of the NAN and INFINITY floating-point values. | n/a n/a'
96
+ - '(no title)
97
+
98
+ CVE-2016-2346 | Allround Automations PL/SQL Developer 11 before 11.0.6 relies
99
+ on unverified HTTP data for updates, which allows man-in-the-middle attackers
100
+ to execute arbitrary code by modifying fields in the client-server data stream.
101
+ | n/a n/a'
102
+ - source_sentence: no title CVE-2018-1091 | In flush_tmregs_to_thread arch/powerpc/kernel/ptrace.c
103
+ in kernel before 4.13.5 guest kernel unprivileged during on a processor feature
104
+ and an erroneous of TM) instructions core path leading to a | Linux kernel
105
+ sentences:
106
+ - '(no title)
107
+
108
+ CVE-2018-1091 | In the flush_tmregs_to_thread function in arch/powerpc/kernel/ptrace.c
109
+ in the Linux kernel before 4.13.5, a guest kernel crash can be triggered from
110
+ unprivileged userspace during a core dump on a POWER host due to a missing processor
111
+ feature check and an erroneous use of transactional memory (TM) instructions in
112
+ the core dump path, leading to a denial of service. | CWE-391 | n/a Linux kernel
113
+ v4.13 and newer'
114
+ - '(no title)
115
+
116
+ CVE-2015-9227 | PHP remote file inclusion vulnerability in the get_file function
117
+ in upload/admin2/controller/report_logs.php in AlegroCart 1.2.8 allows remote
118
+ administrators to execute arbitrary PHP code via a URL in the file_path parameter
119
+ to upload/admin2. | n/a n/a'
120
+ - '(no title)
121
+
122
+ CVE-2014-0623 | Cross-site scripting (XSS) vulnerability in the Self-Service Console
123
+ in EMC RSA Authentication Manager 7.1 before SP4 P32 allows remote attackers to
124
+ inject arbitrary web script or HTML via unspecified vectors, related to a "cross
125
+ frame scripting" issue. | n/a n/a'
126
+ pipeline_tag: sentence-similarity
127
+ library_name: sentence-transformers
128
+ ---
129
+
130
+ # SentenceTransformer based on sentence-transformers/all-MiniLM-L6-v2
131
+
132
+ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [sentence-transformers/all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2). 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.
133
+
134
+ ## Model Details
135
+
136
+ ### Model Description
137
+ - **Model Type:** Sentence Transformer
138
+ - **Base model:** [sentence-transformers/all-MiniLM-L6-v2](https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2) <!-- at revision c9745ed1d9f207416be6d2e6f8de32d1f16199bf -->
139
+ - **Maximum Sequence Length:** 256 tokens
140
+ - **Output Dimensionality:** 384 dimensions
141
+ - **Similarity Function:** Cosine Similarity
142
+ <!-- - **Training Dataset:** Unknown -->
143
+ <!-- - **Language:** Unknown -->
144
+ <!-- - **License:** Unknown -->
145
+
146
+ ### Model Sources
147
+
148
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
149
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
150
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
151
+
152
+ ### Full Model Architecture
153
+
154
+ ```
155
+ SentenceTransformer(
156
+ (0): Transformer({'max_seq_length': 256, 'do_lower_case': False, 'architecture': 'BertModel'})
157
+ (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})
158
+ (2): Normalize()
159
+ )
160
+ ```
161
+
162
+ ## Usage
163
+
164
+ ### Direct Usage (Sentence Transformers)
165
+
166
+ First install the Sentence Transformers library:
167
+
168
+ ```bash
169
+ pip install -U sentence-transformers
170
+ ```
171
+
172
+ Then you can load this model and run inference.
