Initial model upload

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d71b98c7810f32cd761119e72b94f7efaebb9b9cdef0c12959c166256cdbd93
+oid sha256:26b7ca80ef00aa7288a188db5edf163c11a8ac1681371f4c16f17cfd5a6456db
 size 540015464

optimizer.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c4ebc28c2ebca4862776a9f89c3c8de972438cd172a22813b89b57db18dc44d2
+oid sha256:d3739f11dcf2f0ed8e7c54b63ea0dc0f8965d399c942b60f2377e57983e9dff7
 size 1080155450

rng_state.pth

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:19059bd5ccf01b60d1ddae0c675995666bd984bcf457cfdcaf9516d7d8ff43fe
+oid sha256:dfeb8c590edf888671f99518f74ec69b152edc0fe62d6ec619109abb0aabd9ad
 size 14244

scheduler.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:37d2c3099ed2a7c9bdcdc0c3d1058ad81e0894112175f8523a80f541d102e381
+oid sha256:919da8dcf4d5c21cc901f05c12ea815f92297aa610dd89eab8d81c1d5537f624
 size 1064

trainer_state.json

@@ -1,380 +1,562 @@
 {
-  "best_metric": 0.513288052386553,
-  "best_model_checkpoint": "bkai-fine-tuned-legal/checkpoint-240",
-  "epoch": 4.0,
+  "best_metric": 0.16643136739730835,
+  "best_model_checkpoint": "bkai-fine-tuned-legal/checkpoint-282",
+  "epoch": 5.878048780487805,
   "eval_steps": 500,
-  "global_step": 240,
+  "global_step": 282,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 3051630.25,
-      "learning_rate": 2.9970400926424075e-05,
-      "loss": 0.9663,
-      "step": 60
+      "grad_norm": 3041986.25,
+      "learning_rate": 2.9967614628878062e-05,
+      "loss": 1.1542,
+      "step": 48
     },
     {
       "epoch": 1.0,
-      "eval_dim_128_cosine_accuracy@1": 0.2080602426439827,
-      "eval_dim_128_cosine_accuracy@10": 0.6729884256031237,
-      "eval_dim_128_cosine_accuracy@3": 0.4049644401059824,
-      "eval_dim_128_cosine_accuracy@5": 0.49504950495049505,
-      "eval_dim_128_cosine_map@100": 0.34660918730461937,
-      "eval_dim_128_cosine_mrr@10": 0.3358910891089097,
-      "eval_dim_128_cosine_ndcg@10": 0.41500734332231626,
-      "eval_dim_128_cosine_precision@1": 0.2080602426439827,
-      "eval_dim_128_cosine_precision@10": 0.06729884256031238,
-      "eval_dim_128_cosine_precision@3": 0.13498814670199413,
-      "eval_dim_128_cosine_precision@5": 0.09900990099009901,
-      "eval_dim_128_cosine_recall@1": 0.2080602426439827,
-      "eval_dim_128_cosine_recall@10": 0.6729884256031237,
-      "eval_dim_128_cosine_recall@3": 0.4049644401059824,
-      "eval_dim_128_cosine_recall@5": 0.49504950495049505,
-      "eval_dim_256_cosine_accuracy@1": 0.21321991354064984,
-      "eval_dim_256_cosine_accuracy@10": 0.6862362292567284,
-      "eval_dim_256_cosine_accuracy@3": 0.41598103472319065,
-      "eval_dim_256_cosine_accuracy@5": 0.5063450006972529,
-      "eval_dim_256_cosine_map@100": 0.3546268888431031,
-      "eval_dim_256_cosine_mrr@10": 0.34420854057236694,
-      "eval_dim_256_cosine_ndcg@10": 0.424598027833536,
-      "eval_dim_256_cosine_precision@1": 0.21321991354064984,
-      "eval_dim_256_cosine_precision@10": 0.06862362292567285,
-      "eval_dim_256_cosine_precision@3": 0.1386603449077302,
-      "eval_dim_256_cosine_precision@5": 0.10126900013945056,
-      "eval_dim_256_cosine_recall@1": 0.21321991354064984,
-      "eval_dim_256_cosine_recall@10": 0.6862362292567284,
-      "eval_dim_256_cosine_recall@3": 0.41598103472319065,
-      "eval_dim_256_cosine_recall@5": 0.5063450006972529,
-      "eval_dim_512_cosine_accuracy@1": 0.21433551805884815,
-      "eval_dim_512_cosine_accuracy@10": 0.6919537024124948,
-      "eval_dim_512_cosine_accuracy@3": 0.4211407056198578,
-      "eval_dim_512_cosine_accuracy@5": 0.508855110863199,
-      "eval_dim_512_cosine_map@100": 0.3564257538175301,
-      "eval_dim_512_cosine_mrr@10": 0.34621192501543907,
-      "eval_dim_512_cosine_ndcg@10": 0.42739973393443165,
-      "eval_dim_512_cosine_precision@1": 0.21433551805884815,
-      "eval_dim_512_cosine_precision@10": 0.06919537024124948,
-      "eval_dim_512_cosine_precision@3": 0.14038023520661924,
-      "eval_dim_512_cosine_precision@5": 0.1017710221726398,
-      "eval_dim_512_cosine_recall@1": 0.21433551805884815,
-      "eval_dim_512_cosine_recall@10": 0.6919537024124948,
-      "eval_dim_512_cosine_recall@3": 0.4211407056198578,
-      "eval_dim_512_cosine_recall@5": 0.508855110863199,
-      "eval_dim_64_cosine_accuracy@1": 0.19704364802677451,
-      "eval_dim_64_cosine_accuracy@10": 0.6503974341096082,
-      "eval_dim_64_cosine_accuracy@3": 0.3846046576488635,
-      "eval_dim_64_cosine_accuracy@5": 0.48096499790824154,
-      "eval_dim_64_cosine_map@100": 0.3332997375022508,
-      "eval_dim_64_cosine_mrr@10": 0.3223062799237654,
-      "eval_dim_64_cosine_ndcg@10": 0.399385228043632,
-      "eval_dim_64_cosine_precision@1": 0.19704364802677451,
-      "eval_dim_64_cosine_precision@10": 0.06503974341096082,
-      "eval_dim_64_cosine_precision@3": 0.12820155254962115,
-      "eval_dim_64_cosine_precision@5": 0.0961929995816483,
-      "eval_dim_64_cosine_recall@1": 0.19704364802677451,
-      "eval_dim_64_cosine_recall@10": 0.6503974341096082,
-      "eval_dim_64_cosine_recall@3": 0.3846046576488635,
-      "eval_dim_64_cosine_recall@5": 0.48096499790824154,
-      "eval_dim_768_cosine_accuracy@1": 0.22116859573281272,
-      "eval_dim_768_cosine_accuracy@10": 0.6947427137079906,
-      "eval_dim_768_cosine_accuracy@3": 0.4196067494073351,
-      "eval_dim_768_cosine_accuracy@5": 0.5128991772416678,
-      "eval_dim_768_cosine_map@100": 0.3609325312259912,
-      "eval_dim_768_cosine_mrr@10": 0.3508835189353944,
-      "eval_dim_768_cosine_ndcg@10": 0.4315716703139609,
-      "eval_dim_768_cosine_precision@1": 0.22116859573281272,
-      "eval_dim_768_cosine_precision@10": 0.06947427137079905,
-      "eval_dim_768_cosine_precision@3": 0.13986891646911168,
-      "eval_dim_768_cosine_precision@5": 0.10257983544833356,
-      "eval_dim_768_cosine_recall@1": 0.22116859573281272,
-      "eval_dim_768_cosine_recall@10": 0.6947427137079906,
-      "eval_dim_768_cosine_recall@3": 0.4196067494073351,
-      "eval_dim_768_cosine_recall@5": 0.5128991772416678,
-      "eval_loss": 0.30127617716789246,
-      "eval_runtime": 4760.4689,
-      "eval_samples_per_second": 1.507,
-      "eval_sequential_score": 0.399385228043632,
-      "eval_steps_per_second": 0.019,
-      "step": 60
+      "eval_dim_128_cosine_accuracy@1": 0.19871705480407195,
