Hugging Face's logo

ancc
/
Velvet-2B-embedding-news

Sentence Similarity
sentence-transformers
Safetensors
mistral
feature-extraction
Generated from Trainer
dataset_size:152913
loss:BatchAllTripletLoss
Eval Results
text-embeddings-inference
Model card Files Community
ancc commited on
Commit
689eeef
·
verified ·
1 Parent(s): a8def79

Add new SentenceTransformer model

Browse files
Files changed (10) hide show
  1. 1_Pooling/config.json +10 -0
  2. README.md +432 -0
  3. config.json +27 -0
  4. config_sentence_transformers.json +10 -0
  5. model.safetensors +3 -0
  6. modules.json +20 -0
  7. sentence_bert_config.json +4 -0
  8. special_tokens_map.json +30 -0
  9. tokenizer.json +0 -0
  10. tokenizer_config.json +0 -0
1_Pooling/config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "word_embedding_dimension": 2048,
3
+ "pooling_mode_cls_token": false,
4
+ "pooling_mode_mean_tokens": false,
5
+ "pooling_mode_max_tokens": false,
6
+ "pooling_mode_mean_sqrt_len_tokens": false,
7
+ "pooling_mode_weightedmean_tokens": false,
8
+ "pooling_mode_lasttoken": true,
9
+ "include_prompt": true
10
+ }
README.md ADDED
@@ -0,0 +1,432 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - sentence-transformers
4
+ - sentence-similarity
5
+ - feature-extraction
6
+ - generated_from_trainer
7
+ - dataset_size:152913
8
+ - loss:BatchAllTripletLoss
9
+ base_model: Almawave/Velvet-2B
10
+ widget:
11
+ - source_sentence: La crisi non tocca il mercato del “luxury food”, che continua a
12
+ crescere
13
+ sentences:
14
+ - Avviso aggiornamento (1.55)
15
+ - Il fotoritocco sui social per fingere di essere al Mic
16
+ - Turchia, Biberovic giocherà da passaportato nella finestra FIBA
17
+ - source_sentence: 'Miller stagione finita, Pagani: «Rimpiazzo? Valutiamo il mercato
18
+ europeo»'
19
+ sentences:
20
+ - Beautiful, le trame della settimana dal 2 al 7 dicembre
21
+ - 'Corte dei conti: concorso per 8 funzionari a tempo indeterminato'
22
+ - 'Leonardo, AD: incontrato stamani numero uno Airbus su alleanza satellitare'
23
+ - source_sentence: 'Il segreto di Jalen Hurts in una foto sullo smartphone: così ha
24
+ vinto e sconfitto gli scettici'
25
+ sentences:
26
+ - Scarcerato il boss Ernesto Fazzalari, era il latitante più ricercato dopo Messina
27
+ Denaro
28
+ - 'UniVdA: ecco il nuovo master in Psicologia dello sport'
29
+ - 'Gran Turismo 7: arrivano quattro nuove auto con l’update 1.55 [VIDEO]'
30
+ - source_sentence: 'San Vito al Torre, recuperato un antico monumento funerario romano
31
+ dal fiume: la scoperta'
32
+ sentences:
33
+ - Charlotte Casiraghi in pubblico dopo le voci di divorzio
34
+ - Mr. Bezos, la sua non è imparzialità ma viltà
35
+ - 'Elisa Di Francisca a La Talpa: “Sono troppo vera per tenere segreti”'
36
+ - source_sentence: 'NBA, dopo l’addio a Schroeder i Nets promuovono Simmons in quintetto:
37
+ l''idea è correre'
38
+ sentences:
39
+ - 'Picasso a Milano: al Mudec la mostra sulle metamorfosi del maestro'
40
+ - Italia Viva e +Europa non parteciperanno alle elezioni regionali in Liguria
41
+ - 'Achille Costacurta rivela: «Sono stato rinchiuso per un anno e sette mesi in
42
+ un centro penale'
43
+ pipeline_tag: sentence-similarity
44
+ library_name: sentence-transformers
45
+ metrics:
46
+ - average_precision
47
+ - f1
48
+ - precision
49
+ - recall
50
+ - threshold
51
+ model-index:
52
+ - name: SentenceTransformer based on Almawave/Velvet-2B
53
+ results:
54
+ - task:
55
+ type: paraphrase-mining
56
+ name: Paraphrase Mining
57
+ dataset:
58
+ name: Unknown
59
+ type: unknown
60
+ metrics:
61
+ - type: average_precision
62
+ value: 0.5283532795784699
63
+ name: Average Precision
64
+ - type: f1
65
+ value: 0.5502357974952371
66
+ name: F1
67
+ - type: precision
68
+ value: 0.5567564151181899
69
+ name: Precision
70
+ - type: recall
71
+ value: 0.5438661480521084
72
+ name: Recall
73
+ - type: threshold
74
+ value: 0.9310455322265625
75
+ name: Threshold
76
+ ---
77
+
78
+ # SentenceTransformer based on Almawave/Velvet-2B
79
+
80
+ This is a [sentence-transformers](https://www.SBERT.net) model finetuned from [Almawave/Velvet-2B](https://huggingface.co/Almawave/Velvet-2B). It maps sentences & paragraphs to a 2048-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.
