Sentence Similarity
sentence-transformers
Safetensors
English
grt
geometric-rotation-transformer
feature-extraction
circular-pooling
rotational
qwen3
Instructions to use MatterWaveAI/grt-30m-embedding with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use MatterWaveAI/grt-30m-embedding with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("MatterWaveAI/grt-30m-embedding") sentences = [ "That is a happy person", "That is a happy dog", "That is a very happy person", "Today is a sunny day" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [4, 4] - Notebooks
- Google Colab
- Kaggle
Update eval_results.json with 16-way Classical Geometric, Symplectic & Toroidal optimizer ablation
Browse files- eval_results.json +113 -1
eval_results.json
CHANGED
|
@@ -130,5 +130,117 @@
|
|
| 130 |
"stsb_test_spearman_after": 0.6197,
|
| 131 |
"stsb_test_pearson_after": 0.6304
|
| 132 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
}
|
| 134 |
-
}
|
|
|
|
| 130 |
"stsb_test_spearman_after": 0.6197,
|
| 131 |
"stsb_test_pearson_after": 0.6304
|
| 132 |
}
|
| 133 |
+
},
|
| 134 |
+
"optimizer_ablations_16way": {
|
| 135 |
+
"symplectic_adam": {
|
| 136 |
+
"s1_dev_spearman": 0.6198,
|
| 137 |
+
"stsb_dev_spearman_after": 0.736,
|
| 138 |
+
"stsb_dev_pearson_after": 0.7087,
|
| 139 |
+
"stsb_test_spearman_after": 0.6516,
|
| 140 |
+
"stsb_test_pearson_after": 0.6564,
|
| 141 |
+
"rocm_elapsed_sec": 243.28
|
| 142 |
+
},
|
| 143 |
+
"polar_natural_adam": {
|
| 144 |
+
"stsb_dev_spearman_after": 0.7262,
|
| 145 |
+
"stsb_dev_pearson_after": 0.702,
|
| 146 |
+
"stsb_test_spearman_after": 0.6471,
|
| 147 |
+
"stsb_test_pearson_after": 0.6494
|
| 148 |
+
},
|
| 149 |
+
"gyric_adam": {
|
| 150 |
+
"s1_dev_spearman": 0.6221,
|
| 151 |
+
"stsb_dev_spearman_after": 0.7304,
|
| 152 |
+
"stsb_dev_pearson_after": 0.7023,
|
| 153 |
+
"stsb_test_spearman_after": 0.6455,
|
| 154 |
+
"stsb_test_pearson_after": 0.6503,
|
| 155 |
+
"rocm_elapsed_sec": 419.85
|
| 156 |
+
},
|
| 157 |
+
"huygens_adam": {
|
| 158 |
+
"s1_dev_spearman": 0.6383,
|
| 159 |
+
"stsb_dev_spearman_after": 0.7295,
|
| 160 |
+
"stsb_dev_pearson_after": 0.7035,
|
| 161 |
+
"stsb_test_spearman_after": 0.642,
|
| 162 |
+
"stsb_test_pearson_after": 0.6469,
|
| 163 |
+
"rocm_elapsed_sec": 331.8
|
| 164 |
+
},
|
| 165 |
+
"hessian_pair_adam_1x_coupling": {
|
| 166 |
+
"stsb_dev_spearman_after": 0.7242,
|
| 167 |
+
"stsb_dev_pearson_after": 0.6876,
|
| 168 |
+
"stsb_test_spearman_after": 0.6337,
|
| 169 |
+
"stsb_test_pearson_after": 0.6369
|
| 170 |
+
},
|
| 171 |
+
"beltrami_adam": {
|
| 172 |
+
"s1_dev_spearman": 0.6337,
|
| 173 |
+
"stsb_dev_spearman_after": 0.7363,
|
| 174 |
+
"stsb_dev_pearson_after": 0.7068,
|
| 175 |
+
"stsb_test_spearman_after": 0.6333,
|
| 176 |
+
"stsb_test_pearson_after": 0.6328,
|
| 177 |
+
"rocm_elapsed_sec": 321.44
|
| 178 |
+
},
|
| 179 |
+
"hannay_adam": {
|
| 180 |
+
"s1_dev_spearman": 0.6115,
|
| 181 |
+
"stsb_dev_spearman_after": 0.7257,
|
| 182 |
+
"stsb_dev_pearson_after": 0.699,
|
| 183 |
+
"stsb_test_spearman_after": 0.6333,
|
| 184 |
+
"stsb_test_pearson_after": 0.633,
|
| 185 |
+
"rocm_elapsed_sec": 239.35
|
| 186 |
+
},
|
| 187 |
+
"rotor_adam": {
|
| 188 |
+
"stsb_dev_spearman_after": 0.7322,
|
| 189 |
+
"stsb_dev_pearson_after": 0.7099,
|
| 190 |
+
"stsb_test_spearman_after": 0.6329,
|
| 191 |
+
"stsb_test_pearson_after": 0.6309
|
| 192 |
+
},
|
| 193 |
+
"toroidal_lion": {
|
| 194 |
+
"stsb_dev_spearman_after": 0.7289,
|
| 195 |
+
"stsb_dev_pearson_after": 0.697,
|
| 196 |
+
"stsb_test_spearman_after": 0.6319,
|
| 197 |
+
"stsb_test_pearson_after": 0.6341
|
| 198 |
+
},
|
| 199 |
+
"adamw": {
|
| 200 |
+
"stsb_dev_spearman_after": 0.7289,
|
| 201 |
+
"stsb_test_spearman_after": 0.6317,
|
| 202 |
+
"stsb_test_pearson_after": 0.6372
|
| 203 |
+
},
|
| 204 |
+
"poncelet_adam": {
|
| 205 |
+
"s1_dev_spearman": 0.6262,
|
| 206 |
+
"stsb_dev_spearman_after": 0.7275,
|
| 207 |
+
"stsb_dev_pearson_after": 0.6987,
|
| 208 |
+
"stsb_test_spearman_after": 0.6315,
|
| 209 |
+
"stsb_test_pearson_after": 0.6374,
|
| 210 |
+
"rocm_elapsed_sec": 267.34
|
| 211 |
+
},
|
| 212 |
+
"hessian_pair_adam_3x_coupling": {
|
| 213 |
+
"stsb_dev_spearman_after": 0.7278,
|
| 214 |
+
"stsb_dev_pearson_after": 0.6917,
|
| 215 |
+
"stsb_test_spearman_after": 0.6288,
|
| 216 |
+
"stsb_test_pearson_after": 0.6316
|
| 217 |
+
},
|
| 218 |
+
"cyclide_adam": {
|
| 219 |
+
"s1_dev_spearman": 0.6255,
|
| 220 |
+
"stsb_dev_spearman_after": 0.7368,
|
| 221 |
+
"stsb_dev_pearson_after": 0.7179,
|
| 222 |
+
"stsb_test_spearman_after": 0.6287,
|
| 223 |
+
"stsb_test_pearson_after": 0.6476,
|
| 224 |
+
"rocm_elapsed_sec": 380.42
|
| 225 |
+
},
|
| 226 |
+
"rindler_adam": {
|
| 227 |
+
"s1_dev_spearman": 0.6208,
|
| 228 |
+
"stsb_dev_spearman_after": 0.7318,
|
| 229 |
+
"stsb_dev_pearson_after": 0.7069,
|
| 230 |
+
"stsb_test_spearman_after": 0.6264,
|
| 231 |
+
"stsb_test_pearson_after": 0.6353,
|
| 232 |
+
"rocm_elapsed_sec": 271.68
|
| 233 |
+
},
|
| 234 |
+
"pair_adam_3x_coupling": {
|
| 235 |
+
"stsb_dev_spearman_after": 0.7179,
|
| 236 |
+
"stsb_test_spearman_after": 0.6242,
|
| 237 |
+
"stsb_test_pearson_after": 0.633
|
| 238 |
+
},
|
| 239 |
+
"toroidal_muon": {
|
| 240 |
+
"stsb_dev_spearman_after": 0.7361,
|
| 241 |
+
"stsb_dev_pearson_after": 0.7105,
|
| 242 |
+
"stsb_test_spearman_after": 0.6197,
|
| 243 |
+
"stsb_test_pearson_after": 0.6304
|
| 244 |
+
}
|
| 245 |
}
|
| 246 |
+
}
|