+      "eval_dim_128_cosine_accuracy@10": 0.6681076558360062,
+      "eval_dim_128_cosine_accuracy@3": 0.3995258680797657,
+      "eval_dim_128_cosine_accuracy@5": 0.4922604936549993,
+      "eval_dim_128_cosine_map@100": 0.33912660774009473,
+      "eval_dim_128_cosine_mrr@10": 0.32818611337994835,
+      "eval_dim_128_cosine_ndcg@10": 0.40810554118165804,
+      "eval_dim_128_cosine_precision@1": 0.19871705480407195,
+      "eval_dim_128_cosine_precision@10": 0.06681076558360062,
+      "eval_dim_128_cosine_precision@3": 0.1331752893599219,
+      "eval_dim_128_cosine_precision@5": 0.09845209873099986,
+      "eval_dim_128_cosine_recall@1": 0.19871705480407195,
+      "eval_dim_128_cosine_recall@10": 0.6681076558360062,
+      "eval_dim_128_cosine_recall@3": 0.3995258680797657,
+      "eval_dim_128_cosine_recall@5": 0.4922604936549993,
+      "eval_dim_256_cosine_accuracy@1": 0.19983265932227026,
+      "eval_dim_256_cosine_accuracy@10": 0.6777297448054664,
+      "eval_dim_256_cosine_accuracy@3": 0.3998047692093153,
+      "eval_dim_256_cosine_accuracy@5": 0.4954678566448194,
+      "eval_dim_256_cosine_map@100": 0.34191237247567885,
+      "eval_dim_256_cosine_mrr@10": 0.33120438804443747,
+      "eval_dim_256_cosine_ndcg@10": 0.4125662992991588,
+      "eval_dim_256_cosine_precision@1": 0.19983265932227026,
+      "eval_dim_256_cosine_precision@10": 0.06777297448054664,
+      "eval_dim_256_cosine_precision@3": 0.13326825640310508,
+      "eval_dim_256_cosine_precision@5": 0.09909357132896388,
+      "eval_dim_256_cosine_recall@1": 0.19983265932227026,
+      "eval_dim_256_cosine_recall@10": 0.6777297448054664,
+      "eval_dim_256_cosine_recall@3": 0.3998047692093153,
+      "eval_dim_256_cosine_recall@5": 0.4954678566448194,
+      "eval_dim_512_cosine_accuracy@1": 0.20945474829173058,
+      "eval_dim_512_cosine_accuracy@10": 0.6849811741737554,
+      "eval_dim_512_cosine_accuracy@3": 0.41040301213219915,
+      "eval_dim_512_cosine_accuracy@5": 0.5059266490029285,
+      "eval_dim_512_cosine_map@100": 0.3513161520991861,
+      "eval_dim_512_cosine_mrr@10": 0.340864051193851,
+      "eval_dim_512_cosine_ndcg@10": 0.42175585614805716,
+      "eval_dim_512_cosine_precision@1": 0.20945474829173058,
+      "eval_dim_512_cosine_precision@10": 0.06849811741737553,
+      "eval_dim_512_cosine_precision@3": 0.13680100404406637,
+      "eval_dim_512_cosine_precision@5": 0.1011853298005857,
+      "eval_dim_512_cosine_recall@1": 0.20945474829173058,
+      "eval_dim_512_cosine_recall@10": 0.6849811741737554,
+      "eval_dim_512_cosine_recall@3": 0.41040301213219915,
+      "eval_dim_512_cosine_recall@5": 0.5059266490029285,
+      "eval_dim_64_cosine_accuracy@1": 0.18923441639938643,
+      "eval_dim_64_cosine_accuracy@10": 0.6431460047413192,
+      "eval_dim_64_cosine_accuracy@3": 0.38056059127039465,
+      "eval_dim_64_cosine_accuracy@5": 0.4667410403012132,
+      "eval_dim_64_cosine_map@100": 0.32516177346707625,
+      "eval_dim_64_cosine_mrr@10": 0.3137617785923441,
+      "eval_dim_64_cosine_ndcg@10": 0.39102781421972704,
+      "eval_dim_64_cosine_precision@1": 0.18923441639938643,
+      "eval_dim_64_cosine_precision@10": 0.06431460047413191,
+      "eval_dim_64_cosine_precision@3": 0.12685353042346487,
+      "eval_dim_64_cosine_precision@5": 0.09334820806024266,
+      "eval_dim_64_cosine_recall@1": 0.18923441639938643,
+      "eval_dim_64_cosine_recall@10": 0.6431460047413192,
+      "eval_dim_64_cosine_recall@3": 0.38056059127039465,
+      "eval_dim_64_cosine_recall@5": 0.4667410403012132,
+      "eval_dim_768_cosine_accuracy@1": 0.2069446381257844,
+      "eval_dim_768_cosine_accuracy@10": 0.6877701854692512,
+      "eval_dim_768_cosine_accuracy@3": 0.41110026495607305,
+      "eval_dim_768_cosine_accuracy@5": 0.5078789569097755,
+      "eval_dim_768_cosine_map@100": 0.35016286957488246,
+      "eval_dim_768_cosine_mrr@10": 0.33975193293977246,
+      "eval_dim_768_cosine_ndcg@10": 0.42158855710438475,
+      "eval_dim_768_cosine_precision@1": 0.2069446381257844,
+      "eval_dim_768_cosine_precision@10": 0.06877701854692511,
+      "eval_dim_768_cosine_precision@3": 0.13703342165202437,
+      "eval_dim_768_cosine_precision@5": 0.1015757913819551,
+      "eval_dim_768_cosine_recall@1": 0.2069446381257844,
+      "eval_dim_768_cosine_recall@10": 0.6877701854692512,
+      "eval_dim_768_cosine_recall@3": 0.41110026495607305,
+      "eval_dim_768_cosine_recall@5": 0.5078789569097755,
+      "eval_loss": 0.35814541578292847,
+      "eval_runtime": 5308.0606,
+      "eval_samples_per_second": 1.351,
+      "eval_sequential_score": 0.39102781421972704,
+      "eval_steps_per_second": 0.014,
+      "step": 48
     },
     {
       "epoch": 2.0,
-      "grad_norm": 1582530.375,
-      "learning_rate": 2.655769864163684e-05,
-      "loss": 0.2399,
-      "step": 120
+      "grad_norm": 1227518.25,
+      "learning_rate": 2.65047365119413e-05,
+      "loss": 0.3058,
+      "step": 96
     },
     {
       "epoch": 2.0,
-      "eval_dim_128_cosine_accuracy@1": 0.3225491563240831,
-      "eval_dim_128_cosine_accuracy@10": 0.7071538139729466,
-      "eval_dim_128_cosine_accuracy@3": 0.46102356714544696,
-      "eval_dim_128_cosine_accuracy@5": 0.53772137777158,
-      "eval_dim_128_cosine_map@100": 0.43211454127792864,
-      "eval_dim_128_cosine_mrr@10": 0.42227551225947657,
-      "eval_dim_128_cosine_ndcg@10": 0.488553211966058,
-      "eval_dim_128_cosine_precision@1": 0.3225491563240831,
-      "eval_dim_128_cosine_precision@10": 0.07071538139729466,
-      "eval_dim_128_cosine_precision@3": 0.15367452238181564,
-      "eval_dim_128_cosine_precision@5": 0.107544275554316,
-      "eval_dim_128_cosine_recall@1": 0.3225491563240831,
-      "eval_dim_128_cosine_recall@10": 0.7071538139729466,
-      "eval_dim_128_cosine_recall@3": 0.46102356714544696,
-      "eval_dim_128_cosine_recall@5": 0.53772137777158,
-      "eval_dim_256_cosine_accuracy@1": 0.32854553060939895,
-      "eval_dim_256_cosine_accuracy@10": 0.7131501882582625,
-      "eval_dim_256_cosine_accuracy@3": 0.464928182959141,
-      "eval_dim_256_cosine_accuracy@5": 0.5421837958443732,
-      "eval_dim_256_cosine_map@100": 0.4374891918057777,
-      "eval_dim_256_cosine_mrr@10": 0.42779830799981505,
-      "eval_dim_256_cosine_ndcg@10": 0.49410729071963905,