81
+
82
+ ## Model Details
83
+
84
+ ### Model Description
85
+ - **Model Type:** Sentence Transformer
86
+ - **Base model:** [Almawave/Velvet-2B](https://huggingface.co/Almawave/Velvet-2B) <!-- at revision 3b864694ae4d80923ac39cee130c5eeb7d8808b6 -->
87
+ - **Maximum Sequence Length:** 32768 tokens
88
+ - **Output Dimensionality:** 2048 dimensions
89
+ - **Similarity Function:** Cosine Similarity
90
+ <!-- - **Training Dataset:** Unknown -->
91
+ <!-- - **Language:** Unknown -->
92
+ <!-- - **License:** Unknown -->
93
+
94
+ ### Model Sources
95
+
96
+ - **Documentation:** [Sentence Transformers Documentation](https://sbert.net)
97
+ - **Repository:** [Sentence Transformers on GitHub](https://github.com/UKPLab/sentence-transformers)
98
+ - **Hugging Face:** [Sentence Transformers on Hugging Face](https://huggingface.co/models?library=sentence-transformers)
99
+
100
+ ### Full Model Architecture
101
+
102
+ ```
103
+ SentenceTransformer(
104
+ (0): Transformer({'max_seq_length': 32768, 'do_lower_case': False}) with Transformer model: MistralModel
105
+ (1): Pooling({'word_embedding_dimension': 2048, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': True, 'include_prompt': True})
106
+ (2): Normalize()
107
+ )
108
+ ```
109
+
110
+ ## Usage
111
+
112
+ ### Direct Usage (Sentence Transformers)
113
+
114
+ First install the Sentence Transformers library:
115
+
116
+ ```bash
117
+ pip install -U sentence-transformers
118
+ ```
119
+
120
+ Then you can load this model and run inference.
121
+ ```python
122
+ from sentence_transformers import SentenceTransformer
123
+
124
+ # Download from the 🤗 Hub
125
+ model = SentenceTransformer("ancc/Velvet-2B-embedding-news")
126
+ # Run inference
127
+ sentences = [
128
+ "NBA, dopo l’addio a Schroeder i Nets promuovono Simmons in quintetto: l'idea è correre",
129
+ 'Italia Viva e +Europa non parteciperanno alle elezioni regionali in Liguria',
130
+ 'Achille Costacurta rivela: «Sono stato rinchiuso per un anno e sette mesi in un centro penale',
131
+ ]
132
+ embeddings = model.encode(sentences)
133
+ print(embeddings.shape)
134
+ # [3, 2048]
135
+
136
+ # Get the similarity scores for the embeddings
137
+ similarities = model.similarity(embeddings, embeddings)
138
+ print(similarities.shape)
139
+ # [3, 3]
140
+ ```
141
+
142
+ <!--
143
+ ### Direct Usage (Transformers)
144
+
145
+ <details><summary>Click to see the direct usage in Transformers</summary>
146
+
147
+ </details>
148
+ -->
149
+
150
+ <!--
151
+ ### Downstream Usage (Sentence Transformers)
152
+
153
+ You can finetune this model on your own dataset.