173
+ ```python
174
+ from sentence_transformers import SentenceTransformer
175
+
176
+ # Download from the 🤗 Hub
177
+ model = SentenceTransformer("sentence_transformers_model_id")
178
+ # Run inference
179
+ sentences = [
180
+ 'no title CVE-2018-1091 | In flush_tmregs_to_thread arch/powerpc/kernel/ptrace.c in kernel before 4.13.5 guest kernel unprivileged during on a processor feature and an erroneous of TM) instructions core path leading to a | Linux kernel',
181
+ '(no title)\nCVE-2018-1091 | In the flush_tmregs_to_thread function in arch/powerpc/kernel/ptrace.c in the Linux kernel before 4.13.5, a guest kernel crash can be triggered from unprivileged userspace during a core dump on a POWER host due to a missing processor feature check and an erroneous use of transactional memory (TM) instructions in the core dump path, leading to a denial of service. | CWE-391 | n/a Linux kernel v4.13 and newer',
182
+ '(no title)\nCVE-2014-0623 | Cross-site scripting (XSS) vulnerability in the Self-Service Console in EMC RSA Authentication Manager 7.1 before SP4 P32 allows remote attackers to inject arbitrary web script or HTML via unspecified vectors, related to a "cross frame scripting" issue. | n/a n/a',
183
+ ]
184
+ embeddings = model.encode(sentences)
185
+ print(embeddings.shape)
186
+ # [3, 384]
187
+
188
+ # Get the similarity scores for the embeddings
189
+ similarities = model.similarity(embeddings, embeddings)
190
+ print(similarities)
191
+ # tensor([[1.0000, 0.9425, 0.7739],
192
+ # [0.9425, 1.0000, 0.8393],
193
+ # [0.7739, 0.8393, 1.0000]])
194
+ ```
195
+
196
+ <!--
197
+ ### Direct Usage (Transformers)
198
+
199
+ <details><summary>Click to see the direct usage in Transformers</summary>
200
+
201
+ </details>
202
+ -->
203
+
204
+ <!--
205
+ ### Downstream Usage (Sentence Transformers)
206
+
207
+ You can finetune this model on your own dataset.
208
+
209
+ <details><summary>Click to expand</summary>
210
+
211
+ </details>
212
+ -->
213
+
214
+ <!--
215
+ ### Out-of-Scope Use
216
+
217
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
218
+ -->
219
+
220
+ <!--
221
+ ## Bias, Risks and Limitations
222
+
223
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
224
+ -->
225
+
226
+ <!--
227
+ ### Recommendations
228
+
229
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
230
+ -->
231
+
232
+ ## Training Details
233
+
234
+ ### Training Dataset
235
+
236
+ #### Unnamed Dataset
237
+
238
+ * Size: 70,262 training samples
239
+ * Columns: <code>sentence_0</code> and <code>sentence_1</code>
240
+ * Approximate statistics based on the first 1000 samples:
241
+ | | sentence_0 | sentence_1 |
242
+ |:--------|:-----------------------------------------------------------------------------------|:------------------------------------------------------------------------------------|
243
+ | type | string | string |
244
+ | details | <ul><li>min: 3 tokens</li><li>mean: 37.14 tokens</li><li>max: 256 tokens</li></ul> | <ul><li>min: 12 tokens</li><li>mean: 87.48 tokens</li><li>max: 256 tokens</li></ul> |
245
+ * Samples:
246
+ | sentence_0 | sentence_1 |
247
+ |:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
248
+ | <code>title CVE-2016-0950 Connect allows remote attackers the interface unspecified n/a n/a</code> | <code>(no title)<br>CVE-2016-0950 \| Adobe Connect before 9.5.2 allows remote attackers to spoof the user interface via unspecified vectors. \| n/a n/a</code> |
249