-      "eval_dim_256_cosine_precision@1": 0.32854553060939895,
-      "eval_dim_256_cosine_precision@10": 0.07131501882582625,
-      "eval_dim_256_cosine_precision@3": 0.1549760609863803,
-      "eval_dim_256_cosine_precision@5": 0.10843675916887462,
-      "eval_dim_256_cosine_recall@1": 0.32854553060939895,
-      "eval_dim_256_cosine_recall@10": 0.7131501882582625,
-      "eval_dim_256_cosine_recall@3": 0.464928182959141,
-      "eval_dim_256_cosine_recall@5": 0.5421837958443732,
-      "eval_dim_512_cosine_accuracy@1": 0.33286849811741737,
-      "eval_dim_512_cosine_accuracy@10": 0.7142657927764607,
-      "eval_dim_512_cosine_accuracy@3": 0.46966950216148373,
-      "eval_dim_512_cosine_accuracy@5": 0.5445544554455446,
-      "eval_dim_512_cosine_map@100": 0.4410220731771345,
-      "eval_dim_512_cosine_mrr@10": 0.4314777000838923,
-      "eval_dim_512_cosine_ndcg@10": 0.4972638536904392,
-      "eval_dim_512_cosine_precision@1": 0.33286849811741737,
-      "eval_dim_512_cosine_precision@10": 0.07142657927764608,
-      "eval_dim_512_cosine_precision@3": 0.15655650072049457,
-      "eval_dim_512_cosine_precision@5": 0.10891089108910891,
-      "eval_dim_512_cosine_recall@1": 0.33286849811741737,
-      "eval_dim_512_cosine_recall@10": 0.7142657927764607,
-      "eval_dim_512_cosine_recall@3": 0.46966950216148373,
-      "eval_dim_512_cosine_recall@5": 0.5445544554455446,
-      "eval_dim_64_cosine_accuracy@1": 0.306233440245433,
-      "eval_dim_64_cosine_accuracy@10": 0.6906986473295217,
-      "eval_dim_64_cosine_accuracy@3": 0.44205829033607585,
-      "eval_dim_64_cosine_accuracy@5": 0.5208478594338307,
-      "eval_dim_64_cosine_map@100": 0.416048690892794,
-      "eval_dim_64_cosine_mrr@10": 0.4057003738603239,
-      "eval_dim_64_cosine_ndcg@10": 0.47195343698452114,
-      "eval_dim_64_cosine_precision@1": 0.306233440245433,
-      "eval_dim_64_cosine_precision@10": 0.06906986473295217,
-      "eval_dim_64_cosine_precision@3": 0.1473527634453586,
-      "eval_dim_64_cosine_precision@5": 0.10416957188676614,
-      "eval_dim_64_cosine_recall@1": 0.306233440245433,
-      "eval_dim_64_cosine_recall@10": 0.6906986473295217,
-      "eval_dim_64_cosine_recall@3": 0.44205829033607585,
-      "eval_dim_64_cosine_recall@5": 0.5208478594338307,
-      "eval_dim_768_cosine_accuracy@1": 0.33412355320039044,
-      "eval_dim_768_cosine_accuracy@10": 0.7152419467298843,
-      "eval_dim_768_cosine_accuracy@3": 0.47106400780923163,
-      "eval_dim_768_cosine_accuracy@5": 0.5448333565750941,
-      "eval_dim_768_cosine_map@100": 0.44207831994901653,
-      "eval_dim_768_cosine_mrr@10": 0.4325288363846451,
-      "eval_dim_768_cosine_ndcg@10": 0.49828039957821635,
-      "eval_dim_768_cosine_precision@1": 0.33412355320039044,
-      "eval_dim_768_cosine_precision@10": 0.07152419467298843,
-      "eval_dim_768_cosine_precision@3": 0.15702133593641054,
-      "eval_dim_768_cosine_precision@5": 0.10896667131501885,
-      "eval_dim_768_cosine_recall@1": 0.33412355320039044,
-      "eval_dim_768_cosine_recall@10": 0.7152419467298843,
-      "eval_dim_768_cosine_recall@3": 0.47106400780923163,
-      "eval_dim_768_cosine_recall@5": 0.5448333565750941,
-      "eval_loss": 0.18898583948612213,
-      "eval_runtime": 4757.4,
-      "eval_samples_per_second": 1.508,
-      "eval_sequential_score": 0.47195343698452114,
-      "eval_steps_per_second": 0.019,
-      "step": 120
+      "eval_dim_128_cosine_accuracy@1": 0.319341793334263,
+      "eval_dim_128_cosine_accuracy@10": 0.7118951331752894,
+      "eval_dim_128_cosine_accuracy@3": 0.45670059963742854,
+      "eval_dim_128_cosine_accuracy@5": 0.5373030260772557,
+      "eval_dim_128_cosine_map@100": 0.43025589198069303,
+      "eval_dim_128_cosine_mrr@10": 0.4206424465384167,
+      "eval_dim_128_cosine_ndcg@10": 0.48839839347709046,
+      "eval_dim_128_cosine_precision@1": 0.319341793334263,
+      "eval_dim_128_cosine_precision@10": 0.07118951331752894,
+      "eval_dim_128_cosine_precision@3": 0.15223353321247615,
+      "eval_dim_128_cosine_precision@5": 0.10746060521545113,
+      "eval_dim_128_cosine_recall@1": 0.319341793334263,
+      "eval_dim_128_cosine_recall@10": 0.7118951331752894,
+      "eval_dim_128_cosine_recall@3": 0.45670059963742854,
+      "eval_dim_128_cosine_recall@5": 0.5373030260772557,
+      "eval_dim_256_cosine_accuracy@1": 0.3284060800446242,
+      "eval_dim_256_cosine_accuracy@10": 0.709524473574118,
+      "eval_dim_256_cosine_accuracy@3": 0.4618602705340957,
+      "eval_dim_256_cosine_accuracy@5": 0.5356296192999581,
+      "eval_dim_256_cosine_map@100": 0.4364935437376769,
+      "eval_dim_256_cosine_mrr@10": 0.42624382156525586,
+      "eval_dim_256_cosine_ndcg@10": 0.49204715409513783,
+      "eval_dim_256_cosine_precision@1": 0.3284060800446242,
+      "eval_dim_256_cosine_precision@10": 0.07095244735741181,
+      "eval_dim_256_cosine_precision@3": 0.15395342351136523,
+      "eval_dim_256_cosine_precision@5": 0.10712592385999165,
+      "eval_dim_256_cosine_recall@1": 0.3284060800446242,
+      "eval_dim_256_cosine_recall@10": 0.709524473574118,
+      "eval_dim_256_cosine_recall@3": 0.4618602705340957,
+      "eval_dim_256_cosine_recall@5": 0.5356296192999581,
+      "eval_dim_512_cosine_accuracy@1": 0.33663366336633666,
+      "eval_dim_512_cosine_accuracy@10": 0.7125923859991633,
+      "eval_dim_512_cosine_accuracy@3": 0.46841444707851065,
+      "eval_dim_512_cosine_accuracy@5": 0.5441361037512202,
+      "eval_dim_512_cosine_map@100": 0.4433337740824084,
+      "eval_dim_512_cosine_mrr@10": 0.43360332512124483,
+      "eval_dim_512_cosine_ndcg@10": 0.4984772586608768,
+      "eval_dim_512_cosine_precision@1": 0.33663366336633666,
+      "eval_dim_512_cosine_precision@10": 0.07125923859991633,
+      "eval_dim_512_cosine_precision@3": 0.15613814902617024,
+      "eval_dim_512_cosine_precision@5": 0.10882722075024404,
+      "eval_dim_512_cosine_recall@1": 0.33663366336633666,
+      "eval_dim_512_cosine_recall@10": 0.7125923859991633,
+      "eval_dim_512_cosine_recall@3": 0.46841444707851065,
+      "eval_dim_512_cosine_recall@5": 0.5441361037512202,
+      "eval_dim_64_cosine_accuracy@1": 0.3026077255612885,
+      "eval_dim_64_cosine_accuracy@10": 0.689861943940873,
+      "eval_dim_64_cosine_accuracy@3": 0.4437316971133733,
+      "eval_dim_64_cosine_accuracy@5": 0.519313903221308,