154
+
155
+ <details><summary>Click to expand</summary>
156
+
157
+ </details>
158
+ -->
159
+
160
+ <!--
161
+ ### Out-of-Scope Use
162
+
163
+ *List how the model may foreseeably be misused and address what users ought not to do with the model.*
164
+ -->
165
+
166
+ ## Evaluation
167
+
168
+ ### Metrics
169
+
170
+ #### Paraphrase Mining
171
+
172
+ * Evaluated with [<code>ParaphraseMiningEvaluator</code>](https://sbert.net/docs/package_reference/sentence_transformer/evaluation.html#sentence_transformers.evaluation.ParaphraseMiningEvaluator)
173
+
174
+ | Metric | Value |
175
+ |:----------------------|:-----------|
176
+ | **average_precision** | **0.5284** |
177
+ | f1 | 0.5502 |
178
+ | precision | 0.5568 |
179
+ | recall | 0.5439 |
180
+ | threshold | 0.931 |
181
+
182
+ <!--
183
+ ## Bias, Risks and Limitations
184
+
185
+ *What are the known or foreseeable issues stemming from this model? You could also flag here known failure cases or weaknesses of the model.*
186
+ -->
187
+
188
+ <!--
189
+ ### Recommendations
190
+
191
+ *What are recommendations with respect to the foreseeable issues? For example, filtering explicit content.*
192
+ -->
193
+
194
+ ## Training Details
195
+
196
+ ### Training Dataset
197
+
198
+ #### Unnamed Dataset
199
+
200
+ * Size: 152,913 training samples
201
+ * Columns: <code>sentence</code> and <code>label</code>
202
+ * Approximate statistics based on the first 1000 samples:
203
+ | | sentence | label |
204
+ |:--------|:----------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
205
+ | type | string | int |
206
+ | details | <ul><li>min: 3 tokens</li><li>mean: 19.14 tokens</li><li>max: 47 tokens</li></ul> | <ul><li>0: ~0.50%</li><li>1: ~0.50%</li><li>2: ~0.50%</li><li>3: ~0.50%</li><li>4: ~0.50%</li><li>5: ~0.20%</li><li>6: ~0.50%</li><li>7: ~0.50%</li><li>8: ~0.20%</li><li>9: ~0.50%</li><li>11: ~0.50%</li><li>12: ~0.50%</li><li>13: ~0.50%</li><li>15: ~0.50%</li><li>16: ~0.20%</li><li>17: ~0.30%</li><li>18: ~0.50%</li><li>19: ~0.40%</li><li>20: ~0.50%</li><li>21: ~0.50%</li><li>22: ~0.30%</li><li>23: ~0.50%</li><li>24: ~0.50%</li><li>25: ~0.50%</li><li>26: ~0.50%</li><li>27: ~0.50%</li><li>28: ~0.40%</li><li>29: ~0.50%</li><li>30: ~0.20%</li><li>31: ~0.30%</li><li>32: ~0.40%</li><li>33: ~0.50%</li><li>34: ~0.50%</li><li>36: ~0.50%</li><li>38: ~0.50%</li><li>39: ~0.50%</li><li>40: ~0.30%</li><li>41: ~0.20%</li><li>43: ~0.50%</li><li>44: ~0.50%</li><li>45: ~0.50%</li><li>46: ~0.40%</li><li>47: ~0.30%</li><li>48: ~0.30%</li><li>49: ~0.50%</li><li>50: ~0.30%</li><li>51: ~0.50%</li><li>52: ~0.50%</li><li>53: ~0.50%</li><li>54: ~0.50%</li><li>55: ~0.20%</li><li>56: ~0.50%</li><li>57: ~0.50%</li><li>59: ~0.50%</li><li>60: ~0.40%</li><li>62: ~0.50%</li><li>63: ~0.20%</li><li>64: ~0.50%</li><li>65: ~0.40%</li><li>66: ~0.50%</li><li>67: ~0.40%</li><li>68: ~0.50%</li><li>69: ~0.50%</li><li>70: ~0.50%</li><li>71: ~0.20%</li><li>72: ~0.50%</li><li>73: ~0.20%</li><li>74: ~0.50%</li><li>75: ~0.50%</li><li>77: ~0.50%</li><li>78: ~0.30%</li><li>79: ~0.50%</li><li>80: ~0.40%</li><li>81: ~0.30%</li><li>82: ~0.20%</li><li>83: ~0.40%</li><li>84: ~0.30%</li><li>85: ~0.50%</li><li>87: ~0.50%</li><li>88: ~0.20%</li><li>90: ~0.50%</li><li>91: ~0.20%</li><li>92: ~0.50%</li><li>93: ~0.40%</li><li>94: ~0.50%</li><li>95: ~0.50%</li><li>97: ~0.50%</li><li>98: ~0.40%</li><li>99: ~0.50%</li><li>100: ~0.30%</li><li>101: ~0.50%</li><li>103: ~0.20%</li><li>104: ~0.50%</li><li>106: ~0.40%</li><li>107: ~0.20%</li><li>108: ~0.40%</li><li>109: ~0.30%</li><li>110: ~0.50%</li><li>111: ~0.40%</li><li>112: ~0.50%</li><li>113: ~0.30%</li><li>115: ~0.30%</li><li>116: ~0.50%</li><li>117: ~0.50%</li><li>118: ~0.30%</li><li>120: ~0.30%</li><li>121: ~0.50%</li><li>122: ~0.20%</li><li>123: ~0.40%</li><li>124: ~0.30%</li><li>125: ~0.40%</li><li>126: ~0.50%</li><li>127: ~0.50%</li><li>128: ~0.50%</li><li>129: ~0.50%</li><li>130: ~0.50%</li><li>131: ~0.50%</li><li>132: ~0.40%</li><li>133: ~0.50%</li><li>134: ~0.30%</li><li>135: ~0.50%</li><li>136: ~0.30%</li><li>137: ~0.30%</li><li>138: ~0.50%</li><li>139: ~0.50%</li><li>140: ~0.20%</li><li>141: ~0.50%</li><li>142: ~0.50%</li><li>143: ~0.50%</li><li>144: ~0.30%</li><li>145: ~0.30%</li><li>146: ~0.30%</li><li>147: ~0.50%</li><li>148: ~0.30%</li><li>149: ~0.20%</li><li>150: ~0.50%</li><li>151: ~0.30%</li><li>152: ~0.20%</li><li>153: ~0.50%</li><li>154: ~0.50%</li><li>155: ~0.40%</li><li>156: ~0.20%</li><li>158: ~0.50%</li><li>159: ~0.50%</li><li>160: ~0.50%</li><li>161: ~0.40%</li><li>162: ~0.20%</li><li>163: ~0.40%</li><li>164: ~0.20%</li><li>165: ~0.50%</li><li>166: ~0.50%</li><li>167: ~0.50%</li><li>168: ~0.20%</li><li>169: ~0.50%</li><li>170: ~0.50%</li><li>171: ~0.20%</li><li>172: ~0.50%</li><li>173: ~0.50%</li><li>174: ~0.40%</li><li>175: ~0.50%</li><li>176: ~0.30%</li><li>178: ~0.50%</li><li>180: ~0.20%</li><li>182: ~0.50%</li><li>183: ~0.50%</li><li>184: ~0.30%</li><li>185: ~0.50%</li><li>187: ~0.20%</li><li>188: ~0.50%</li><li>189: ~0.50%</li><li>191: ~0.50%</li><li>192: ~0.50%</li><li>193: ~0.50%</li><li>194: ~0.20%</li><li>195: ~0.30%</li><li>196: ~0.50%</li><li>197: ~0.50%</li><li>199: ~0.50%</li><li>200: ~0.50%</li><li>201: ~0.20%</li><li>202: ~0.50%</li><li>203: ~0.50%</li><li>204: ~0.30%</li><li>205: ~0.50%</li><li>206: ~0.20%</li><li>207: ~0.40%</li><li>208: ~0.50%</li><li>209: ~0.30%</li><li>210: ~0.30%</li><li>211: ~0.50%</li><li>212: ~0.20%</li><li>213: ~0.50%</li><li>214: ~0.50%</li><li>215: ~0.40%</li><li>216: ~0.50%</li><li>217: ~0.40%</li><li>218: ~0.50%</li><li>219: ~0.50%</li><li>220: ~0.20%</li><li>221: ~0.50%</li><li>222: ~0.40%</li><li>223: ~0.50%</li><li>224: ~0.30%</li><li>225: ~0.40%</li><li>227: ~0.20%</li><li>228: ~0.30%</li><li>230: ~0.40%</li><li>231: ~0.40%</li><li>232: ~0.50%</li><li>233: ~0.50%</li><li>234: ~0.50%</li><li>235: ~0.50%</li><li>236: ~0.40%</li><li>237: ~0.50%</li><li>238: ~0.40%</li><li>239: ~0.50%</li><li>240: ~0.50%</li><li>241: ~0.50%</li><li>242: ~0.30%</li><li>243: ~0.50%</li><li>244: ~0.50%</li><li>245: ~0.50%</li><li>246: ~0.50%</li><li>247: ~0.20%</li><li>248: ~0.50%</li><li>249: ~0.50%</li><li>250: ~0.50%</li><li>251: ~0.50%</li><li>252: ~0.20%</li><li>253: ~0.50%</li><li>254: ~0.50%</li><li>255: ~0.30%</li><li>256: ~0.50%</li><li>257: ~0.50%</li><li>258: ~0.20%</li><li>259: ~0.50%</li><li>260: ~0.30%</li><li>261: ~0.50%</li><li>262: ~0.50%</li><li>263: ~0.20%</li><li>264: ~0.50%</li><li>265: ~0.20%</li></ul> |
207
+ * Samples:
208
+ | sentence | label |
209
+ |:-------------------------------------------------------------------------------------------|:---------------|
210
+ | <code>MERCATO LBA - Treviso, Giofrè: "Mercato in continua osservazione, vedremo..."</code> | <code>0</code> |
211
+ | <code>Ky Bowman: Non sono soddisfatto delle mie performance</code> | <code>0</code> |
212
+ | <code>LBA - Treviso, Giofrè: "Sabato la Reggiana, dobbiamo vincere. Punto"</code> | <code>0</code> |
213
+ * Loss: [<code>BatchAllTripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#batchalltripletloss)
214
+
215
+ ### Evaluation Dataset
216
+
217
+ #### Unnamed Dataset
218
+
219