+ | <code>CVE-2018-0769 Edge 10 1511 1607,, and Server an arbitrary in context of the current the scripting memory, aka "Memory This ID from CVE-2018-0758 CVE-2018-0762 CVE-2018-0768 CVE-2018-0773, CVE-2018-0776,, CVE-2018-0778 Microsoft</code> | <code>(no title)<br>CVE-2018-0769 \| Microsoft Edge in Windows 10 Gold, 1511, 1607, 1703, 1709, and Windows Server 2016 allows an attacker to execute arbitrary code in the context of the current user, due to how the scripting engine handles objects in memory, aka "Scripting Engine Memory Corruption Vulnerability". This CVE ID is unique from CVE-2018-0758, CVE-2018-0762, CVE-2018-0768, CVE-2018-0770, CVE-2018-0772, CVE-2018-0773, CVE-2018-0774, CVE-2018-0775, CVE-2018-0776, CVE-2018-0777, CVE-2018-0778, and CVE-2018-0781. \| Microsoft Corporation Microsoft Edge</code> |
250
+ | <code>(CVE-2018-11306</code> | <code>(no title)<br>CVE-2018-11306</code> |
251
+ * Loss: [<code>DenoisingAutoEncoderLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#denoisingautoencoderloss)
252
+
253
+ ### Training Hyperparameters
254
+ #### Non-Default Hyperparameters
255
+
256
+ - `num_train_epochs`: 1
257
+ - `multi_dataset_batch_sampler`: round_robin
258
+
259
+ #### All Hyperparameters
260
+ <details><summary>Click to expand</summary>
261
+
262
+ - `overwrite_output_dir`: False
263
+ - `do_predict`: False
264
+ - `eval_strategy`: no
265
+ - `prediction_loss_only`: True
266
+ - `per_device_train_batch_size`: 8
267
+ - `per_device_eval_batch_size`: 8
268
+ - `per_gpu_train_batch_size`: None
269
+ - `per_gpu_eval_batch_size`: None
270
+ - `gradient_accumulation_steps`: 1
271
+ - `eval_accumulation_steps`: None
272
+ - `torch_empty_cache_steps`: None
273
+ - `learning_rate`: 5e-05
274
+ - `weight_decay`: 0.0
275
+ - `adam_beta1`: 0.9
276
+ - `adam_beta2`: 0.999
277
+ - `adam_epsilon`: 1e-08
278
+ - `max_grad_norm`: 1
279
+ - `num_train_epochs`: 1
280
+ - `max_steps`: -1
281
+ - `lr_scheduler_type`: linear
282
+ - `lr_scheduler_kwargs`: None
283
+ - `warmup_ratio`: 0.0
284
+ - `warmup_steps`: 0
285
+ - `log_level`: passive
286
+ - `log_level_replica`: warning
287
+ - `log_on_each_node`: True
288
+ - `logging_nan_inf_filter`: True
289
+ - `save_safetensors`: True
290
+ - `save_on_each_node`: False
291
+ - `save_only_model`: False
292
+ - `restore_callback_states_from_checkpoint`: False
293
+ - `no_cuda`: False
294
+ - `use_cpu`: False
295
+ - `use_mps_device`: False
296
+ - `seed`: 42
297
+ - `data_seed`: None
298
+ - `jit_mode_eval`: False
299
+ - `bf16`: False
300
+ - `fp16`: False
301
+ - `fp16_opt_level`: O1
302
+ - `half_precision_backend`: auto
303
+ - `bf16_full_eval`: False
304
+ - `fp16_full_eval`: False
305
+ - `tf32`: None
306
+ - `local_rank`: 0
307
+ - `ddp_backend`: None
308
+ - `tpu_num_cores`: None
309
+ - `tpu_metrics_debug`: False
310
+ - `debug`: []
311
+ - `dataloader_drop_last`: False
312
+ - `dataloader_num_workers`: 0
313
+ - `dataloader_prefetch_factor`: None
314
+ - `past_index`: -1
315
+ - `disable_tqdm`: False
316
+ - `remove_unused_columns`: True
317
+ - `label_names`: None
318
+ - `load_best_model_at_end`: False
319
+ - `ignore_data_skip`: False
320
+ - `fsdp`: []
321
+ - `fsdp_min_num_params`: 0
322
+ - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
323
+ - `fsdp_transformer_layer_cls_to_wrap`: None
324
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
325
+ - `parallelism_config`: None
326
+ - `deepspeed`: None
327
+ - `label_smoothing_factor`: 0.0
328
+ - `optim`: adamw_torch_fused
329
+ - `optim_args`: None
330
+ - `adafactor`: False
331
+ - `group_by_length`: False
332
+ - `length_column_name`: length
333
+ - `project`: huggingface
334
+ - `trackio_space_id`: trackio
335
+ - `ddp_find_unused_parameters`: None
336
+ - `ddp_bucket_cap_mb`: None
337
+ - `ddp_broadcast_buffers`: False
338
+ - `dataloader_pin_memory`: True
339
+ - `dataloader_persistent_workers`: False
340
+ - `skip_memory_metrics`: True
341
+ - `use_legacy_prediction_loop`: False
342
+ - `push_to_hub`: False
343
+ - `resume_from_checkpoint`: None
344