+      "eval_dim_64_cosine_map@100": 0.4140292225757397,
+      "eval_dim_64_cosine_mrr@10": 0.4037352829850398,
+      "eval_dim_64_cosine_ndcg@10": 0.47036063500166436,
+      "eval_dim_64_cosine_precision@1": 0.3026077255612885,
+      "eval_dim_64_cosine_precision@10": 0.0689861943940873,
+      "eval_dim_64_cosine_precision@3": 0.14791056570445774,
+      "eval_dim_64_cosine_precision@5": 0.10386278064426162,
+      "eval_dim_64_cosine_recall@1": 0.3026077255612885,
+      "eval_dim_64_cosine_recall@10": 0.689861943940873,
+      "eval_dim_64_cosine_recall@3": 0.4437316971133733,
+      "eval_dim_64_cosine_recall@5": 0.519313903221308,
+      "eval_dim_768_cosine_accuracy@1": 0.3332868498117417,
+      "eval_dim_768_cosine_accuracy@10": 0.7166364523776322,
+      "eval_dim_768_cosine_accuracy@3": 0.46841444707851065,
+      "eval_dim_768_cosine_accuracy@5": 0.546227862222842,
+      "eval_dim_768_cosine_map@100": 0.44212169088004494,
+      "eval_dim_768_cosine_mrr@10": 0.4324117975177812,
+      "eval_dim_768_cosine_ndcg@10": 0.498508349299699,
+      "eval_dim_768_cosine_precision@1": 0.3332868498117417,
+      "eval_dim_768_cosine_precision@10": 0.07166364523776322,
+      "eval_dim_768_cosine_precision@3": 0.1561381490261702,
+      "eval_dim_768_cosine_precision@5": 0.1092455724445684,
+      "eval_dim_768_cosine_recall@1": 0.3332868498117417,
+      "eval_dim_768_cosine_recall@10": 0.7166364523776322,
+      "eval_dim_768_cosine_recall@3": 0.46841444707851065,
+      "eval_dim_768_cosine_recall@5": 0.546227862222842,
+      "eval_loss": 0.2100292444229126,
+      "eval_runtime": 5303.3966,
+      "eval_samples_per_second": 1.352,
+      "eval_sequential_score": 0.47036063500166436,
+      "eval_steps_per_second": 0.014,
+      "step": 96
     },
     {
       "epoch": 3.0,
-      "grad_norm": 792357.3125,
-      "learning_rate": 1.8730348307472828e-05,
-      "loss": 0.1032,
-      "step": 180
+      "grad_norm": 751449.875,
+      "learning_rate": 1.8611820248535433e-05,
+      "loss": 0.1395,
+      "step": 144
     },
     {
       "epoch": 3.0,
-      "eval_dim_128_cosine_accuracy@1": 0.3468135545948961,
-      "eval_dim_128_cosine_accuracy@10": 0.7173337052015061,
-      "eval_dim_128_cosine_accuracy@3": 0.46520708408869055,
-      "eval_dim_128_cosine_accuracy@5": 0.5407892901966252,
-      "eval_dim_128_cosine_map@100": 0.447677926706174,
-      "eval_dim_128_cosine_mrr@10": 0.43798207285517343,
-      "eval_dim_128_cosine_ndcg@10": 0.5026334575249692,
-      "eval_dim_128_cosine_precision@1": 0.3468135545948961,
-      "eval_dim_128_cosine_precision@10": 0.07173337052015061,
-      "eval_dim_128_cosine_precision@3": 0.15506902802956352,
-      "eval_dim_128_cosine_precision@5": 0.10815785803932505,
-      "eval_dim_128_cosine_recall@1": 0.3468135545948961,
-      "eval_dim_128_cosine_recall@10": 0.7173337052015061,
-      "eval_dim_128_cosine_recall@3": 0.46520708408869055,
-      "eval_dim_128_cosine_recall@5": 0.5407892901966252,
-      "eval_dim_256_cosine_accuracy@1": 0.35043926927904057,
-      "eval_dim_256_cosine_accuracy@10": 0.7156602984242086,
-      "eval_dim_256_cosine_accuracy@3": 0.4700878538558081,
-      "eval_dim_256_cosine_accuracy@5": 0.544972807139869,
-      "eval_dim_256_cosine_map@100": 0.4507378853512356,
-      "eval_dim_256_cosine_mrr@10": 0.44098468699988747,
-      "eval_dim_256_cosine_ndcg@10": 0.5046872072161374,
-      "eval_dim_256_cosine_precision@1": 0.35043926927904057,
-      "eval_dim_256_cosine_precision@10": 0.07156602984242087,
-      "eval_dim_256_cosine_precision@3": 0.15669595128526936,
-      "eval_dim_256_cosine_precision@5": 0.1089945614279738,
-      "eval_dim_256_cosine_recall@1": 0.35043926927904057,
-      "eval_dim_256_cosine_recall@10": 0.7156602984242086,
-      "eval_dim_256_cosine_recall@3": 0.4700878538558081,
-      "eval_dim_256_cosine_recall@5": 0.544972807139869,
-      "eval_dim_512_cosine_accuracy@1": 0.35169432436201364,
-      "eval_dim_512_cosine_accuracy@10": 0.7231906289220471,
-      "eval_dim_512_cosine_accuracy@3": 0.4700878538558081,
-      "eval_dim_512_cosine_accuracy@5": 0.5448333565750941,
-      "eval_dim_512_cosine_map@100": 0.45212818188900633,
-      "eval_dim_512_cosine_mrr@10": 0.44279272997722396,
-      "eval_dim_512_cosine_ndcg@10": 0.5076751608428753,
-      "eval_dim_512_cosine_precision@1": 0.35169432436201364,
-      "eval_dim_512_cosine_precision@10": 0.07231906289220472,
-      "eval_dim_512_cosine_precision@3": 0.15669595128526936,
-      "eval_dim_512_cosine_precision@5": 0.10896667131501885,
-      "eval_dim_512_cosine_recall@1": 0.35169432436201364,
-      "eval_dim_512_cosine_recall@10": 0.7231906289220471,
-      "eval_dim_512_cosine_recall@3": 0.4700878538558081,
-      "eval_dim_512_cosine_recall@5": 0.5448333565750941,
-      "eval_dim_64_cosine_accuracy@1": 0.3356575094129131,
-      "eval_dim_64_cosine_accuracy@10": 0.6976711755682611,
-      "eval_dim_64_cosine_accuracy@3": 0.4523776321294101,
-      "eval_dim_64_cosine_accuracy@5": 0.5307488495328406,
-      "eval_dim_64_cosine_map@100": 0.43559811808588345,
-      "eval_dim_64_cosine_mrr@10": 0.42538852476797007,
-      "eval_dim_64_cosine_ndcg@10": 0.48852868771634256,
-      "eval_dim_64_cosine_precision@1": 0.3356575094129131,
-      "eval_dim_64_cosine_precision@10": 0.0697671175568261,
-      "eval_dim_64_cosine_precision@3": 0.1507925440431367,
-      "eval_dim_64_cosine_precision@5": 0.10614976990656813,
-      "eval_dim_64_cosine_recall@1": 0.3356575094129131,
-      "eval_dim_64_cosine_recall@10": 0.6976711755682611,
-      "eval_dim_64_cosine_recall@3": 0.4523776321294101,
-      "eval_dim_64_cosine_recall@5": 0.5307488495328406,
-      "eval_dim_768_cosine_accuracy@1": 0.3568539952586808,
-      "eval_dim_768_cosine_accuracy@10": 0.7237484311811463,
-      "eval_dim_768_cosine_accuracy@3": 0.473434667410403,
-      "eval_dim_768_cosine_accuracy@5": 0.5530609398968066,
-      "eval_dim_768_cosine_map@100": 0.456489701012935,
-      "eval_dim_768_cosine_mrr@10": 0.44698465822437466,
-      "eval_dim_768_cosine_ndcg@10": 0.5110884733352402,
-      "eval_dim_768_cosine_precision@1": 0.3568539952586808,
-      "eval_dim_768_cosine_precision@10": 0.07237484311811462,
-      "eval_dim_768_cosine_precision@3": 0.15781155580346765,
-      "eval_dim_768_cosine_precision@5": 0.11061218797936133,