+ * Size: 9,310 evaluation samples
220
+ * Columns: <code>sentence</code> and <code>label</code>
221
+ * Approximate statistics based on the first 1000 samples:
222
+ | | sentence | label |
223
+ |:--------|:----------------------------------------------------------------------------------|:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
224
+ | type | string | int |
225
+ | details | <ul><li>min: 4 tokens</li><li>mean: 19.04 tokens</li><li>max: 78 tokens</li></ul> | <ul><li>0: ~0.50%</li><li>1: ~0.30%</li><li>2: ~0.50%</li><li>4: ~0.50%</li><li>5: ~0.20%</li><li>6: ~0.30%</li><li>7: ~0.50%</li><li>8: ~0.50%</li><li>9: ~0.50%</li><li>10: ~0.50%</li><li>12: ~0.40%</li><li>13: ~0.30%</li><li>14: ~0.30%</li><li>15: ~0.40%</li><li>16: ~0.20%</li><li>17: ~0.50%</li><li>18: ~0.40%</li><li>19: ~0.40%</li><li>20: ~0.50%</li><li>21: ~0.50%</li><li>22: ~0.50%</li><li>23: ~0.50%</li><li>24: ~0.20%</li><li>25: ~0.20%</li><li>26: ~0.50%</li><li>27: ~0.50%</li><li>28: ~0.50%</li><li>30: ~0.50%</li><li>31: ~0.20%</li><li>32: ~0.30%</li><li>33: ~0.50%</li><li>34: ~0.20%</li><li>35: ~0.50%</li><li>37: ~0.50%</li><li>38: ~0.20%</li><li>39: ~0.50%</li><li>40: ~0.50%</li><li>41: ~0.20%</li><li>42: ~0.30%</li><li>43: ~0.30%</li><li>44: ~0.50%</li><li>45: ~0.30%</li><li>46: ~0.50%</li><li>47: ~0.50%</li><li>48: ~0.30%</li><li>49: ~0.40%</li><li>50: ~0.30%</li><li>51: ~0.20%</li><li>52: ~0.50%</li><li>53: ~0.20%</li><li>54: ~0.30%</li><li>55: ~0.50%</li><li>56: ~0.50%</li><li>57: ~0.20%</li><li>58: ~0.50%</li><li>59: ~0.50%</li><li>60: ~0.50%</li><li>61: ~0.30%</li><li>62: ~0.30%</li><li>63: ~0.20%</li><li>65: ~0.50%</li><li>67: ~0.40%</li><li>68: ~0.50%</li><li>69: ~0.50%</li><li>70: ~0.20%</li><li>71: ~0.40%</li><li>72: ~0.50%</li><li>73: ~0.50%</li><li>74: ~0.30%</li><li>77: ~0.20%</li><li>78: ~0.50%</li><li>79: ~0.50%</li><li>80: ~0.20%</li><li>81: ~0.50%</li><li>82: ~0.20%</li><li>83: ~0.50%</li><li>84: ~0.50%</li><li>85: ~0.50%</li><li>89: ~0.20%</li><li>90: ~0.50%</li><li>91: ~0.50%</li><li>92: ~0.30%</li><li>93: ~0.40%</li><li>94: ~0.50%</li><li>95: ~0.30%</li><li>96: ~0.20%</li><li>97: ~0.50%</li><li>98: ~0.20%</li><li>99: ~0.50%</li><li>100: ~0.50%</li><li>101: ~0.50%</li><li>102: ~0.50%</li><li>103: ~0.30%</li><li>104: ~0.50%</li><li>105: ~0.50%</li><li>106: ~0.50%</li><li>107: ~0.50%</li><li>108: ~0.50%</li><li>109: ~0.50%</li><li>110: ~0.20%</li><li>111: ~0.50%</li><li>112: ~0.50%</li><li>113: ~0.20%</li><li>114: ~0.50%</li><li>115: ~0.40%</li><li>116: ~0.50%</li><li>117: ~0.20%</li><li>118: ~0.50%</li><li>120: ~0.50%</li><li>121: ~0.30%</li><li>122: ~0.50%</li><li>123: ~0.40%</li><li>124: ~0.20%</li><li>125: ~0.50%</li><li>126: ~0.20%</li><li>129: ~0.50%</li><li>130: ~0.30%</li><li>131: ~0.50%</li><li>132: ~0.40%</li><li>133: ~0.50%</li><li>134: ~0.50%</li><li>135: ~0.50%</li><li>136: ~0.20%</li><li>137: ~0.30%</li><li>138: ~0.50%</li><li>140: ~0.30%</li><li>141: ~0.30%</li><li>142: ~0.50%</li><li>143: ~0.40%</li><li>144: ~0.50%</li><li>145: ~0.50%</li><li>146: ~0.30%</li><li>147: ~0.20%</li><li>148: ~0.50%</li><li>149: ~0.40%</li><li>150: ~0.50%</li><li>151: ~0.20%</li><li>152: ~0.50%</li><li>154: ~0.20%</li><li>155: ~0.20%</li><li>156: ~0.50%</li><li>157: ~0.50%</li><li>158: ~0.50%</li><li>159: ~0.50%</li><li>160: ~0.20%</li><li>162: ~0.50%</li><li>163: ~0.20%</li><li>164: ~0.20%</li><li>165: ~0.20%</li><li>166: ~0.20%</li><li>167: ~0.30%</li><li>168: ~0.50%</li><li>169: ~0.50%</li><li>170: ~0.50%</li><li>171: ~0.40%</li><li>172: ~0.30%</li><li>173: ~0.50%</li><li>174: ~0.50%</li><li>175: ~0.30%</li><li>176: ~0.50%</li><li>178: ~0.50%</li><li>179: ~0.20%</li><li>180: ~0.50%</li><li>181: ~0.50%</li><li>182: ~0.30%</li><li>183: ~0.40%</li><li>184: ~0.20%</li><li>185: ~0.50%</li><li>186: ~0.50%</li><li>187: ~0.50%</li><li>188: ~0.20%</li><li>189: ~0.50%</li><li>191: ~0.20%</li><li>192: ~0.50%</li><li>193: ~0.20%</li><li>194: ~0.20%</li><li>195: ~0.40%</li><li>196: ~0.20%</li><li>197: ~0.20%</li><li>198: ~0.50%</li><li>199: ~0.40%</li><li>200: ~0.50%</li><li>201: ~0.50%</li><li>202: ~0.50%</li><li>203: ~0.40%</li><li>205: ~0.50%</li><li>206: ~0.50%</li><li>207: ~0.50%</li><li>208: ~0.50%</li><li>209: ~0.50%</li><li>210: ~0.50%</li><li>211: ~0.40%</li><li>212: ~0.50%</li><li>213: ~0.50%</li><li>214: ~0.20%</li><li>215: ~0.20%</li><li>216: ~0.50%</li><li>217: ~0.50%</li><li>218: ~0.50%</li><li>219: ~0.30%</li><li>220: ~0.50%</li><li>221: ~0.50%</li><li>223: ~0.50%</li><li>224: ~0.50%</li><li>225: ~0.50%</li><li>227: ~0.20%</li><li>228: ~0.50%</li><li>230: ~0.20%</li><li>231: ~0.50%</li><li>232: ~0.50%</li><li>234: ~0.30%</li><li>235: ~0.40%</li><li>236: ~0.50%</li><li>237: ~0.50%</li><li>238: ~0.20%</li><li>239: ~0.50%</li><li>240: ~0.20%</li><li>241: ~0.50%</li><li>242: ~0.20%</li><li>243: ~0.20%</li><li>244: ~0.20%</li><li>246: ~0.50%</li><li>247: ~0.20%</li><li>248: ~0.50%</li><li>249: ~0.50%</li><li>250: ~0.40%</li><li>251: ~0.50%</li><li>252: ~0.50%</li><li>253: ~0.20%</li><li>254: ~0.50%</li><li>255: ~0.50%</li><li>256: ~0.50%</li><li>257: ~0.50%</li><li>258: ~0.20%</li><li>259: ~0.20%</li><li>260: ~0.50%</li><li>261: ~0.50%</li><li>262: ~0.20%</li><li>263: ~0.40%</li><li>264: ~0.50%</li><li>265: ~0.50%</li><li>266: ~0.50%</li><li>267: ~0.50%</li><li>268: ~0.20%</li><li>269: ~0.20%</li><li>270: ~0.40%</li><li>272: ~0.30%</li><li>273: ~0.20%</li><li>274: ~0.50%</li><li>275: ~0.50%</li><li>276: ~0.40%</li><li>277: ~0.30%</li><li>278: ~0.30%</li></ul> |
226
+ * Samples:
227
+ | sentence | label |
228
+ |:----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:---------------|
229
+ | <code>Supplenze: come funzionano i contratti fino al 31 dicembre 2024 e il calcolo del punteggio?</code> | <code>0</code> |
230
+ | <code>Docente non abilitato assunto a tempo determinato da concorso PNRR1: in quale scuola "andrò a finire" se nella mia si perde un posto?</code> | <code>0</code> |
231
+ | <code>Docenti non abilitati nominati dopo il 31 agosto da graduatorie pubblicate prima: otterranno sede di titolarità all’esito delle operazioni di mobilità [Chiarimenti]</code> | <code>0</code> |
232
+ * Loss: [<code>BatchAllTripletLoss</code>](https://sbert.net/docs/package_reference/sentence_transformer/losses.html#batchalltripletloss)
233
+
234
+ ### Training Hyperparameters
235
+ #### Non-Default Hyperparameters
236
+
237
+ - `per_device_train_batch_size`: 32
238
+ - `num_train_epochs`: 4
239
+ - `lr_scheduler_type`: cosine
240
+ - `warmup_ratio`: 0.25
241
+ - `seed`: 17
242
+ - `data_seed`: 17
243
+ - `bf16`: True
244
+ - `batch_sampler`: group_by_label
245
+
246
+ #### All Hyperparameters
247
+ <details><summary>Click to expand</summary>
248
+
249
+ - `overwrite_output_dir`: False
250
+ - `do_predict`: False
251
+ - `eval_strategy`: no
252
+ - `prediction_loss_only`: True
253
+ - `per_device_train_batch_size`: 32
254
+ - `per_device_eval_batch_size`: 8
255
+ - `per_gpu_train_batch_size`: None
256
+ - `per_gpu_eval_batch_size`: None
257
+ - `gradient_accumulation_steps`: 1
258
+ - `eval_accumulation_steps`: None
259
+ - `torch_empty_cache_steps`: None
260
+ - `learning_rate`: 5e-05
261