+ - `hub_model_id`: None
345
+ - `hub_strategy`: every_save
346
+ - `hub_private_repo`: None
347
+ - `hub_always_push`: False
348
+ - `hub_revision`: None
349
+ - `gradient_checkpointing`: False
350
+ - `gradient_checkpointing_kwargs`: None
351
+ - `include_inputs_for_metrics`: False
352
+ - `include_for_metrics`: []
353
+ - `eval_do_concat_batches`: True
354
+ - `fp16_backend`: auto
355
+ - `push_to_hub_model_id`: None
356
+ - `push_to_hub_organization`: None
357
+ - `mp_parameters`:
358
+ - `auto_find_batch_size`: False
359
+ - `full_determinism`: False
360
+ - `torchdynamo`: None
361
+ - `ray_scope`: last
362
+ - `ddp_timeout`: 1800
363
+ - `torch_compile`: False
364
+ - `torch_compile_backend`: None
365
+ - `torch_compile_mode`: None
366
+ - `include_tokens_per_second`: False
367
+ - `include_num_input_tokens_seen`: no
368
+ - `neftune_noise_alpha`: None
369
+ - `optim_target_modules`: None
370
+ - `batch_eval_metrics`: False
371
+ - `eval_on_start`: False
372
+ - `use_liger_kernel`: False
373
+ - `liger_kernel_config`: None
374
+ - `eval_use_gather_object`: False
375
+ - `average_tokens_across_devices`: True
376
+ - `prompts`: None
377
+ - `batch_sampler`: batch_sampler
378
+ - `multi_dataset_batch_sampler`: round_robin
379
+ - `router_mapping`: {}
380
+ - `learning_rate_mapping`: {}
381
+
382
+ </details>
383
+
384
+ ### Training Logs
385
+ | Epoch | Step | Training Loss |
386
+ |:------:|:----:|:-------------:|
387
+ | 0.0569 | 500 | 10.0337 |
388
+ | 0.1139 | 1000 | 6.2851 |
389
+ | 0.1708 | 1500 | 5.1002 |
390
+ | 0.2277 | 2000 | 4.6124 |
391
+ | 0.2846 | 2500 | 4.2173 |
392
+ | 0.3416 | 3000 | 3.9048 |
393
+ | 0.3985 | 3500 | 3.6551 |
394
+ | 0.4554 | 4000 | 3.4511 |
395
+ | 0.5124 | 4500 | 3.314 |
396
+ | 0.5693 | 5000 | 3.1761 |
397
+ | 0.6262 | 5500 | 3.0345 |
398
+ | 0.6831 | 6000 | 2.9177 |
399
+ | 0.7401 | 6500 | 2.8082 |
400
+ | 0.7970 | 7000 | 2.7573 |
401
+ | 0.8539 | 7500 | 2.6312 |
402
+ | 0.9109 | 8000 | 2.615 |
403
+ | 0.9678 | 8500 | 2.5313 |
404
+
405
+
406
+ ### Framework Versions
407
+ - Python: 3.13.12
408
+ - Sentence Transformers: 5.1.1
409
+ - Transformers: 4.57.6
410
+ - PyTorch: 2.10.0
411
+ - Accelerate: 1.12.0
412
+ - Datasets: 4.5.0
413
+ - Tokenizers: 0.22.2
414
+
415
+ ## Citation
416
+
417
+ ### BibTeX
418
+
419
+ #### Sentence Transformers
420
+ ```bibtex
421
+ @inproceedings{reimers-2019-sentence-bert,
422
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
423
+ author = "Reimers, Nils and Gurevych, Iryna",
424
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
425
+ month = "11",
426
+ year = "2019",
427
+ publisher = "Association for Computational Linguistics",
428
+ url = "https://arxiv.org/abs/1908.10084",
429
+ }
430
+ ```
431
+
432
+ #### DenoisingAutoEncoderLoss
433
+ ```bibtex
434
+ @inproceedings{wang-2021-TSDAE,
435
+ title = "TSDAE: Using Transformer-based Sequential Denoising Auto-Encoderfor Unsupervised Sentence Embedding Learning",
436
+ author = "Wang, Kexin and Reimers, Nils and Gurevych, Iryna",
437
+ booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021",
438
+ month = nov,
439
+ year = "2021",
440
+ address = "Punta Cana, Dominican Republic",
441
+ publisher = "Association for Computational Linguistics",
442
+ pages = "671--688",
443
+ url = "https://arxiv.org/abs/2104.06979",
444
+ }
445
+ ```
446
+
447
+ <!--
448
+ ## Glossary
449
+
450
+ *Clearly define terms in order to be accessible across audiences.*
451
+ -->
452
+
453
+ <!--
454
+ ## Model Card Authors
455
+
456
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
457
+ -->
458
+
459
+ <!--
460
+ ## Model Card Contact
461
+
462
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
463
+ -->
cve/domain_adapted_model/config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "BertModel"
4
+ ],
5
+ "attention_probs_dropout_prob": 0.1,
6
+ "classifier_dropout": null,
7
+ "dtype": "float32",
8
+ "gradient_checkpointing": false,
9
+ "hidden_act": "gelu",
10
+ "hidden_dropout_prob": 0.1,
11
+ "hidden_size": 384,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 1536,
14
+ "layer_norm_eps": 1e-12,
15
+ "max_position_embeddings": 512,
16
+ "model_type": "bert",
17
+ "num_attention_heads": 12,
18
+ "num_hidden_layers": 6,
19
+ "pad_token_id": 0,
20
+ "position_embedding_type": "absolute",
21
+ "transformers_version": "4.57.6",
22
+ "type_vocab_size": 2,
23
+ "use_cache": true,
24
+ "vocab_size": 30522
25
+ }
cve/domain_adapted_model/config_sentence_transformers.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "5.1.1",
4
+ "transformers": "4.57.6",
5
+ "pytorch": "2.10.0"
6
+ },
7
+ "model_type": "SentenceTransformer",
8
+ "prompts": {
9
+ "query": "",
10
+ "document": ""
11
+ },
12
+ "default_prompt_name": null,
13
+ "similarity_fn_name": "cosine"
14
+ }
cve/domain_adapted_model/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2399d23821b60aa1c82c91d74ede3aa55d4bdd10216310cae0e6d2a4c8875522
3
+ size 90864192
cve/domain_adapted_model/modules.json ADDED
@@ -0,0 +1,20 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "idx": 0,
4
+ "name": "0",
5
+ "path": "",
6
+ "type": "sentence_transformers.models.Transformer"
7
+ },
8
+ {
9
+ "idx": 1,
10
+ "name": "1",
11
+ "path": "1_Pooling",
12
+ "type": "sentence_transformers.models.Pooling"
13
+ },
14
+ {
15
+ "idx": 2,
16
+ "name": "2",
17
+ "path": "2_Normalize",
18
+ "type": "sentence_transformers.models.Normalize"
19
+ }
20
+ ]
cve/domain_adapted_model/sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 256,
3
+ "do_lower_case": false
4
+ }
cve/domain_adapted_model/special_tokens_map.json ADDED
@@ -0,0 +1,37 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cls_token": {
3
+ "content": "[CLS]",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "mask_token": {
10
+ "content": "[MASK]",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "[PAD]",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "sep_token": {
24
+ "content": "[SEP]",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ },
30
+ "unk_token": {
31
+ "content": "[UNK]",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false
36
+ }
37
+ }
cve/domain_adapted_model/tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
cve/domain_adapted_model/tokenizer_config.json ADDED
@@ -0,0 +1,65 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "0": {
4
+ "content": "[PAD]",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "100": {
12
+ "content": "[UNK]",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "101": {
20
+ "content": "[CLS]",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "102": {
28
+ "content": "[SEP]",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "103": {
36
+ "content": "[MASK]",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ }
43
+ },
44
+ "clean_up_tokenization_spaces": false,
45
+ "cls_token": "[CLS]",
46
+ "do_basic_tokenize": true,
47
+ "do_lower_case": true,
48
+ "extra_special_tokens": {},
49
+ "mask_token": "[MASK]",
50
+ "max_length": 128,
51
+ "model_max_length": 256,
52
+ "never_split": null,
53
+ "pad_to_multiple_of": null,
54
+ "pad_token": "[PAD]",
55
+ "pad_token_type_id": 0,
56
+ "padding_side": "right",
57
+ "sep_token": "[SEP]",
58
+ "stride": 0,
59
+ "strip_accents": null,
60
+ "tokenize_chinese_chars": true,
61
+ "tokenizer_class": "BertTokenizer",
62
+ "truncation_side": "right",
63
+ "truncation_strategy": "longest_first",
64
+ "unk_token": "[UNK]"
65
+ }
cve/domain_adapted_model/vocab.txt ADDED
The diff for this file is too large to render. See raw diff