-      "eval_dim_768_cosine_recall@1": 0.3568539952586808,
-      "eval_dim_768_cosine_recall@10": 0.7237484311811463,
-      "eval_dim_768_cosine_recall@3": 0.473434667410403,
-      "eval_dim_768_cosine_recall@5": 0.5530609398968066,
-      "eval_loss": 0.15947215259075165,
-      "eval_runtime": 4772.3881,
-      "eval_samples_per_second": 1.503,
-      "eval_sequential_score": 0.48852868771634256,
-      "eval_steps_per_second": 0.019,
-      "step": 180
+      "eval_dim_128_cosine_accuracy@1": 0.3420722353925533,
+      "eval_dim_128_cosine_accuracy@10": 0.7111978803514154,
+      "eval_dim_128_cosine_accuracy@3": 0.4580951052851764,
+      "eval_dim_128_cosine_accuracy@5": 0.5392553339841026,
+      "eval_dim_128_cosine_map@100": 0.44322012540075556,
+      "eval_dim_128_cosine_mrr@10": 0.43320030192154113,
+      "eval_dim_128_cosine_ndcg@10": 0.49761764615839976,
+      "eval_dim_128_cosine_precision@1": 0.3420722353925533,
+      "eval_dim_128_cosine_precision@10": 0.07111978803514155,
+      "eval_dim_128_cosine_precision@3": 0.15269836842839213,
+      "eval_dim_128_cosine_precision@5": 0.10785106679682054,
+      "eval_dim_128_cosine_recall@1": 0.3420722353925533,
+      "eval_dim_128_cosine_recall@10": 0.7111978803514154,
+      "eval_dim_128_cosine_recall@3": 0.4580951052851764,
+      "eval_dim_128_cosine_recall@5": 0.5392553339841026,
+      "eval_dim_256_cosine_accuracy@1": 0.34876586250174313,
+      "eval_dim_256_cosine_accuracy@10": 0.7184493097197043,
+      "eval_dim_256_cosine_accuracy@3": 0.4705062055501325,
+      "eval_dim_256_cosine_accuracy@5": 0.5427415981034723,
+      "eval_dim_256_cosine_map@100": 0.4497903349951724,
+      "eval_dim_256_cosine_mrr@10": 0.4402565004991461,
+      "eval_dim_256_cosine_ndcg@10": 0.5046482555536308,
+      "eval_dim_256_cosine_precision@1": 0.34876586250174313,
+      "eval_dim_256_cosine_precision@10": 0.07184493097197044,
+      "eval_dim_256_cosine_precision@3": 0.15683540185004413,
+      "eval_dim_256_cosine_precision@5": 0.10854831962069447,
+      "eval_dim_256_cosine_recall@1": 0.34876586250174313,
+      "eval_dim_256_cosine_recall@10": 0.7184493097197043,
+      "eval_dim_256_cosine_recall@3": 0.4705062055501325,
+      "eval_dim_256_cosine_recall@5": 0.5427415981034723,
+      "eval_dim_512_cosine_accuracy@1": 0.3525310277506624,
+      "eval_dim_512_cosine_accuracy@10": 0.7208199693208758,
+      "eval_dim_512_cosine_accuracy@3": 0.47190071119788035,
+      "eval_dim_512_cosine_accuracy@5": 0.5545948961093292,
+      "eval_dim_512_cosine_map@100": 0.4537559279114653,
+      "eval_dim_512_cosine_mrr@10": 0.44413881528997956,
+      "eval_dim_512_cosine_ndcg@10": 0.5083169852417319,
+      "eval_dim_512_cosine_precision@1": 0.3525310277506624,
+      "eval_dim_512_cosine_precision@10": 0.07208199693208758,
+      "eval_dim_512_cosine_precision@3": 0.1573002370659601,
+      "eval_dim_512_cosine_precision@5": 0.11091897922186586,
+      "eval_dim_512_cosine_recall@1": 0.3525310277506624,
+      "eval_dim_512_cosine_recall@10": 0.7208199693208758,
+      "eval_dim_512_cosine_recall@3": 0.47190071119788035,
+      "eval_dim_512_cosine_recall@5": 0.5545948961093292,
+      "eval_dim_64_cosine_accuracy@1": 0.3289638823037233,
+      "eval_dim_64_cosine_accuracy@10": 0.6980895272625854,
+      "eval_dim_64_cosine_accuracy@3": 0.447217961232743,
+      "eval_dim_64_cosine_accuracy@5": 0.5223818156463533,
+      "eval_dim_64_cosine_map@100": 0.4300263980717357,
+      "eval_dim_64_cosine_mrr@10": 0.4197134401568931,
+      "eval_dim_64_cosine_ndcg@10": 0.48411999159781,
+      "eval_dim_64_cosine_precision@1": 0.3289638823037233,
+      "eval_dim_64_cosine_precision@10": 0.06980895272625855,
+      "eval_dim_64_cosine_precision@3": 0.14907265374424766,
+      "eval_dim_64_cosine_precision@5": 0.10447636312927068,
+      "eval_dim_64_cosine_recall@1": 0.3289638823037233,
+      "eval_dim_64_cosine_recall@10": 0.6980895272625854,
+      "eval_dim_64_cosine_recall@3": 0.447217961232743,
+      "eval_dim_64_cosine_recall@5": 0.5223818156463533,
+      "eval_dim_768_cosine_accuracy@1": 0.35392553339841026,
+      "eval_dim_768_cosine_accuracy@10": 0.7202621670617766,
+      "eval_dim_768_cosine_accuracy@3": 0.47078510667968204,
+      "eval_dim_768_cosine_accuracy@5": 0.5458095105285177,
+      "eval_dim_768_cosine_map@100": 0.45349669058599684,
+      "eval_dim_768_cosine_mrr@10": 0.4438007583454529,
+      "eval_dim_768_cosine_ndcg@10": 0.5077758391371117,
+      "eval_dim_768_cosine_precision@1": 0.35392553339841026,
+      "eval_dim_768_cosine_precision@10": 0.07202621670617766,
+      "eval_dim_768_cosine_precision@3": 0.15692836889322734,
+      "eval_dim_768_cosine_precision@5": 0.10916190210570352,
+      "eval_dim_768_cosine_recall@1": 0.35392553339841026,
+      "eval_dim_768_cosine_recall@10": 0.7202621670617766,
+      "eval_dim_768_cosine_recall@3": 0.47078510667968204,
+      "eval_dim_768_cosine_recall@5": 0.5458095105285177,
+      "eval_loss": 0.1820719689130783,
+      "eval_runtime": 5286.0245,
+      "eval_samples_per_second": 1.357,
+      "eval_sequential_score": 0.48411999159781,
+      "eval_steps_per_second": 0.014,
+      "step": 144
     },
     {
       "epoch": 4.0,
-      "grad_norm": 528732.6875,
-      "learning_rate": 9.478131709729831e-06,
-      "loss": 0.0562,
-      "step": 240
+      "grad_norm": 609946.5,
+      "learning_rate": 9.328128849810832e-06,
+      "loss": 0.0828,
+      "step": 192
     },
     {
       "epoch": 4.0,
-      "eval_dim_128_cosine_accuracy@1": 0.35016036814949103,
-      "eval_dim_128_cosine_accuracy@10": 0.71803095802538,
+      "eval_dim_128_cosine_accuracy@1": 0.34876586250174313,
+      "eval_dim_128_cosine_accuracy@10": 0.7169153535071817,
+      "eval_dim_128_cosine_accuracy@3": 0.4691116999023846,
+      "eval_dim_128_cosine_accuracy@5": 0.5466462139171664,
+      "eval_dim_128_cosine_map@100": 0.44992557552198303,
+      "eval_dim_128_cosine_mrr@10": 0.4401795813384156,
+      "eval_dim_128_cosine_ndcg@10": 0.504316835829398,
+      "eval_dim_128_cosine_precision@1": 0.34876586250174313,
+      "eval_dim_128_cosine_precision@10": 0.07169153535071818,
+      "eval_dim_128_cosine_precision@3": 0.15637056663412818,
+      "eval_dim_128_cosine_precision@5": 0.10932924278343327,
+      "eval_dim_128_cosine_recall@1": 0.34876586250174313,
+      "eval_dim_128_cosine_recall@10": 0.7169153535071817,
+      "eval_dim_128_cosine_recall@3": 0.4691116999023846,
+      "eval_dim_128_cosine_recall@5": 0.5466462139171664,
+      "eval_dim_256_cosine_accuracy@1": 0.3514154232324641,
+      "eval_dim_256_cosine_accuracy@10": 0.7198438153674522,
+      "eval_dim_256_cosine_accuracy@3": 0.47482917305815087,
+      "eval_dim_256_cosine_accuracy@5": 0.5479012690001395,
+      "eval_dim_256_cosine_map@100": 0.45302713856879157,
+      "eval_dim_256_cosine_mrr@10": 0.44321866291257,
+      "eval_dim_256_cosine_ndcg@10": 0.5073024414696454,
+      "eval_dim_256_cosine_precision@1": 0.3514154232324641,
+      "eval_dim_256_cosine_precision@10": 0.07198438153674523,
+      "eval_dim_256_cosine_precision@3": 0.15827639101938362,
+      "eval_dim_256_cosine_precision@5": 0.10958025380002788,
+      "eval_dim_256_cosine_recall@1": 0.3514154232324641,
+      "eval_dim_256_cosine_recall@10": 0.7198438153674522,
+      "eval_dim_256_cosine_recall@3": 0.47482917305815087,
+      "eval_dim_256_cosine_recall@5": 0.5479012690001395,
+      "eval_dim_512_cosine_accuracy@1": 0.35629619299958165,
+      "eval_dim_512_cosine_accuracy@10": 0.7229117277924976,
+      "eval_dim_512_cosine_accuracy@3": 0.46966950216148373,
+      "eval_dim_512_cosine_accuracy@5": 0.5479012690001395,
+      "eval_dim_512_cosine_map@100": 0.4551128486578315,
+      "eval_dim_512_cosine_mrr@10": 0.4454170567962228,
+      "eval_dim_512_cosine_ndcg@10": 0.5095844549678528,
+      "eval_dim_512_cosine_precision@1": 0.35629619299958165,
+      "eval_dim_512_cosine_precision@10": 0.07229117277924975,
+      "eval_dim_512_cosine_precision@3": 0.1565565007204946,
+      "eval_dim_512_cosine_precision@5": 0.10958025380002788,
+      "eval_dim_512_cosine_recall@1": 0.35629619299958165,
+      "eval_dim_512_cosine_recall@10": 0.7229117277924976,
+      "eval_dim_512_cosine_recall@3": 0.46966950216148373,
+      "eval_dim_512_cosine_recall@5": 0.5479012690001395,
+      "eval_dim_64_cosine_accuracy@1": 0.33649421280156183,
+      "eval_dim_64_cosine_accuracy@10": 0.7029702970297029,
+      "eval_dim_64_cosine_accuracy@3": 0.45432994003625715,
+      "eval_dim_64_cosine_accuracy@5": 0.5300515967089666,
+      "eval_dim_64_cosine_map@100": 0.43606205857055896,
+      "eval_dim_64_cosine_mrr@10": 0.4258440632795675,
+      "eval_dim_64_cosine_ndcg@10": 0.48996806351717825,
+      "eval_dim_64_cosine_precision@1": 0.33649421280156183,
+      "eval_dim_64_cosine_precision@10": 0.0702970297029703,
+      "eval_dim_64_cosine_precision@3": 0.15144331334541905,
+      "eval_dim_64_cosine_precision@5": 0.10601031934179334,
+      "eval_dim_64_cosine_recall@1": 0.33649421280156183,
+      "eval_dim_64_cosine_recall@10": 0.7029702970297029,
+      "eval_dim_64_cosine_recall@3": 0.45432994003625715,
+      "eval_dim_64_cosine_recall@5": 0.5300515967089666,
+      "eval_dim_768_cosine_accuracy@1": 0.356714544693906,
+      "eval_dim_768_cosine_accuracy@10": 0.7251429368288942,
+      "eval_dim_768_cosine_accuracy@3": 0.47315576628085343,
+      "eval_dim_768_cosine_accuracy@5": 0.5499930274717613,
+      "eval_dim_768_cosine_map@100": 0.4565653595396862,
+      "eval_dim_768_cosine_mrr@10": 0.44696689487862346,
+      "eval_dim_768_cosine_ndcg@10": 0.5113293441302604,
+      "eval_dim_768_cosine_precision@1": 0.356714544693906,
+      "eval_dim_768_cosine_precision@10": 0.07251429368288942,
+      "eval_dim_768_cosine_precision@3": 0.15771858876028447,
+      "eval_dim_768_cosine_precision@5": 0.10999860549435224,
+      "eval_dim_768_cosine_recall@1": 0.356714544693906,
+      "eval_dim_768_cosine_recall@10": 0.7251429368288942,
+      "eval_dim_768_cosine_recall@3": 0.47315576628085343,
+      "eval_dim_768_cosine_recall@5": 0.5499930274717613,
+      "eval_loss": 0.17269685864448547,
+      "eval_runtime": 5288.939,
+      "eval_samples_per_second": 1.356,
+      "eval_sequential_score": 0.48996806351717825,
+      "eval_steps_per_second": 0.014,
+      "step": 192
+    },
+    {
+      "epoch": 5.0,
+      "grad_norm": 373272.25,
+      "learning_rate": 2.2284601387084617e-06,
+      "loss": 0.0601,
+      "step": 240
+    },
+    {
+      "epoch": 5.0,
+      "eval_dim_128_cosine_accuracy@1": 0.345558499511923,
+      "eval_dim_128_cosine_accuracy@10": 0.7213777715799748,
       "eval_dim_128_cosine_accuracy@3": 0.466880490865988,
-      "eval_dim_128_cosine_accuracy@5": 0.5453911588341933,
-      "eval_dim_128_cosine_map@100": 0.45032612042753656,
-      "eval_dim_128_cosine_mrr@10": 0.44050070500007754,
-      "eval_dim_128_cosine_ndcg@10": 0.5047593926261945,
-      "eval_dim_128_cosine_precision@1": 0.35016036814949103,
-      "eval_dim_128_cosine_precision@10": 0.07180309580253802,
+      "eval_dim_128_cosine_accuracy@5": 0.5456700599637428,
+      "eval_dim_128_cosine_map@100": 0.44781889706523814,
+      "eval_dim_128_cosine_mrr@10": 0.4382224037293079,
+      "eval_dim_128_cosine_ndcg@10": 0.5037796273775794,
+      "eval_dim_128_cosine_precision@1": 0.345558499511923,
+      "eval_dim_128_cosine_precision@10": 0.07213777715799749,
       "eval_dim_128_cosine_precision@3": 0.15562683028866267,
-      "eval_dim_128_cosine_precision@5": 0.10907823176683866,
-      "eval_dim_128_cosine_recall@1": 0.35016036814949103,
-      "eval_dim_128_cosine_recall@10": 0.71803095802538,
+      "eval_dim_128_cosine_precision@5": 0.10913401199274857,
+      "eval_dim_128_cosine_recall@1": 0.345558499511923,
+      "eval_dim_128_cosine_recall@10": 0.7213777715799748,
       "eval_dim_128_cosine_recall@3": 0.466880490865988,
-      "eval_dim_128_cosine_recall@5": 0.5453911588341933,
-      "eval_dim_256_cosine_accuracy@1": 0.34932366476084226,
-      "eval_dim_256_cosine_accuracy@10": 0.7227722772277227,
-      "eval_dim_256_cosine_accuracy@3": 0.46660158973643845,
-      "eval_dim_256_cosine_accuracy@5": 0.5445544554455446,
-      "eval_dim_256_cosine_map@100": 0.4503116262429447,
-      "eval_dim_256_cosine_mrr@10": 0.44069499505282583,
-      "eval_dim_256_cosine_ndcg@10": 0.5059436181895386,
-      "eval_dim_256_cosine_precision@1": 0.34932366476084226,
-      "eval_dim_256_cosine_precision@10": 0.07227722772277227,
-      "eval_dim_256_cosine_precision@3": 0.15553386324547946,
-      "eval_dim_256_cosine_precision@5": 0.10891089108910891,
-      "eval_dim_256_cosine_recall@1": 0.34932366476084226,
-      "eval_dim_256_cosine_recall@10": 0.7227722772277227,
-      "eval_dim_256_cosine_recall@3": 0.46660158973643845,
-      "eval_dim_256_cosine_recall@5": 0.5445544554455446,
-      "eval_dim_512_cosine_accuracy@1": 0.3550411379166086,
-      "eval_dim_512_cosine_accuracy@10": 0.7234695300515968,
-      "eval_dim_512_cosine_accuracy@3": 0.4693906010319342,
-      "eval_dim_512_cosine_accuracy@5": 0.5484590712592387,
-      "eval_dim_512_cosine_map@100": 0.45428241375668854,
-      "eval_dim_512_cosine_mrr@10": 0.4447460339595335,
-      "eval_dim_512_cosine_ndcg@10": 0.5091968606671677,
-      "eval_dim_512_cosine_precision@1": 0.3550411379166086,
-      "eval_dim_512_cosine_precision@10": 0.07234695300515968,
-      "eval_dim_512_cosine_precision@3": 0.1564635336773114,
-      "eval_dim_512_cosine_precision@5": 0.10969181425184772,
-      "eval_dim_512_cosine_recall@1": 0.3550411379166086,
-      "eval_dim_512_cosine_recall@10": 0.7234695300515968,
-      "eval_dim_512_cosine_recall@3": 0.4693906010319342,
-      "eval_dim_512_cosine_recall@5": 0.5484590712592387,
-      "eval_dim_64_cosine_accuracy@1": 0.33760981731976014,
-      "eval_dim_64_cosine_accuracy@10": 0.7082694184911449,
-      "eval_dim_64_cosine_accuracy@3": 0.45349323664760843,
-      "eval_dim_64_cosine_accuracy@5": 0.5308883000976153,
-      "eval_dim_64_cosine_map@100": 0.4383018870756752,
-      "eval_dim_64_cosine_mrr@10": 0.4285745828103948,
-      "eval_dim_64_cosine_ndcg@10": 0.49330694449266793,
-      "eval_dim_64_cosine_precision@1": 0.33760981731976014,
-      "eval_dim_64_cosine_precision@10": 0.07082694184911449,
-      "eval_dim_64_cosine_precision@3": 0.1511644122158695,
-      "eval_dim_64_cosine_precision@5": 0.10617766001952307,
-      "eval_dim_64_cosine_recall@1": 0.33760981731976014,
-      "eval_dim_64_cosine_recall@10": 0.7082694184911449,
-      "eval_dim_64_cosine_recall@3": 0.45349323664760843,
-      "eval_dim_64_cosine_recall@5": 0.5308883000976153,
-      "eval_dim_768_cosine_accuracy@1": 0.36061916050760007,
-      "eval_dim_768_cosine_accuracy@10": 0.7238878817459211,
-      "eval_dim_768_cosine_accuracy@3": 0.478036536047971,
-      "eval_dim_768_cosine_accuracy@5": 0.5529214893320318,
-      "eval_dim_768_cosine_map@100": 0.45954079808905995,
-      "eval_dim_768_cosine_mrr@10": 0.44983946583793183,
-      "eval_dim_768_cosine_ndcg@10": 0.513288052386553,
-      "eval_dim_768_cosine_precision@1": 0.36061916050760007,
-      "eval_dim_768_cosine_precision@10": 0.07238878817459211,
-      "eval_dim_768_cosine_precision@3": 0.15934551201599031,
-      "eval_dim_768_cosine_precision@5": 0.11058429786640636,
-      "eval_dim_768_cosine_recall@1": 0.36061916050760007,
-      "eval_dim_768_cosine_recall@10": 0.7238878817459211,
-      "eval_dim_768_cosine_recall@3": 0.478036536047971,
-      "eval_dim_768_cosine_recall@5": 0.5529214893320318,
-      "eval_loss": 0.14603222906589508,
-      "eval_runtime": 4765.1388,
-      "eval_samples_per_second": 1.505,
-      "eval_sequential_score": 0.49330694449266793,
-      "eval_steps_per_second": 0.019,
+      "eval_dim_128_cosine_recall@5": 0.5456700599637428,
+      "eval_dim_256_cosine_accuracy@1": 0.3511365221029145,
+      "eval_dim_256_cosine_accuracy@10": 0.7241667828754706,
+      "eval_dim_256_cosine_accuracy@3": 0.47399246966950215,
+      "eval_dim_256_cosine_accuracy@5": 0.5540370938502301,
+      "eval_dim_256_cosine_map@100": 0.45382807101318545,
+      "eval_dim_256_cosine_mrr@10": 0.444310748982345,
+      "eval_dim_256_cosine_ndcg@10": 0.509228248886466,
+      "eval_dim_256_cosine_precision@1": 0.3511365221029145,
+      "eval_dim_256_cosine_precision@10": 0.07241667828754707,
+      "eval_dim_256_cosine_precision@3": 0.15799748988983406,
+      "eval_dim_256_cosine_precision@5": 0.11080741877004602,
+      "eval_dim_256_cosine_recall@1": 0.3511365221029145,
+      "eval_dim_256_cosine_recall@10": 0.7241667828754706,
+      "eval_dim_256_cosine_recall@3": 0.47399246966950215,
+      "eval_dim_256_cosine_recall@5": 0.5540370938502301,
+      "eval_dim_512_cosine_accuracy@1": 0.35545948961093293,
+      "eval_dim_512_cosine_accuracy@10": 0.7241667828754706,
+      "eval_dim_512_cosine_accuracy@3": 0.47622367870589877,
+      "eval_dim_512_cosine_accuracy@5": 0.5550132478036536,
+      "eval_dim_512_cosine_map@100": 0.4566662947498507,
+      "eval_dim_512_cosine_mrr@10": 0.4469697724299597,
+      "eval_dim_512_cosine_ndcg@10": 0.5112339641773941,
+      "eval_dim_512_cosine_precision@1": 0.35545948961093293,
+      "eval_dim_512_cosine_precision@10": 0.07241667828754705,
+      "eval_dim_512_cosine_precision@3": 0.1587412262352996,
+      "eval_dim_512_cosine_precision@5": 0.11100264956073073,
+      "eval_dim_512_cosine_recall@1": 0.35545948961093293,
+      "eval_dim_512_cosine_recall@10": 0.7241667828754706,
+      "eval_dim_512_cosine_recall@3": 0.47622367870589877,
+      "eval_dim_512_cosine_recall@5": 0.5550132478036536,
+      "eval_dim_64_cosine_accuracy@1": 0.3371914656254358,
+      "eval_dim_64_cosine_accuracy@10": 0.7071538139729466,
+      "eval_dim_64_cosine_accuracy@3": 0.45725840189652767,
+      "eval_dim_64_cosine_accuracy@5": 0.5389764328545531,
+      "eval_dim_64_cosine_map@100": 0.43880820327735404,
+      "eval_dim_64_cosine_mrr@10": 0.428855420753344,
+      "eval_dim_64_cosine_ndcg@10": 0.4933751447627624,
+      "eval_dim_64_cosine_precision@1": 0.3371914656254358,
+      "eval_dim_64_cosine_precision@10": 0.07071538139729466,
+      "eval_dim_64_cosine_precision@3": 0.15241946729884256,
+      "eval_dim_64_cosine_precision@5": 0.10779528657091063,
+      "eval_dim_64_cosine_recall@1": 0.3371914656254358,
+      "eval_dim_64_cosine_recall@10": 0.7071538139729466,
+      "eval_dim_64_cosine_recall@3": 0.45725840189652767,
+      "eval_dim_64_cosine_recall@5": 0.5389764328545531,
+      "eval_dim_768_cosine_accuracy@1": 0.35406498396318503,
+      "eval_dim_768_cosine_accuracy@10": 0.7283502998187142,
+      "eval_dim_768_cosine_accuracy@3": 0.4713429089387812,
+      "eval_dim_768_cosine_accuracy@5": 0.552782038767257,
+      "eval_dim_768_cosine_map@100": 0.45544037871042936,
+      "eval_dim_768_cosine_mrr@10": 0.4460218406146454,
+      "eval_dim_768_cosine_ndcg@10": 0.5113413299327996,
+      "eval_dim_768_cosine_precision@1": 0.35406498396318503,
+      "eval_dim_768_cosine_precision@10": 0.07283502998187143,
+      "eval_dim_768_cosine_precision@3": 0.15711430297959372,
+      "eval_dim_768_cosine_precision@5": 0.1105564077534514,
+      "eval_dim_768_cosine_recall@1": 0.35406498396318503,
+      "eval_dim_768_cosine_recall@10": 0.7283502998187142,
+      "eval_dim_768_cosine_recall@3": 0.4713429089387812,
+      "eval_dim_768_cosine_recall@5": 0.552782038767257,
+      "eval_loss": 0.166685089468956,
+      "eval_runtime": 5294.4091,
+      "eval_samples_per_second": 1.355,
+      "eval_sequential_score": 0.4933751447627624,
+      "eval_steps_per_second": 0.014,
       "step": 240
+    },
+    {
+      "epoch": 5.878048780487805,
+      "grad_norm": 715763.1875,
+      "learning_rate": 0.0,
+      "loss": 0.0521,
+      "step": 282
+    },
+    {
+      "epoch": 5.878048780487805,
+      "eval_dim_128_cosine_accuracy@1": 0.34667410403012133,
+      "eval_dim_128_cosine_accuracy@10": 0.7259796402175429,
+      "eval_dim_128_cosine_accuracy@3": 0.46869334820806025,
+      "eval_dim_128_cosine_accuracy@5": 0.548877422953563,
+      "eval_dim_128_cosine_map@100": 0.44958294087462464,
+      "eval_dim_128_cosine_mrr@10": 0.44026242161439555,
+      "eval_dim_128_cosine_ndcg@10": 0.5064194393114915,
+      "eval_dim_128_cosine_precision@1": 0.34667410403012133,
+      "eval_dim_128_cosine_precision@10": 0.0725979640217543,
+      "eval_dim_128_cosine_precision@3": 0.15623111606935342,
+      "eval_dim_128_cosine_precision@5": 0.1097754845907126,
+      "eval_dim_128_cosine_recall@1": 0.34667410403012133,
+      "eval_dim_128_cosine_recall@10": 0.7259796402175429,
+      "eval_dim_128_cosine_recall@3": 0.46869334820806025,
+      "eval_dim_128_cosine_recall@5": 0.548877422953563,
+      "eval_dim_256_cosine_accuracy@1": 0.35099707153813975,
+      "eval_dim_256_cosine_accuracy@10": 0.719286013108353,
+      "eval_dim_256_cosine_accuracy@3": 0.47455027192860133,
+      "eval_dim_256_cosine_accuracy@5": 0.5502719286013108,
+      "eval_dim_256_cosine_map@100": 0.45309491401564994,
+      "eval_dim_256_cosine_mrr@10": 0.4432182202123648,
+      "eval_dim_256_cosine_ndcg@10": 0.5072173461870104,
+      "eval_dim_256_cosine_precision@1": 0.35099707153813975,
+      "eval_dim_256_cosine_precision@10": 0.0719286013108353,
+      "eval_dim_256_cosine_precision@3": 0.15818342397620047,
+      "eval_dim_256_cosine_precision@5": 0.11005438572026217,
+      "eval_dim_256_cosine_recall@1": 0.35099707153813975,
+      "eval_dim_256_cosine_recall@10": 0.719286013108353,
+      "eval_dim_256_cosine_recall@3": 0.47455027192860133,
+      "eval_dim_256_cosine_recall@5": 0.5502719286013108,
+      "eval_dim_512_cosine_accuracy@1": 0.3569934458234556,
+      "eval_dim_512_cosine_accuracy@10": 0.7248640356993445,
+      "eval_dim_512_cosine_accuracy@3": 0.4744108213638265,
+      "eval_dim_512_cosine_accuracy@5": 0.5523636870729326,
+      "eval_dim_512_cosine_map@100": 0.45691118758008836,
+      "eval_dim_512_cosine_mrr@10": 0.4473125330641728,
+      "eval_dim_512_cosine_ndcg@10": 0.5115413503217677,
+      "eval_dim_512_cosine_precision@1": 0.3569934458234556,
+      "eval_dim_512_cosine_precision@10": 0.07248640356993445,
+      "eval_dim_512_cosine_precision@3": 0.15813694045460885,
+      "eval_dim_512_cosine_precision@5": 0.11047273741458653,
+      "eval_dim_512_cosine_recall@1": 0.3569934458234556,
+      "eval_dim_512_cosine_recall@10": 0.7248640356993445,
+      "eval_dim_512_cosine_recall@3": 0.4744108213638265,
+      "eval_dim_512_cosine_recall@5": 0.5523636870729326,
+      "eval_dim_64_cosine_accuracy@1": 0.3380281690140845,
+      "eval_dim_64_cosine_accuracy@10": 0.7046437038070004,
+      "eval_dim_64_cosine_accuracy@3": 0.4548877422953563,
+      "eval_dim_64_cosine_accuracy@5": 0.5322828057453632,
+      "eval_dim_64_cosine_map@100": 0.4382982729365051,
+      "eval_dim_64_cosine_mrr@10": 0.4281275109402298,
+      "eval_dim_64_cosine_ndcg@10": 0.49214355622918665,
+      "eval_dim_64_cosine_precision@1": 0.3380281690140845,
+      "eval_dim_64_cosine_precision@10": 0.07046437038070004,
+      "eval_dim_64_cosine_precision@3": 0.1516292474317854,
+      "eval_dim_64_cosine_precision@5": 0.10645656114907265,
+      "eval_dim_64_cosine_recall@1": 0.3380281690140845,
+      "eval_dim_64_cosine_recall@10": 0.7046437038070004,
+      "eval_dim_64_cosine_recall@3": 0.4548877422953563,
+      "eval_dim_64_cosine_recall@5": 0.5322828057453632,
+      "eval_dim_768_cosine_accuracy@1": 0.35490168735183375,
+      "eval_dim_768_cosine_accuracy@10": 0.7261190907823176,
+      "eval_dim_768_cosine_accuracy@3": 0.4723190628922047,
+      "eval_dim_768_cosine_accuracy@5": 0.5504113791660856,
+      "eval_dim_768_cosine_map@100": 0.455135537492312,
+      "eval_dim_768_cosine_mrr@10": 0.44549231583118165,
+      "eval_dim_768_cosine_ndcg@10": 0.5104063813142831,
+      "eval_dim_768_cosine_precision@1": 0.35490168735183375,
+      "eval_dim_768_cosine_precision@10": 0.07261190907823176,
+      "eval_dim_768_cosine_precision@3": 0.1574396876307349,
+      "eval_dim_768_cosine_precision@5": 0.11008227583321714,
+      "eval_dim_768_cosine_recall@1": 0.35490168735183375,
+      "eval_dim_768_cosine_recall@10": 0.7261190907823176,
+      "eval_dim_768_cosine_recall@3": 0.4723190628922047,
+      "eval_dim_768_cosine_recall@5": 0.5504113791660856,
+      "eval_loss": 0.16643136739730835,
+      "eval_runtime": 5293.6409,
+      "eval_samples_per_second": 1.355,
+      "eval_sequential_score": 0.49214355622918665,
+      "eval_steps_per_second": 0.014,
+      "step": 282
     }
   ],
   "logging_steps": 500,
-  "max_steps": 354,
+  "max_steps": 282,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 6,
   "save_steps": 500,
@@ -385,7 +567,7 @@
         "early_stopping_threshold": 0.001
       },
       "attributes": {
-        "early_stopping_patience_counter": 0
+        "early_stopping_patience_counter": 1
       }
     },
     "TrainerControl": {
@@ -394,13 +576,13 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
   "total_flos": 0.0,
-  "train_batch_size": 80,
+  "train_batch_size": 100,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a1704bcfcfd1965a4c663d038235420f54fec65277e92be8502fddf96743aea5
+oid sha256:f36f3242f7ba54c2450351b52935014d0ea597a3fc4c27c34bb913b3491658f2
 size 5624