+ - `weight_decay`: 0.0
262
+ - `adam_beta1`: 0.9
263
+ - `adam_beta2`: 0.999
264
+ - `adam_epsilon`: 1e-08
265
+ - `max_grad_norm`: 1.0
266
+ - `num_train_epochs`: 4
267
+ - `max_steps`: -1
268
+ - `lr_scheduler_type`: cosine
269
+ - `lr_scheduler_kwargs`: {}
270
+ - `warmup_ratio`: 0.25
271
+ - `warmup_steps`: 0
272
+ - `log_level`: passive
273
+ - `log_level_replica`: warning
274
+ - `log_on_each_node`: True
275
+ - `logging_nan_inf_filter`: True
276
+ - `save_safetensors`: True
277
+ - `save_on_each_node`: False
278
+ - `save_only_model`: False
279
+ - `restore_callback_states_from_checkpoint`: False
280
+ - `no_cuda`: False
281
+ - `use_cpu`: False
282
+ - `use_mps_device`: False
283
+ - `seed`: 17
284
+ - `data_seed`: 17
285
+ - `jit_mode_eval`: False
286
+ - `use_ipex`: False
287
+ - `bf16`: True
288
+ - `fp16`: False
289
+ - `fp16_opt_level`: O1
290
+ - `half_precision_backend`: auto
291
+ - `bf16_full_eval`: False
292
+ - `fp16_full_eval`: False
293
+ - `tf32`: None
294
+ - `local_rank`: 0
295
+ - `ddp_backend`: None
296
+ - `tpu_num_cores`: None
297
+ - `tpu_metrics_debug`: False
298
+ - `debug`: []
299
+ - `dataloader_drop_last`: False
300
+ - `dataloader_num_workers`: 0
301
+ - `dataloader_prefetch_factor`: None
302
+ - `past_index`: -1
303
+ - `disable_tqdm`: False
304
+ - `remove_unused_columns`: True
305
+ - `label_names`: None
306
+ - `load_best_model_at_end`: False
307
+ - `ignore_data_skip`: False
308
+ - `fsdp`: []
309
+ - `fsdp_min_num_params`: 0
310
+ - `fsdp_config`: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
311
+ - `fsdp_transformer_layer_cls_to_wrap`: None
312
+ - `accelerator_config`: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
313
+ - `deepspeed`: None
314
+ - `label_smoothing_factor`: 0.0
315
+ - `optim`: adamw_torch
316
+ - `optim_args`: None
317
+ - `adafactor`: False
318
+ - `group_by_length`: False
319
+ - `length_column_name`: length
320
+ - `ddp_find_unused_parameters`: None
321
+ - `ddp_bucket_cap_mb`: None
322
+ - `ddp_broadcast_buffers`: False
323
+ - `dataloader_pin_memory`: True
324
+ - `dataloader_persistent_workers`: False
325
+ - `skip_memory_metrics`: True
326
+ - `use_legacy_prediction_loop`: False
327
+ - `push_to_hub`: False
328
+ - `resume_from_checkpoint`: None
329
+ - `hub_model_id`: None
330
+ - `hub_strategy`: every_save
331
+ - `hub_private_repo`: None
332
+ - `hub_always_push`: False
333
+ - `gradient_checkpointing`: False
334
+ - `gradient_checkpointing_kwargs`: None
335
+ - `include_inputs_for_metrics`: False
336
+ - `include_for_metrics`: []
337
+ - `eval_do_concat_batches`: True
338
+ - `fp16_backend`: auto
339
+ - `push_to_hub_model_id`: None
340
+ - `push_to_hub_organization`: None
341
+ - `mp_parameters`:
342
+ - `auto_find_batch_size`: False
343
+ - `full_determinism`: False
344
+ - `torchdynamo`: None
345
+ - `ray_scope`: last
346
+ - `ddp_timeout`: 1800
347
+ - `torch_compile`: False
348
+ - `torch_compile_backend`: None
349
+ - `torch_compile_mode`: None
350
+ - `dispatch_batches`: None
351
+ - `split_batches`: None
352
+ - `include_tokens_per_second`: False
353
+ - `include_num_input_tokens_seen`: False
354
+ - `neftune_noise_alpha`: None
355
+ - `optim_target_modules`: None
356
+ - `batch_eval_metrics`: False
357
+ - `eval_on_start`: False
358
+ - `use_liger_kernel`: False
359
+ - `eval_use_gather_object`: False
360
+ - `average_tokens_across_devices`: False
361
+ - `prompts`: None
362
+ - `batch_sampler`: group_by_label
363
+ - `multi_dataset_batch_sampler`: proportional
364
+
365
+ </details>
366
+
367
+ ### Training Logs
368
+ | Epoch | Step | Training Loss | Validation Loss | average_precision |
369
+ |:------:|:-----:|:-------------:|:---------------:|:-----------------:|
370
+ | 0.0002 | 1 | 0.0521 | - | - |
371
+ | 0.8002 | 3824 | 0.0195 | - | - |
372
+ | 1.7732 | 7648 | 0.0071 | - | - |
373
+ | 2.7462 | 11472 | 0.0052 | - | - |
374
+ | 3.7192 | 15296 | 0.0046 | - | - |
375
+ | 3.8272 | 15812 | - | 0.0011 | 0.5284 |
376
+
377
+
378
+ ### Framework Versions
379
+ - Python: 3.12.8
380
+ - Sentence Transformers: 3.4.1
381
+ - Transformers: 4.48.3
382
+ - PyTorch: 2.6.0+cu124
383
+ - Accelerate: 1.3.0
384
+ - Datasets: 3.2.0
385
+ - Tokenizers: 0.21.0
386
+
387
+ ## Citation
388
+
389
+ ### BibTeX
390
+
391
+ #### Sentence Transformers
392
+ ```bibtex
393
+ @inproceedings{reimers-2019-sentence-bert,
394
+ title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
395
+ author = "Reimers, Nils and Gurevych, Iryna",
396
+ booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
397
+ month = "11",
398
+ year = "2019",
399
+ publisher = "Association for Computational Linguistics",
400
+ url = "https://arxiv.org/abs/1908.10084",
401
+ }
402
+ ```
403
+
404
+ #### BatchAllTripletLoss
405
+ ```bibtex
406
+ @misc{hermans2017defense,
407
+ title={In Defense of the Triplet Loss for Person Re-Identification},
408
+ author={Alexander Hermans and Lucas Beyer and Bastian Leibe},
409
+ year={2017},
410
+ eprint={1703.07737},
411
+ archivePrefix={arXiv},
412
+ primaryClass={cs.CV}
413
+ }
414
+ ```
415
+
416
+ <!--
417
+ ## Glossary
418
+
419
+ *Clearly define terms in order to be accessible across audiences.*
420
+ -->
421
+
422
+ <!--
423
+ ## Model Card Authors
424
+
425
+ *Lists the people who create the model card, providing recognition and accountability for the detailed work that goes into its construction.*
426
+ -->
427
+
428
+ <!--
429
+ ## Model Card Contact
430
+
431
+ *Provides a way for people who have updates to the Model Card, suggestions, or questions, to contact the Model Card authors.*
432
+ -->
config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "Almawave/Velvet-2B",
3
+ "architectures": [
4
+ "MistralModel"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "eos_token_id": 2,
9
+ "head_dim": 64,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 2048,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 8192,
14
+ "max_position_embeddings": 32768,
15
+ "model_type": "mistral",
16
+ "num_attention_heads": 32,
17
+ "num_hidden_layers": 28,
18
+ "num_key_value_heads": 8,
19
+ "rms_norm_eps": 1e-05,
20
+ "rope_theta": 100000.0,
21
+ "sliding_window": null,
22
+ "tie_word_embeddings": false,
23
+ "torch_dtype": "bfloat16",
24
+ "transformers_version": "4.48.3",
25
+ "use_cache": false,
26
+ "vocab_size": 126976
27
+ }
config_sentence_transformers.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "__version__": {
3
+ "sentence_transformers": "3.4.1",
4
+ "transformers": "4.48.3",
5
+ "pytorch": "2.6.0+cu124"
6
+ },
7
+ "prompts": {},
8
+ "default_prompt_name": null,
9
+ "similarity_fn_name": "cosine"
10
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:67b5116d7db9697ff853fd6a9717a137c827ff9959918b2ce90c1bcba5cfcf0c
3
+ size 3926129848
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
+ ]
sentence_bert_config.json ADDED
@@ -0,0 +1,4 @@
 
 
 
 
 
1
+ {
2
+ "max_seq_length": 32768,
3
+ "do_lower_case": false
4
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
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
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff