update ablation study models

.gitattributes

@@ -186,3 +186,96 @@ modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4.512/checkpoint-50
 .git/lfs/objects/c9/29/c9292f8084df08f44633c95111d71cf5055e02e10995567cc733f2bef07d3113 filter=lfs diff=lfs merge=lfs -text
 .git/lfs/objects/7e/97/7e973922297f0c493940a90076a687d8de67be2aa6db71915dd86d5c61411430 filter=lfs diff=lfs merge=lfs -text
 modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4.512/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_quarter.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_low.neg_zero.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_half.neg_zero.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-15000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-15000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-20000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-20000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-10000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-10000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-flatten.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_low.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_zero.neg_high.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-4000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-4000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-2000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-2000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-3000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-3000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-1000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-1000/model.safetensors filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-5000/optimizer.pt filter=lfs diff=lfs merge=lfs -text
+modernbert-crux-researchy-pos_high.neg_zero.b64_n512.1e-4/checkpoint-5000/model.safetensors filter=lfs diff=lfs merge=lfs -text

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/config.json

@@ -0,0 +1,45 @@
+{
+  "architectures": [
+    "ModernBertModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 50281,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "embedding_dropout": 0.0,
+  "eos_token_id": 50282,
+  "global_attn_every_n_layers": 3,
+  "global_rope_theta": 160000.0,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "layer_norm_eps": 1e-05,
+  "local_attention": 128,
+  "local_rope_theta": 10000.0,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "repad_logits_with_grad": false,
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.0",
+  "vocab_size": 50368
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5bd56a1262b3564d49c1dc685e5e53ebad9fb0773925dd5a93b507f36dc9f7c7
+size 298041696

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6de8ed923ca2086d0104e84c05300e12e9d0899e81e5aecff0ec9bd73b35052c
+size 596170443

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3044a568f2c48f55afab6e63cf0c046d3586784e77e39b1af825abecfb2cdba8
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b50176e6a2e04d27f26849df53fc28c50b0357c3f1bc26ca6b6bb8848991144
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fbda571d0d08add9abed7e3b3b11bb38b8d472a92fd8409d00660f44b646db10
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a5ce74a6352b774e1ca5b1ca2e8b3e684a2cc29987576c968c710783e2b33305
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:053b181ac2db5ceae48310494587c492212456802136bc3852c62ccc7b9ff38c
+size 1465

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/trainer_state.json

@@ -0,0 +1,799 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.7127583749109052,
+  "eval_steps": 1000.0,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval/nano_beir.arguana": 0.5201586889046041,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.climate_fever": 0.23997409265084552,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.dbpedia_entity": 0.34286887314902237,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fever": 0.6213496755517833,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fiqa": 0.4542903465876373,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.hotpotqa": 0.6560051657630788,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nfcorpus": 0.3328835143785305,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nq": 0.4570725655218753,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.quora": 0.832098944676056,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scidocs": 0.25908236715595256,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scifact": 0.7391066781087818,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.webis_touche2020": 0.30431141846717047,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/avg": 0.4799335275762782,
+      "step": 0
+    },
+    {
+      "epoch": 0.007127583749109052,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8e-06,
+      "loss": 1.5314,
+      "step": 10
+    },
+    {
+      "epoch": 0.014255167498218105,
+      "grad_norm": 11.125,
+      "learning_rate": 3.8e-06,
+      "loss": 1.4333,
+      "step": 20
+    },
+    {
+      "epoch": 0.021382751247327157,
+      "grad_norm": 22.75,
+      "learning_rate": 5.8e-06,
+      "loss": 1.6002,
+      "step": 30
+    },
+    {
+      "epoch": 0.02851033499643621,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5692,
+      "step": 40
+    },
+    {
+      "epoch": 0.03563791874554526,
+      "grad_norm": 14.125,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.2221,
+      "step": 50
+    },
+    {
+      "epoch": 0.042765502494654314,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3981,
+      "step": 60
+    },
+    {
+      "epoch": 0.04989308624376337,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3844,
+      "step": 70
+    },
+    {
+      "epoch": 0.05702066999287242,
+      "grad_norm": 12.375,
+      "learning_rate": 1.58e-05,
+      "loss": 1.6756,
+      "step": 80
+    },
+    {
+      "epoch": 0.06414825374198146,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.78e-05,
+      "loss": 1.3874,
+      "step": 90
+    },
+    {
+      "epoch": 0.07127583749109052,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.3705,
+      "step": 100
+    },
+    {
+      "epoch": 0.07840342124019957,
+      "grad_norm": 10.0,
+      "learning_rate": 2.18e-05,
+      "loss": 1.3897,
+      "step": 110
+    },
+    {
+      "epoch": 0.08553100498930863,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.38e-05,
+      "loss": 1.4046,
+      "step": 120
+    },
+    {
+      "epoch": 0.09265858873841767,
+      "grad_norm": 8.25,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1084,
+      "step": 130
+    },
+    {
+      "epoch": 0.09978617248752673,
+      "grad_norm": 7.96875,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.1872,
+      "step": 140
+    },
+    {
+      "epoch": 0.10691375623663578,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.98e-05,
+      "loss": 1.4332,
+      "step": 150
+    },
+    {
+      "epoch": 0.11404133998574484,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.18e-05,
+      "loss": 1.2686,
+      "step": 160
+    },
+    {
+      "epoch": 0.12116892373485388,
+      "grad_norm": 12.625,
+      "learning_rate": 3.38e-05,
+      "loss": 0.9751,
+      "step": 170
+    },
+    {
+      "epoch": 0.12829650748396293,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.58e-05,
+      "loss": 1.1097,
+      "step": 180
+    },
+    {
+      "epoch": 0.13542409123307197,
+      "grad_norm": 10.6875,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.0446,
+      "step": 190
+    },
+    {
+      "epoch": 0.14255167498218105,
+      "grad_norm": 8.625,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.2052,
+      "step": 200
+    },
+    {
+      "epoch": 0.1496792587312901,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.18e-05,
+      "loss": 1.0325,
+      "step": 210
+    },
+    {
+      "epoch": 0.15680684248039914,
+      "grad_norm": 9.1875,
+      "learning_rate": 4.38e-05,
+      "loss": 0.9825,
+      "step": 220
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.58e-05,
+      "loss": 1.0758,
+      "step": 230
+    },
+    {
+      "epoch": 0.17106200997861726,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.78e-05,
+      "loss": 1.076,
+      "step": 240
+    },
+    {
+      "epoch": 0.1781895937277263,
+      "grad_norm": 8.25,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.0498,
+      "step": 250
+    },
+    {
+      "epoch": 0.18531717747683535,
+      "grad_norm": 7.9375,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.9616,
+      "step": 260
+    },
+    {
+      "epoch": 0.1924447612259444,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 1.0782,
+      "step": 270
+    },
+    {
+      "epoch": 0.19957234497505347,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.9648,
+      "step": 280
+    },
+    {
+      "epoch": 0.2066999287241625,
+      "grad_norm": 9.375,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.9675,
+      "step": 290
+    },
+    {
+      "epoch": 0.21382751247327156,
+      "grad_norm": 7.53125,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.9063,
+      "step": 300
+    },
+    {
+      "epoch": 0.2209550962223806,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.18e-05,
+      "loss": 0.8631,
+      "step": 310
+    },
+    {
+      "epoch": 0.22808267997148968,
+      "grad_norm": 7.28125,
+      "learning_rate": 6.38e-05,
+      "loss": 0.9451,
+      "step": 320
+    },
+    {
+      "epoch": 0.23521026372059872,
+      "grad_norm": 8.9375,
+      "learning_rate": 6.58e-05,
+      "loss": 1.0807,
+      "step": 330
+    },
+    {
+      "epoch": 0.24233784746970777,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.9637,
+      "step": 340
+    },
+    {
+      "epoch": 0.2494654312188168,
+      "grad_norm": 6.25,
+      "learning_rate": 6.98e-05,
+      "loss": 0.9216,
+      "step": 350
+    },
+    {
+      "epoch": 0.25659301496792586,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.18e-05,
+      "loss": 0.9588,
+      "step": 360
+    },
+    {
+      "epoch": 0.2637205987170349,
+      "grad_norm": 6.65625,
+      "learning_rate": 7.38e-05,
+      "loss": 0.8696,
+      "step": 370
+    },
+    {
+      "epoch": 0.27084818246614395,
+      "grad_norm": 6.0625,
+      "learning_rate": 7.58e-05,
+      "loss": 0.951,
+      "step": 380
+    },
+    {
+      "epoch": 0.27797576621525305,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.9381,
+      "step": 390
+    },
+    {
+      "epoch": 0.2851033499643621,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.98e-05,
+      "loss": 0.9314,
+      "step": 400
+    },
+    {
+      "epoch": 0.29223093371347114,
+      "grad_norm": 5.84375,
+      "learning_rate": 8.18e-05,
+      "loss": 0.9212,
+      "step": 410
+    },
+    {
+      "epoch": 0.2993585174625802,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.38e-05,
+      "loss": 0.8897,
+      "step": 420
+    },
+    {
+      "epoch": 0.30648610121168923,
+      "grad_norm": 5.46875,
+      "learning_rate": 8.58e-05,
+      "loss": 0.9813,
+      "step": 430
+    },
+    {
+      "epoch": 0.3136136849607983,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.78e-05,
+      "loss": 0.9023,
+      "step": 440
+    },
+    {
+      "epoch": 0.3207412687099073,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.98e-05,
+      "loss": 0.882,
+      "step": 450
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.9943,
+      "step": 460
+    },
+    {
+      "epoch": 0.33499643620812547,
+      "grad_norm": 12.0,
+      "learning_rate": 9.38e-05,
+      "loss": 0.8606,
+      "step": 470
+    },
+    {
+      "epoch": 0.3421240199572345,
+      "grad_norm": 6.125,
+      "learning_rate": 9.58e-05,
+      "loss": 0.9137,
+      "step": 480
+    },
+    {
+      "epoch": 0.34925160370634356,
+      "grad_norm": 6.25,
+      "learning_rate": 9.78e-05,
+      "loss": 0.8677,
+      "step": 490
+    },
+    {
+      "epoch": 0.3563791874554526,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.98e-05,
+      "loss": 0.9161,
+      "step": 500
+    },
+    {
+      "epoch": 0.36350677120456165,
+      "grad_norm": 6.125,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.9176,
+      "step": 510
+    },
+    {
+      "epoch": 0.3706343549536707,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.999560138895238e-05,
+      "loss": 0.871,
+      "step": 520
+    },
+    {
+      "epoch": 0.37776193870277974,
+      "grad_norm": 6.28125,
+      "learning_rate": 9.99897530200195e-05,
+      "loss": 0.9097,
+      "step": 530
+    },
+    {
+      "epoch": 0.3848895224518888,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.998146822104943e-05,
+      "loss": 0.8721,
+      "step": 540
+    },
+    {
+      "epoch": 0.3920171062009979,
+      "grad_norm": 7.15625,
+      "learning_rate": 9.997074739583162e-05,
+      "loss": 0.9282,
+      "step": 550
+    },
+    {
+      "epoch": 0.39914468995010693,
+      "grad_norm": 5.375,
+      "learning_rate": 9.995759106688393e-05,
+      "loss": 0.8213,
+      "step": 560
+    },
+    {
+      "epoch": 0.406272273699216,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.99419998754273e-05,
+      "loss": 0.8527,
+      "step": 570
+    },
+    {
+      "epoch": 0.413399857448325,
+      "grad_norm": 5.75,
+      "learning_rate": 9.992397458135438e-05,
+      "loss": 0.944,
+      "step": 580
+    },
+    {
+      "epoch": 0.42052744119743407,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.990351606319261e-05,
+      "loss": 0.9269,
+      "step": 590
+    },
+    {
+      "epoch": 0.4276550249465431,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.8261,
+      "step": 600
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 6.4375,
+      "learning_rate": 9.985530346162299e-05,
+      "loss": 1.0049,
+      "step": 610
+    },
+    {
+      "epoch": 0.4419101924447612,
+      "grad_norm": 6.65625,
+      "learning_rate": 9.982755172802933e-05,
+      "loss": 0.9438,
+      "step": 620
+    },
+    {
+      "epoch": 0.44903777619387025,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.979737146986064e-05,
+      "loss": 0.9527,
+      "step": 630
+    },
+    {
+      "epoch": 0.45616535994297935,
+      "grad_norm": 6.375,
+      "learning_rate": 9.976476415806013e-05,
+      "loss": 0.8917,
+      "step": 640
+    },
+    {
+      "epoch": 0.4632929436920884,
+      "grad_norm": 4.625,
+      "learning_rate": 9.972973138186217e-05,
+      "loss": 0.8554,
+      "step": 650
+    },
+    {
+      "epoch": 0.47042052744119744,
+      "grad_norm": 5.59375,
+      "learning_rate": 9.969227484871484e-05,
+      "loss": 0.978,
+      "step": 660
+    },
+    {
+      "epoch": 0.4775481111903065,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.965239638419672e-05,
+      "loss": 0.774,
+      "step": 670
+    },
+    {
+      "epoch": 0.48467569493941554,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.961009793192792e-05,
+      "loss": 0.8388,
+      "step": 680
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.8888,
+      "step": 690
+    },
+    {
+      "epoch": 0.4989308624376336,
+      "grad_norm": 6.03125,
+      "learning_rate": 9.951824942825215e-05,
+      "loss": 0.7606,
+      "step": 700
+    },
+    {
+      "epoch": 0.5060584461867427,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.946870385341167e-05,
+      "loss": 0.9804,
+      "step": 710
+    },
+    {
+      "epoch": 0.5131860299358517,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.941674724373531e-05,
+      "loss": 0.8832,
+      "step": 720
+    },
+    {
+      "epoch": 0.5203136136849608,
+      "grad_norm": 5.0,
+      "learning_rate": 9.936238213151491e-05,
+      "loss": 0.8385,
+      "step": 730
+    },
+    {
+      "epoch": 0.5274411974340698,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.930561116642935e-05,
+      "loss": 0.9106,
+      "step": 740
+    },
+    {
+      "epoch": 0.5345687811831789,
+      "grad_norm": 6.3125,
+      "learning_rate": 9.924643711541539e-05,
+      "loss": 0.807,
+      "step": 750
+    },
+    {
+      "epoch": 0.5416963649322879,
+      "grad_norm": 4.0,
+      "learning_rate": 9.918486286253279e-05,
+      "loss": 0.7932,
+      "step": 760
+    },
+    {
+      "epoch": 0.5488239486813971,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.912089140882376e-05,
+      "loss": 0.7533,
+      "step": 770
+    },
+    {
+      "epoch": 0.5559515324305061,
+      "grad_norm": 5.125,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.8676,
+      "step": 780
+    },
+    {
+      "epoch": 0.5630791161796151,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.898576948712426e-05,
+      "loss": 0.8783,
+      "step": 790
+    },
+    {
+      "epoch": 0.5702066999287242,
+      "grad_norm": 5.5625,
+      "learning_rate": 9.891462560478562e-05,
+      "loss": 0.9352,
+      "step": 800
+    },
+    {
+      "epoch": 0.5773342836778332,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.884109769260325e-05,
+      "loss": 0.8814,
+      "step": 810
+    },
+    {
+      "epoch": 0.5844618674269423,
+      "grad_norm": 5.96875,
+      "learning_rate": 9.876518933422386e-05,
+      "loss": 0.8252,
+      "step": 820
+    },
+    {
+      "epoch": 0.5915894511760513,
+      "grad_norm": 4.96875,
+      "learning_rate": 9.868690422931372e-05,
+      "loss": 0.8695,
+      "step": 830
+    },
+    {
+      "epoch": 0.5987170349251604,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.860624619337844e-05,
+      "loss": 0.8821,
+      "step": 840
+    },
+    {
+      "epoch": 0.6058446186742694,
+      "grad_norm": 6.375,
+      "learning_rate": 9.852321915757687e-05,
+      "loss": 0.8968,
+      "step": 850
+    },
+    {
+      "epoch": 0.6129722024233785,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.843782716852963e-05,
+      "loss": 0.9109,
+      "step": 860
+    },
+    {
+      "epoch": 0.6200997861724875,
+      "grad_norm": 6.625,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.8701,
+      "step": 870
+    },
+    {
+      "epoch": 0.6272273699215966,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.82599650933e-05,
+      "loss": 0.8649,
+      "step": 880
+    },
+    {
+      "epoch": 0.6343549536707056,
+      "grad_norm": 5.25,
+      "learning_rate": 9.816750367586425e-05,
+      "loss": 0.8578,
+      "step": 890
+    },
+    {
+      "epoch": 0.6414825374198146,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.807269464225355e-05,
+      "loss": 0.9067,
+      "step": 900
+    },
+    {
+      "epoch": 0.6486101211689237,
+      "grad_norm": 6.8125,
+      "learning_rate": 9.797554261332645e-05,
+      "loss": 0.7862,
+      "step": 910
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 5.875,
+      "learning_rate": 9.787605232413574e-05,
+      "loss": 0.8275,
+      "step": 920
+    },
+    {
+      "epoch": 0.6628652886671419,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.777422862369783e-05,
+      "loss": 0.9155,
+      "step": 930
+    },
+    {
+      "epoch": 0.6699928724162509,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.767007647475617e-05,
+      "loss": 0.8526,
+      "step": 940
+    },
+    {
+      "epoch": 0.67712045616536,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.756360095353956e-05,
+      "loss": 0.8123,
+      "step": 950
+    },
+    {
+      "epoch": 0.684248039914469,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.7986,
+      "step": 960
+    },
+    {
+      "epoch": 0.6913756236635781,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.73437006651333e-05,
+      "loss": 0.9203,
+      "step": 970
+    },
+    {
+      "epoch": 0.6985032074126871,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.723028661557345e-05,
+      "loss": 0.7805,
+      "step": 980
+    },
+    {
+      "epoch": 0.7056307911617962,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.711457062847595e-05,
+      "loss": 0.8683,
+      "step": 990
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.699655834367478e-05,
+      "loss": 0.8515,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-1000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1afec67e7d18e7478623434fcf8c1bcc7719c01c4f7874089687d7138953afff
+size 6097

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/config.json

@@ -0,0 +1,45 @@
+{
+  "architectures": [
+    "ModernBertModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 50281,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "embedding_dropout": 0.0,
+  "eos_token_id": 50282,
+  "global_attn_every_n_layers": 3,
+  "global_rope_theta": 160000.0,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "layer_norm_eps": 1e-05,
+  "local_attention": 128,
+  "local_rope_theta": 10000.0,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "repad_logits_with_grad": false,
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.0",
+  "vocab_size": 50368
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24ce79534477c2159ef6b96ac6d5331aa78395ddcb84d9ad63259b832e3f3b71
+size 298041696

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:83f07c64dc00bceb661a5db33fbba0361af657ed1cd530bc5850eb23fed589d7
+size 596170443

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:18bb6b6e7da6250d22c25b95c64287be4aea598d9d97ef67b0fd69b9be869ed7
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3cff16289615ba210b601be2162794d1949cb1d62132099a4cf62330c43649a5
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c6caa3980c911491b74ec2e96e48a78cd7c3365dd32116c61642a4b6839a2da1
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4e93d0d6464a92d227e9235c134b2224cf28ea45ba375cf82e6949c632e2b0d5
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff65ca8b6e225d1d82cc63887ee464e8c896624115a4dc693eae1fff20815699
+size 1465

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/trainer_state.json

@@ -0,0 +1,1564 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.4255167498218104,
+  "eval_steps": 1000.0,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval/nano_beir.arguana": 0.5201586889046041,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.climate_fever": 0.23997409265084552,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.dbpedia_entity": 0.34286887314902237,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fever": 0.6213496755517833,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fiqa": 0.4542903465876373,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.hotpotqa": 0.6560051657630788,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nfcorpus": 0.3328835143785305,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nq": 0.4570725655218753,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.quora": 0.832098944676056,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scidocs": 0.25908236715595256,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scifact": 0.7391066781087818,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.webis_touche2020": 0.30431141846717047,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/avg": 0.4799335275762782,
+      "step": 0
+    },
+    {
+      "epoch": 0.007127583749109052,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8e-06,
+      "loss": 1.5314,
+      "step": 10
+    },
+    {
+      "epoch": 0.014255167498218105,
+      "grad_norm": 11.125,
+      "learning_rate": 3.8e-06,
+      "loss": 1.4333,
+      "step": 20
+    },
+    {
+      "epoch": 0.021382751247327157,
+      "grad_norm": 22.75,
+      "learning_rate": 5.8e-06,
+      "loss": 1.6002,
+      "step": 30
+    },
+    {
+      "epoch": 0.02851033499643621,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5692,
+      "step": 40
+    },
+    {
+      "epoch": 0.03563791874554526,
+      "grad_norm": 14.125,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.2221,
+      "step": 50
+    },
+    {
+      "epoch": 0.042765502494654314,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3981,
+      "step": 60
+    },
+    {
+      "epoch": 0.04989308624376337,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3844,
+      "step": 70
+    },
+    {
+      "epoch": 0.05702066999287242,
+      "grad_norm": 12.375,
+      "learning_rate": 1.58e-05,
+      "loss": 1.6756,
+      "step": 80
+    },
+    {
+      "epoch": 0.06414825374198146,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.78e-05,
+      "loss": 1.3874,
+      "step": 90
+    },
+    {
+      "epoch": 0.07127583749109052,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.3705,
+      "step": 100
+    },
+    {
+      "epoch": 0.07840342124019957,
+      "grad_norm": 10.0,
+      "learning_rate": 2.18e-05,
+      "loss": 1.3897,
+      "step": 110
+    },
+    {
+      "epoch": 0.08553100498930863,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.38e-05,
+      "loss": 1.4046,
+      "step": 120
+    },
+    {
+      "epoch": 0.09265858873841767,
+      "grad_norm": 8.25,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1084,
+      "step": 130
+    },
+    {
+      "epoch": 0.09978617248752673,
+      "grad_norm": 7.96875,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.1872,
+      "step": 140
+    },
+    {
+      "epoch": 0.10691375623663578,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.98e-05,
+      "loss": 1.4332,
+      "step": 150
+    },
+    {
+      "epoch": 0.11404133998574484,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.18e-05,
+      "loss": 1.2686,
+      "step": 160
+    },
+    {
+      "epoch": 0.12116892373485388,
+      "grad_norm": 12.625,
+      "learning_rate": 3.38e-05,
+      "loss": 0.9751,
+      "step": 170
+    },
+    {
+      "epoch": 0.12829650748396293,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.58e-05,
+      "loss": 1.1097,
+      "step": 180
+    },
+    {
+      "epoch": 0.13542409123307197,
+      "grad_norm": 10.6875,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.0446,
+      "step": 190
+    },
+    {
+      "epoch": 0.14255167498218105,
+      "grad_norm": 8.625,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.2052,
+      "step": 200
+    },
+    {
+      "epoch": 0.1496792587312901,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.18e-05,
+      "loss": 1.0325,
+      "step": 210
+    },
+    {
+      "epoch": 0.15680684248039914,
+      "grad_norm": 9.1875,
+      "learning_rate": 4.38e-05,
+      "loss": 0.9825,
+      "step": 220
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.58e-05,
+      "loss": 1.0758,
+      "step": 230
+    },
+    {
+      "epoch": 0.17106200997861726,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.78e-05,
+      "loss": 1.076,
+      "step": 240
+    },
+    {
+      "epoch": 0.1781895937277263,
+      "grad_norm": 8.25,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.0498,
+      "step": 250
+    },
+    {
+      "epoch": 0.18531717747683535,
+      "grad_norm": 7.9375,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.9616,
+      "step": 260
+    },
+    {
+      "epoch": 0.1924447612259444,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 1.0782,
+      "step": 270
+    },
+    {
+      "epoch": 0.19957234497505347,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.9648,
+      "step": 280
+    },
+    {
+      "epoch": 0.2066999287241625,
+      "grad_norm": 9.375,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.9675,
+      "step": 290
+    },
+    {
+      "epoch": 0.21382751247327156,
+      "grad_norm": 7.53125,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.9063,
+      "step": 300
+    },
+    {
+      "epoch": 0.2209550962223806,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.18e-05,
+      "loss": 0.8631,
+      "step": 310
+    },
+    {
+      "epoch": 0.22808267997148968,
+      "grad_norm": 7.28125,
+      "learning_rate": 6.38e-05,
+      "loss": 0.9451,
+      "step": 320
+    },
+    {
+      "epoch": 0.23521026372059872,
+      "grad_norm": 8.9375,
+      "learning_rate": 6.58e-05,
+      "loss": 1.0807,
+      "step": 330
+    },
+    {
+      "epoch": 0.24233784746970777,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.9637,
+      "step": 340
+    },
+    {
+      "epoch": 0.2494654312188168,
+      "grad_norm": 6.25,
+      "learning_rate": 6.98e-05,
+      "loss": 0.9216,
+      "step": 350
+    },
+    {
+      "epoch": 0.25659301496792586,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.18e-05,
+      "loss": 0.9588,
+      "step": 360
+    },
+    {
+      "epoch": 0.2637205987170349,
+      "grad_norm": 6.65625,
+      "learning_rate": 7.38e-05,
+      "loss": 0.8696,
+      "step": 370
+    },
+    {
+      "epoch": 0.27084818246614395,
+      "grad_norm": 6.0625,
+      "learning_rate": 7.58e-05,
+      "loss": 0.951,
+      "step": 380
+    },
+    {
+      "epoch": 0.27797576621525305,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.9381,
+      "step": 390
+    },
+    {
+      "epoch": 0.2851033499643621,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.98e-05,
+      "loss": 0.9314,
+      "step": 400
+    },
+    {
+      "epoch": 0.29223093371347114,
+      "grad_norm": 5.84375,
+      "learning_rate": 8.18e-05,
+      "loss": 0.9212,
+      "step": 410
+    },
+    {
+      "epoch": 0.2993585174625802,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.38e-05,
+      "loss": 0.8897,
+      "step": 420
+    },
+    {
+      "epoch": 0.30648610121168923,
+      "grad_norm": 5.46875,
+      "learning_rate": 8.58e-05,
+      "loss": 0.9813,
+      "step": 430
+    },
+    {
+      "epoch": 0.3136136849607983,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.78e-05,
+      "loss": 0.9023,
+      "step": 440
+    },
+    {
+      "epoch": 0.3207412687099073,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.98e-05,
+      "loss": 0.882,
+      "step": 450
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.9943,
+      "step": 460
+    },
+    {
+      "epoch": 0.33499643620812547,
+      "grad_norm": 12.0,
+      "learning_rate": 9.38e-05,
+      "loss": 0.8606,
+      "step": 470
+    },
+    {
+      "epoch": 0.3421240199572345,
+      "grad_norm": 6.125,
+      "learning_rate": 9.58e-05,
+      "loss": 0.9137,
+      "step": 480
+    },
+    {
+      "epoch": 0.34925160370634356,
+      "grad_norm": 6.25,
+      "learning_rate": 9.78e-05,
+      "loss": 0.8677,
+      "step": 490
+    },
+    {
+      "epoch": 0.3563791874554526,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.98e-05,
+      "loss": 0.9161,
+      "step": 500
+    },
+    {
+      "epoch": 0.36350677120456165,
+      "grad_norm": 6.125,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.9176,
+      "step": 510
+    },
+    {
+      "epoch": 0.3706343549536707,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.999560138895238e-05,
+      "loss": 0.871,
+      "step": 520
+    },
+    {
+      "epoch": 0.37776193870277974,
+      "grad_norm": 6.28125,
+      "learning_rate": 9.99897530200195e-05,
+      "loss": 0.9097,
+      "step": 530
+    },
+    {
+      "epoch": 0.3848895224518888,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.998146822104943e-05,
+      "loss": 0.8721,
+      "step": 540
+    },
+    {
+      "epoch": 0.3920171062009979,
+      "grad_norm": 7.15625,
+      "learning_rate": 9.997074739583162e-05,
+      "loss": 0.9282,
+      "step": 550
+    },
+    {
+      "epoch": 0.39914468995010693,
+      "grad_norm": 5.375,
+      "learning_rate": 9.995759106688393e-05,
+      "loss": 0.8213,
+      "step": 560
+    },
+    {
+      "epoch": 0.406272273699216,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.99419998754273e-05,
+      "loss": 0.8527,
+      "step": 570
+    },
+    {
+      "epoch": 0.413399857448325,
+      "grad_norm": 5.75,
+      "learning_rate": 9.992397458135438e-05,
+      "loss": 0.944,
+      "step": 580
+    },
+    {
+      "epoch": 0.42052744119743407,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.990351606319261e-05,
+      "loss": 0.9269,
+      "step": 590
+    },
+    {
+      "epoch": 0.4276550249465431,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.8261,
+      "step": 600
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 6.4375,
+      "learning_rate": 9.985530346162299e-05,
+      "loss": 1.0049,
+      "step": 610
+    },
+    {
+      "epoch": 0.4419101924447612,
+      "grad_norm": 6.65625,
+      "learning_rate": 9.982755172802933e-05,
+      "loss": 0.9438,
+      "step": 620
+    },
+    {
+      "epoch": 0.44903777619387025,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.979737146986064e-05,
+      "loss": 0.9527,
+      "step": 630
+    },
+    {
+      "epoch": 0.45616535994297935,
+      "grad_norm": 6.375,
+      "learning_rate": 9.976476415806013e-05,
+      "loss": 0.8917,
+      "step": 640
+    },
+    {
+      "epoch": 0.4632929436920884,
+      "grad_norm": 4.625,
+      "learning_rate": 9.972973138186217e-05,
+      "loss": 0.8554,
+      "step": 650
+    },
+    {
+      "epoch": 0.47042052744119744,
+      "grad_norm": 5.59375,
+      "learning_rate": 9.969227484871484e-05,
+      "loss": 0.978,
+      "step": 660
+    },
+    {
+      "epoch": 0.4775481111903065,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.965239638419672e-05,
+      "loss": 0.774,
+      "step": 670
+    },
+    {
+      "epoch": 0.48467569493941554,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.961009793192792e-05,
+      "loss": 0.8388,
+      "step": 680
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.8888,
+      "step": 690
+    },
+    {
+      "epoch": 0.4989308624376336,
+      "grad_norm": 6.03125,
+      "learning_rate": 9.951824942825215e-05,
+      "loss": 0.7606,
+      "step": 700
+    },
+    {
+      "epoch": 0.5060584461867427,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.946870385341167e-05,
+      "loss": 0.9804,
+      "step": 710
+    },
+    {
+      "epoch": 0.5131860299358517,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.941674724373531e-05,
+      "loss": 0.8832,
+      "step": 720
+    },
+    {
+      "epoch": 0.5203136136849608,
+      "grad_norm": 5.0,
+      "learning_rate": 9.936238213151491e-05,
+      "loss": 0.8385,
+      "step": 730
+    },
+    {
+      "epoch": 0.5274411974340698,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.930561116642935e-05,
+      "loss": 0.9106,
+      "step": 740
+    },
+    {
+      "epoch": 0.5345687811831789,
+      "grad_norm": 6.3125,
+      "learning_rate": 9.924643711541539e-05,
+      "loss": 0.807,
+      "step": 750
+    },
+    {
+      "epoch": 0.5416963649322879,
+      "grad_norm": 4.0,
+      "learning_rate": 9.918486286253279e-05,
+      "loss": 0.7932,
+      "step": 760
+    },
+    {
+      "epoch": 0.5488239486813971,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.912089140882376e-05,
+      "loss": 0.7533,
+      "step": 770
+    },
+    {
+      "epoch": 0.5559515324305061,
+      "grad_norm": 5.125,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.8676,
+      "step": 780
+    },
+    {
+      "epoch": 0.5630791161796151,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.898576948712426e-05,
+      "loss": 0.8783,
+      "step": 790
+    },
+    {
+      "epoch": 0.5702066999287242,
+      "grad_norm": 5.5625,
+      "learning_rate": 9.891462560478562e-05,
+      "loss": 0.9352,
+      "step": 800
+    },
+    {
+      "epoch": 0.5773342836778332,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.884109769260325e-05,
+      "loss": 0.8814,
+      "step": 810
+    },
+    {
+      "epoch": 0.5844618674269423,
+      "grad_norm": 5.96875,
+      "learning_rate": 9.876518933422386e-05,
+      "loss": 0.8252,
+      "step": 820
+    },
+    {
+      "epoch": 0.5915894511760513,
+      "grad_norm": 4.96875,
+      "learning_rate": 9.868690422931372e-05,
+      "loss": 0.8695,
+      "step": 830
+    },
+    {
+      "epoch": 0.5987170349251604,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.860624619337844e-05,
+      "loss": 0.8821,
+      "step": 840
+    },
+    {
+      "epoch": 0.6058446186742694,
+      "grad_norm": 6.375,
+      "learning_rate": 9.852321915757687e-05,
+      "loss": 0.8968,
+      "step": 850
+    },
+    {
+      "epoch": 0.6129722024233785,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.843782716852963e-05,
+      "loss": 0.9109,
+      "step": 860
+    },
+    {
+      "epoch": 0.6200997861724875,
+      "grad_norm": 6.625,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.8701,
+      "step": 870
+    },
+    {
+      "epoch": 0.6272273699215966,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.82599650933e-05,
+      "loss": 0.8649,
+      "step": 880
+    },
+    {
+      "epoch": 0.6343549536707056,
+      "grad_norm": 5.25,
+      "learning_rate": 9.816750367586425e-05,
+      "loss": 0.8578,
+      "step": 890
+    },
+    {
+      "epoch": 0.6414825374198146,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.807269464225355e-05,
+      "loss": 0.9067,
+      "step": 900
+    },
+    {
+      "epoch": 0.6486101211689237,
+      "grad_norm": 6.8125,
+      "learning_rate": 9.797554261332645e-05,
+      "loss": 0.7862,
+      "step": 910
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 5.875,
+      "learning_rate": 9.787605232413574e-05,
+      "loss": 0.8275,
+      "step": 920
+    },
+    {
+      "epoch": 0.6628652886671419,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.777422862369783e-05,
+      "loss": 0.9155,
+      "step": 930
+    },
+    {
+      "epoch": 0.6699928724162509,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.767007647475617e-05,
+      "loss": 0.8526,
+      "step": 940
+    },
+    {
+      "epoch": 0.67712045616536,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.756360095353956e-05,
+      "loss": 0.8123,
+      "step": 950
+    },
+    {
+      "epoch": 0.684248039914469,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.7986,
+      "step": 960
+    },
+    {
+      "epoch": 0.6913756236635781,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.73437006651333e-05,
+      "loss": 0.9203,
+      "step": 970
+    },
+    {
+      "epoch": 0.6985032074126871,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.723028661557345e-05,
+      "loss": 0.7805,
+      "step": 980
+    },
+    {
+      "epoch": 0.7056307911617962,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.711457062847595e-05,
+      "loss": 0.8683,
+      "step": 990
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.699655834367478e-05,
+      "loss": 0.8515,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.arguana": 0.49904762134680597,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.climate_fever": 0.2710368630075979,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.dbpedia_entity": 0.32627448935866016,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fever": 0.6251056136949217,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fiqa": 0.4264627976119544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.hotpotqa": 0.67890984670666,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nfcorpus": 0.33575253271342886,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nq": 0.41529049449791683,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.quora": 0.8400060722791932,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scidocs": 0.25236583848912914,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scifact": 0.7248783076751154,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.webis_touche2020": 0.29804501841705877,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/avg": 0.4744312913165369,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7198859586600143,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.68762555129222e-05,
+      "loss": 0.8998,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7270135424091233,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.675366799960841e-05,
+      "loss": 0.8763,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7341411261582323,
+      "grad_norm": 4.875,
+      "learning_rate": 9.662880177847594e-05,
+      "loss": 0.8851,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7412687099073414,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.65016629353282e-05,
+      "loss": 0.8247,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7483962936564504,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.742,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7555238774055595,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.624059227972075e-05,
+      "loss": 0.8707,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7626514611546685,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.610667319147647e-05,
+      "loss": 0.7698,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7697790449037776,
+      "grad_norm": 6.1875,
+      "learning_rate": 9.597050692902765e-05,
+      "loss": 0.9318,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7769066286528866,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.583210012892582e-05,
+      "loss": 0.752,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7840342124019958,
+      "grad_norm": 6.0,
+      "learning_rate": 9.569145953692315e-05,
+      "loss": 0.9631,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7911617961511048,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.55485920076437e-05,
+      "loss": 0.8623,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7982893799002139,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.540350450424926e-05,
+      "loss": 0.8677,
+      "step": 1120
+    },
+    {
+      "epoch": 0.8054169636493229,
+      "grad_norm": 4.03125,
+      "learning_rate": 9.525620409810007e-05,
+      "loss": 0.7795,
+      "step": 1130
+    },
+    {
+      "epoch": 0.812544547398432,
+      "grad_norm": 4.125,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.8011,
+      "step": 1140
+    },
+    {
+      "epoch": 0.819672131147541,
+      "grad_norm": 4.875,
+      "learning_rate": 9.495499340189728e-05,
+      "loss": 0.7984,
+      "step": 1150
+    },
+    {
+      "epoch": 0.82679971489665,
+      "grad_norm": 4.84375,
+      "learning_rate": 9.480109779242805e-05,
+      "loss": 0.8049,
+      "step": 1160
+    },
+    {
+      "epoch": 0.8339272986457591,
+      "grad_norm": 5.15625,
+      "learning_rate": 9.464501864065735e-05,
+      "loss": 0.7839,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8410548823948681,
+      "grad_norm": 4.78125,
+      "learning_rate": 9.448676355366282e-05,
+      "loss": 0.7809,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8481824661439772,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.432634024457414e-05,
+      "loss": 0.7688,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8553100498930862,
+      "grad_norm": 7.75,
+      "learning_rate": 9.416375653219709e-05,
+      "loss": 0.8638,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8624376336421953,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.399902034063242e-05,
+      "loss": 0.8409,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 5.875,
+      "learning_rate": 9.383213969888972e-05,
+      "loss": 0.916,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8766928011404134,
+      "grad_norm": 5.40625,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.9114,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8838203848895224,
+      "grad_norm": 4.5,
+      "learning_rate": 9.34919777030994e-05,
+      "loss": 0.8744,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8909479686386315,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.331871292806759e-05,
+      "loss": 0.8471,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8980755523877405,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.314333686008125e-05,
+      "loss": 0.8014,
+      "step": 1260
+    },
+    {
+      "epoch": 0.9052031361368497,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.296585804672252e-05,
+      "loss": 0.8587,
+      "step": 1270
+    },
+    {
+      "epoch": 0.9123307198859587,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.278628513805838e-05,
+      "loss": 0.8003,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9194583036350678,
+      "grad_norm": 6.09375,
+      "learning_rate": 9.260462688621905e-05,
+      "loss": 0.7404,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9265858873841768,
+      "grad_norm": 5.5,
+      "learning_rate": 9.242089214497145e-05,
+      "loss": 0.722,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9337134711332858,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.223508986928766e-05,
+      "loss": 0.7005,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9408410548823949,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.7987,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9479686386315039,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.1857319037902e-05,
+      "loss": 0.7872,
+      "step": 1330
+    },
+    {
+      "epoch": 0.955096222380613,
+      "grad_norm": 4.875,
+      "learning_rate": 9.16653688942175e-05,
+      "loss": 0.8632,
+      "step": 1340
+    },
+    {
+      "epoch": 0.962223806129722,
+      "grad_norm": 6.96875,
+      "learning_rate": 9.147138803923416e-05,
+      "loss": 0.8165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9693513898788311,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.127538592730519e-05,
+      "loss": 0.7802,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9764789736279401,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.107737211129701e-05,
+      "loss": 0.8668,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9836065573770492,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.087735624212365e-05,
+      "loss": 0.7785,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9907341411261582,
+      "grad_norm": 6.25,
+      "learning_rate": 9.06753480682764e-05,
+      "loss": 0.8952,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9978617248752673,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.047135743534868e-05,
+      "loss": 0.826,
+      "step": 1400
+    },
+    {
+      "epoch": 1.0049893086243764,
+      "grad_norm": 6.625,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.9168,
+      "step": 1410
+    },
+    {
+      "epoch": 1.0121168923734853,
+      "grad_norm": 5.625,
+      "learning_rate": 9.005746865725207e-05,
+      "loss": 0.8506,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0192444761225945,
+      "grad_norm": 4.875,
+      "learning_rate": 8.984759068443831e-05,
+      "loss": 0.9257,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0263720598717034,
+      "grad_norm": 5.96875,
+      "learning_rate": 8.963577059627118e-05,
+      "loss": 0.8302,
+      "step": 1440
+    },
+    {
+      "epoch": 1.0334996436208126,
+      "grad_norm": 5.375,
+      "learning_rate": 8.942201871656291e-05,
+      "loss": 0.7898,
+      "step": 1450
+    },
+    {
+      "epoch": 1.0406272273699215,
+      "grad_norm": 5.75,
+      "learning_rate": 8.920634546327858e-05,
+      "loss": 0.7682,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0477548111190307,
+      "grad_norm": 4.9375,
+      "learning_rate": 8.898876134802826e-05,
+      "loss": 0.8378,
+      "step": 1470
+    },
+    {
+      "epoch": 1.0548823948681396,
+      "grad_norm": 4.15625,
+      "learning_rate": 8.87692769755548e-05,
+      "loss": 0.7113,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0620099786172488,
+      "grad_norm": 5.71875,
+      "learning_rate": 8.854790304321681e-05,
+      "loss": 0.7187,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0691375623663577,
+      "grad_norm": 5.21875,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.7843,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0762651461154669,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.80995297483286e-05,
+      "loss": 0.7253,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0833927298645758,
+      "grad_norm": 4.78125,
+      "learning_rate": 8.787255223886019e-05,
+      "loss": 0.8077,
+      "step": 1520
+    },
+    {
+      "epoch": 1.090520313613685,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.764372887462586e-05,
+      "loss": 0.8465,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0976478973627941,
+      "grad_norm": 4.875,
+      "learning_rate": 8.741307080815357e-05,
+      "loss": 0.7467,
+      "step": 1540
+    },
+    {
+      "epoch": 1.104775481111903,
+      "grad_norm": 5.0,
+      "learning_rate": 8.718058928139205e-05,
+      "loss": 0.7919,
+      "step": 1550
+    },
+    {
+      "epoch": 1.1119030648610122,
+      "grad_norm": 5.78125,
+      "learning_rate": 8.694629562516294e-05,
+      "loss": 0.7399,
+      "step": 1560
+    },
+    {
+      "epoch": 1.1190306486101211,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.671020125860851e-05,
+      "loss": 0.8305,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1261582323592303,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.647231768863512e-05,
+      "loss": 0.8986,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1332858161083392,
+      "grad_norm": 5.65625,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.8041,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1404133998574484,
+      "grad_norm": 5.5625,
+      "learning_rate": 8.599122940150795e-05,
+      "loss": 0.8434,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1475409836065573,
+      "grad_norm": 5.9375,
+      "learning_rate": 8.574804813191858e-05,
+      "loss": 0.7681,
+      "step": 1610
+    },
+    {
+      "epoch": 1.1546685673556665,
+      "grad_norm": 4.71875,
+      "learning_rate": 8.550312455289625e-05,
+      "loss": 0.6835,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1617961511047754,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.525647060167063e-05,
+      "loss": 0.8917,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1689237348538846,
+      "grad_norm": 5.25,
+      "learning_rate": 8.500809829980734e-05,
+      "loss": 0.8296,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1760513186029935,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.4758019752622e-05,
+      "loss": 0.8147,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1831789023521027,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.450624714859016e-05,
+      "loss": 0.7656,
+      "step": 1660
+    },
+    {
+      "epoch": 1.1903064861012118,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.425279275875336e-05,
+      "loss": 0.8931,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1974340698503207,
+      "grad_norm": 4.21875,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.7844,
+      "step": 1680
+    },
+    {
+      "epoch": 1.2045616535994297,
+      "grad_norm": 5.53125,
+      "learning_rate": 8.374088811506819e-05,
+      "loss": 0.7334,
+      "step": 1690
+    },
+    {
+      "epoch": 1.2116892373485388,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.348246281072998e-05,
+      "loss": 0.7509,
+      "step": 1700
+    },
+    {
+      "epoch": 1.218816821097648,
+      "grad_norm": 5.5,
+      "learning_rate": 8.322240561839109e-05,
+      "loss": 0.8287,
+      "step": 1710
+    },
+    {
+      "epoch": 1.225944404846757,
+      "grad_norm": 5.4375,
+      "learning_rate": 8.296072921287217e-05,
+      "loss": 0.8383,
+      "step": 1720
+    },
+    {
+      "epoch": 1.233071988595866,
+      "grad_norm": 6.6875,
+      "learning_rate": 8.269744634791208e-05,
+      "loss": 0.7443,
+      "step": 1730
+    },
+    {
+      "epoch": 1.240199572344975,
+      "grad_norm": 6.0625,
+      "learning_rate": 8.243256985554621e-05,
+      "loss": 0.7868,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2473271560940842,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.21661126454811e-05,
+      "loss": 0.7662,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2544547398431931,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.189808770446528e-05,
+      "loss": 0.7558,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2615823235923023,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.6894,
+      "step": 1770
+    },
+    {
+      "epoch": 1.2687099073414112,
+      "grad_norm": 5.15625,
+      "learning_rate": 8.135738695798376e-05,
+      "loss": 0.7782,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2758374910905204,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.108473750550965e-05,
+      "loss": 0.756,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2829650748396293,
+      "grad_norm": 4.125,
+      "learning_rate": 8.081057302678352e-05,
+      "loss": 0.763,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2900926585887385,
+      "grad_norm": 10.5,
+      "learning_rate": 8.053490688419533e-05,
+      "loss": 0.7255,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2972202423378474,
+      "grad_norm": 4.65625,
+      "learning_rate": 8.02577525133239e-05,
+      "loss": 0.7615,
+      "step": 1820
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 5.15625,
+      "learning_rate": 7.997912342228232e-05,
+      "loss": 0.8092,
+      "step": 1830
+    },
+    {
+      "epoch": 1.3114754098360657,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.969903319105935e-05,
+      "loss": 0.7503,
+      "step": 1840
+    },
+    {
+      "epoch": 1.3186029935851746,
+      "grad_norm": 4.5,
+      "learning_rate": 7.941749547085777e-05,
+      "loss": 0.8121,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3257305773342836,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.8773,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3328581610833927,
+      "grad_norm": 5.6875,
+      "learning_rate": 7.885013252040359e-05,
+      "loss": 0.8202,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3399857448325019,
+      "grad_norm": 6.15625,
+      "learning_rate": 7.856433494262077e-05,
+      "loss": 0.8359,
+      "step": 1880
+    },
+    {
+      "epoch": 1.3471133285816108,
+      "grad_norm": 5.875,
+      "learning_rate": 7.827714517945115e-05,
+      "loss": 0.9009,
+      "step": 1890
+    },
+    {
+      "epoch": 1.3542409123307197,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.798857722811856e-05,
+      "loss": 0.7515,
+      "step": 1900
+    },
+    {
+      "epoch": 1.361368496079829,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.769864515301787e-05,
+      "loss": 0.8321,
+      "step": 1910
+    },
+    {
+      "epoch": 1.368496079828938,
+      "grad_norm": 6.875,
+      "learning_rate": 7.740736308502938e-05,
+      "loss": 0.8211,
+      "step": 1920
+    },
+    {
+      "epoch": 1.375623663578047,
+      "grad_norm": 3.921875,
+      "learning_rate": 7.711474522083015e-05,
+      "loss": 0.7751,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3827512473271562,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.682080582220206e-05,
+      "loss": 0.8174,
+      "step": 1940
+    },
+    {
+      "epoch": 1.389878831076265,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.7058,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3970064148253742,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.622901979013717e-05,
+      "loss": 0.7263,
+      "step": 1960
+    },
+    {
+      "epoch": 1.4041339985744832,
+      "grad_norm": 5.5625,
+      "learning_rate": 7.593120199951671e-05,
+      "loss": 0.7503,
+      "step": 1970
+    },
+    {
+      "epoch": 1.4112615823235923,
+      "grad_norm": 5.34375,
+      "learning_rate": 7.563212035869425e-05,
+      "loss": 0.87,
+      "step": 1980
+    },
+    {
+      "epoch": 1.4183891660727013,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.533178944448705e-05,
+      "loss": 0.7866,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.503022389460014e-05,
+      "loss": 0.787,
+      "step": 2000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-2000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1afec67e7d18e7478623434fcf8c1bcc7719c01c4f7874089687d7138953afff
+size 6097

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/config.json

@@ -0,0 +1,45 @@
+{
+  "architectures": [
+    "ModernBertModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 50281,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "embedding_dropout": 0.0,
+  "eos_token_id": 50282,
+  "global_attn_every_n_layers": 3,
+  "global_rope_theta": 160000.0,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "layer_norm_eps": 1e-05,
+  "local_attention": 128,
+  "local_rope_theta": 10000.0,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "repad_logits_with_grad": false,
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.0",
+  "vocab_size": 50368
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0904555a863ff6d0ac55c9e1d292dce677e6a2180be364f66d90ff64f37d5dec
+size 298041696

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fed561d644f364b4cc3ae8de2596c1b176c1c8a72072585e4390c0165e177921
+size 596170443

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac0e770c61e3acef7bd144efee65a450ed487b244ed293a8dd801394ffcae775
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21354c9980d6ba769d430eb9962ffba457eb04a66cb90e33b0934d3157fae7cf
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d574a8567b1636a399633d2b20273fdb3be4b888e37f0d331cd695b973f10463
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cca2b80df90b78b3ba9d9d2951c5b7aa80fc0a44b4512b495de6de40396fcb6e
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01a8f419d963b829163d14296852a07ad341e5e63e49d44de7c0a10c9369d217
+size 1465

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/trainer_state.json

@@ -0,0 +1,2329 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.1382751247327154,
+  "eval_steps": 1000.0,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval/nano_beir.arguana": 0.5201586889046041,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.climate_fever": 0.23997409265084552,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.dbpedia_entity": 0.34286887314902237,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fever": 0.6213496755517833,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fiqa": 0.4542903465876373,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.hotpotqa": 0.6560051657630788,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nfcorpus": 0.3328835143785305,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nq": 0.4570725655218753,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.quora": 0.832098944676056,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scidocs": 0.25908236715595256,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scifact": 0.7391066781087818,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.webis_touche2020": 0.30431141846717047,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/avg": 0.4799335275762782,
+      "step": 0
+    },
+    {
+      "epoch": 0.007127583749109052,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8e-06,
+      "loss": 1.5314,
+      "step": 10
+    },
+    {
+      "epoch": 0.014255167498218105,
+      "grad_norm": 11.125,
+      "learning_rate": 3.8e-06,
+      "loss": 1.4333,
+      "step": 20
+    },
+    {
+      "epoch": 0.021382751247327157,
+      "grad_norm": 22.75,
+      "learning_rate": 5.8e-06,
+      "loss": 1.6002,
+      "step": 30
+    },
+    {
+      "epoch": 0.02851033499643621,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5692,
+      "step": 40
+    },
+    {
+      "epoch": 0.03563791874554526,
+      "grad_norm": 14.125,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.2221,
+      "step": 50
+    },
+    {
+      "epoch": 0.042765502494654314,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3981,
+      "step": 60
+    },
+    {
+      "epoch": 0.04989308624376337,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3844,
+      "step": 70
+    },
+    {
+      "epoch": 0.05702066999287242,
+      "grad_norm": 12.375,
+      "learning_rate": 1.58e-05,
+      "loss": 1.6756,
+      "step": 80
+    },
+    {
+      "epoch": 0.06414825374198146,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.78e-05,
+      "loss": 1.3874,
+      "step": 90
+    },
+    {
+      "epoch": 0.07127583749109052,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.3705,
+      "step": 100
+    },
+    {
+      "epoch": 0.07840342124019957,
+      "grad_norm": 10.0,
+      "learning_rate": 2.18e-05,
+      "loss": 1.3897,
+      "step": 110
+    },
+    {
+      "epoch": 0.08553100498930863,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.38e-05,
+      "loss": 1.4046,
+      "step": 120
+    },
+    {
+      "epoch": 0.09265858873841767,
+      "grad_norm": 8.25,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1084,
+      "step": 130
+    },
+    {
+      "epoch": 0.09978617248752673,
+      "grad_norm": 7.96875,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.1872,
+      "step": 140
+    },
+    {
+      "epoch": 0.10691375623663578,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.98e-05,
+      "loss": 1.4332,
+      "step": 150
+    },
+    {
+      "epoch": 0.11404133998574484,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.18e-05,
+      "loss": 1.2686,
+      "step": 160
+    },
+    {
+      "epoch": 0.12116892373485388,
+      "grad_norm": 12.625,
+      "learning_rate": 3.38e-05,
+      "loss": 0.9751,
+      "step": 170
+    },
+    {
+      "epoch": 0.12829650748396293,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.58e-05,
+      "loss": 1.1097,
+      "step": 180
+    },
+    {
+      "epoch": 0.13542409123307197,
+      "grad_norm": 10.6875,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.0446,
+      "step": 190
+    },
+    {
+      "epoch": 0.14255167498218105,
+      "grad_norm": 8.625,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.2052,
+      "step": 200
+    },
+    {
+      "epoch": 0.1496792587312901,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.18e-05,
+      "loss": 1.0325,
+      "step": 210
+    },
+    {
+      "epoch": 0.15680684248039914,
+      "grad_norm": 9.1875,
+      "learning_rate": 4.38e-05,
+      "loss": 0.9825,
+      "step": 220
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.58e-05,
+      "loss": 1.0758,
+      "step": 230
+    },
+    {
+      "epoch": 0.17106200997861726,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.78e-05,
+      "loss": 1.076,
+      "step": 240
+    },
+    {
+      "epoch": 0.1781895937277263,
+      "grad_norm": 8.25,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.0498,
+      "step": 250
+    },
+    {
+      "epoch": 0.18531717747683535,
+      "grad_norm": 7.9375,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.9616,
+      "step": 260
+    },
+    {
+      "epoch": 0.1924447612259444,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 1.0782,
+      "step": 270
+    },
+    {
+      "epoch": 0.19957234497505347,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.9648,
+      "step": 280
+    },
+    {
+      "epoch": 0.2066999287241625,
+      "grad_norm": 9.375,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.9675,
+      "step": 290
+    },
+    {
+      "epoch": 0.21382751247327156,
+      "grad_norm": 7.53125,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.9063,
+      "step": 300
+    },
+    {
+      "epoch": 0.2209550962223806,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.18e-05,
+      "loss": 0.8631,
+      "step": 310
+    },
+    {
+      "epoch": 0.22808267997148968,
+      "grad_norm": 7.28125,
+      "learning_rate": 6.38e-05,
+      "loss": 0.9451,
+      "step": 320
+    },
+    {
+      "epoch": 0.23521026372059872,
+      "grad_norm": 8.9375,
+      "learning_rate": 6.58e-05,
+      "loss": 1.0807,
+      "step": 330
+    },
+    {
+      "epoch": 0.24233784746970777,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.9637,
+      "step": 340
+    },
+    {
+      "epoch": 0.2494654312188168,
+      "grad_norm": 6.25,
+      "learning_rate": 6.98e-05,
+      "loss": 0.9216,
+      "step": 350
+    },
+    {
+      "epoch": 0.25659301496792586,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.18e-05,
+      "loss": 0.9588,
+      "step": 360
+    },
+    {
+      "epoch": 0.2637205987170349,
+      "grad_norm": 6.65625,
+      "learning_rate": 7.38e-05,
+      "loss": 0.8696,
+      "step": 370
+    },
+    {
+      "epoch": 0.27084818246614395,
+      "grad_norm": 6.0625,
+      "learning_rate": 7.58e-05,
+      "loss": 0.951,
+      "step": 380
+    },
+    {
+      "epoch": 0.27797576621525305,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.9381,
+      "step": 390
+    },
+    {
+      "epoch": 0.2851033499643621,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.98e-05,
+      "loss": 0.9314,
+      "step": 400
+    },
+    {
+      "epoch": 0.29223093371347114,
+      "grad_norm": 5.84375,
+      "learning_rate": 8.18e-05,
+      "loss": 0.9212,
+      "step": 410
+    },
+    {
+      "epoch": 0.2993585174625802,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.38e-05,
+      "loss": 0.8897,
+      "step": 420
+    },
+    {
+      "epoch": 0.30648610121168923,
+      "grad_norm": 5.46875,
+      "learning_rate": 8.58e-05,
+      "loss": 0.9813,
+      "step": 430
+    },
+    {
+      "epoch": 0.3136136849607983,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.78e-05,
+      "loss": 0.9023,
+      "step": 440
+    },
+    {
+      "epoch": 0.3207412687099073,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.98e-05,
+      "loss": 0.882,
+      "step": 450
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.9943,
+      "step": 460
+    },
+    {
+      "epoch": 0.33499643620812547,
+      "grad_norm": 12.0,
+      "learning_rate": 9.38e-05,
+      "loss": 0.8606,
+      "step": 470
+    },
+    {
+      "epoch": 0.3421240199572345,
+      "grad_norm": 6.125,
+      "learning_rate": 9.58e-05,
+      "loss": 0.9137,
+      "step": 480
+    },
+    {
+      "epoch": 0.34925160370634356,
+      "grad_norm": 6.25,
+      "learning_rate": 9.78e-05,
+      "loss": 0.8677,
+      "step": 490
+    },
+    {
+      "epoch": 0.3563791874554526,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.98e-05,
+      "loss": 0.9161,
+      "step": 500
+    },
+    {
+      "epoch": 0.36350677120456165,
+      "grad_norm": 6.125,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.9176,
+      "step": 510
+    },
+    {
+      "epoch": 0.3706343549536707,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.999560138895238e-05,
+      "loss": 0.871,
+      "step": 520
+    },
+    {
+      "epoch": 0.37776193870277974,
+      "grad_norm": 6.28125,
+      "learning_rate": 9.99897530200195e-05,
+      "loss": 0.9097,
+      "step": 530
+    },
+    {
+      "epoch": 0.3848895224518888,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.998146822104943e-05,
+      "loss": 0.8721,
+      "step": 540
+    },
+    {
+      "epoch": 0.3920171062009979,
+      "grad_norm": 7.15625,
+      "learning_rate": 9.997074739583162e-05,
+      "loss": 0.9282,
+      "step": 550
+    },
+    {
+      "epoch": 0.39914468995010693,
+      "grad_norm": 5.375,
+      "learning_rate": 9.995759106688393e-05,
+      "loss": 0.8213,
+      "step": 560
+    },
+    {
+      "epoch": 0.406272273699216,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.99419998754273e-05,
+      "loss": 0.8527,
+      "step": 570
+    },
+    {
+      "epoch": 0.413399857448325,
+      "grad_norm": 5.75,
+      "learning_rate": 9.992397458135438e-05,
+      "loss": 0.944,
+      "step": 580
+    },
+    {
+      "epoch": 0.42052744119743407,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.990351606319261e-05,
+      "loss": 0.9269,
+      "step": 590
+    },
+    {
+      "epoch": 0.4276550249465431,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.8261,
+      "step": 600
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 6.4375,
+      "learning_rate": 9.985530346162299e-05,
+      "loss": 1.0049,
+      "step": 610
+    },
+    {
+      "epoch": 0.4419101924447612,
+      "grad_norm": 6.65625,
+      "learning_rate": 9.982755172802933e-05,
+      "loss": 0.9438,
+      "step": 620
+    },
+    {
+      "epoch": 0.44903777619387025,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.979737146986064e-05,
+      "loss": 0.9527,
+      "step": 630
+    },
+    {
+      "epoch": 0.45616535994297935,
+      "grad_norm": 6.375,
+      "learning_rate": 9.976476415806013e-05,
+      "loss": 0.8917,
+      "step": 640
+    },
+    {
+      "epoch": 0.4632929436920884,
+      "grad_norm": 4.625,
+      "learning_rate": 9.972973138186217e-05,
+      "loss": 0.8554,
+      "step": 650
+    },
+    {
+      "epoch": 0.47042052744119744,
+      "grad_norm": 5.59375,
+      "learning_rate": 9.969227484871484e-05,
+      "loss": 0.978,
+      "step": 660
+    },
+    {
+      "epoch": 0.4775481111903065,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.965239638419672e-05,
+      "loss": 0.774,
+      "step": 670
+    },
+    {
+      "epoch": 0.48467569493941554,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.961009793192792e-05,
+      "loss": 0.8388,
+      "step": 680
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.8888,
+      "step": 690
+    },
+    {
+      "epoch": 0.4989308624376336,
+      "grad_norm": 6.03125,
+      "learning_rate": 9.951824942825215e-05,
+      "loss": 0.7606,
+      "step": 700
+    },
+    {
+      "epoch": 0.5060584461867427,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.946870385341167e-05,
+      "loss": 0.9804,
+      "step": 710
+    },
+    {
+      "epoch": 0.5131860299358517,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.941674724373531e-05,
+      "loss": 0.8832,
+      "step": 720
+    },
+    {
+      "epoch": 0.5203136136849608,
+      "grad_norm": 5.0,
+      "learning_rate": 9.936238213151491e-05,
+      "loss": 0.8385,
+      "step": 730
+    },
+    {
+      "epoch": 0.5274411974340698,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.930561116642935e-05,
+      "loss": 0.9106,
+      "step": 740
+    },
+    {
+      "epoch": 0.5345687811831789,
+      "grad_norm": 6.3125,
+      "learning_rate": 9.924643711541539e-05,
+      "loss": 0.807,
+      "step": 750
+    },
+    {
+      "epoch": 0.5416963649322879,
+      "grad_norm": 4.0,
+      "learning_rate": 9.918486286253279e-05,
+      "loss": 0.7932,
+      "step": 760
+    },
+    {
+      "epoch": 0.5488239486813971,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.912089140882376e-05,
+      "loss": 0.7533,
+      "step": 770
+    },
+    {
+      "epoch": 0.5559515324305061,
+      "grad_norm": 5.125,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.8676,
+      "step": 780
+    },
+    {
+      "epoch": 0.5630791161796151,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.898576948712426e-05,
+      "loss": 0.8783,
+      "step": 790
+    },
+    {
+      "epoch": 0.5702066999287242,
+      "grad_norm": 5.5625,
+      "learning_rate": 9.891462560478562e-05,
+      "loss": 0.9352,
+      "step": 800
+    },
+    {
+      "epoch": 0.5773342836778332,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.884109769260325e-05,
+      "loss": 0.8814,
+      "step": 810
+    },
+    {
+      "epoch": 0.5844618674269423,
+      "grad_norm": 5.96875,
+      "learning_rate": 9.876518933422386e-05,
+      "loss": 0.8252,
+      "step": 820
+    },
+    {
+      "epoch": 0.5915894511760513,
+      "grad_norm": 4.96875,
+      "learning_rate": 9.868690422931372e-05,
+      "loss": 0.8695,
+      "step": 830
+    },
+    {
+      "epoch": 0.5987170349251604,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.860624619337844e-05,
+      "loss": 0.8821,
+      "step": 840
+    },
+    {
+      "epoch": 0.6058446186742694,
+      "grad_norm": 6.375,
+      "learning_rate": 9.852321915757687e-05,
+      "loss": 0.8968,
+      "step": 850
+    },
+    {
+      "epoch": 0.6129722024233785,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.843782716852963e-05,
+      "loss": 0.9109,
+      "step": 860
+    },
+    {
+      "epoch": 0.6200997861724875,
+      "grad_norm": 6.625,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.8701,
+      "step": 870
+    },
+    {
+      "epoch": 0.6272273699215966,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.82599650933e-05,
+      "loss": 0.8649,
+      "step": 880
+    },
+    {
+      "epoch": 0.6343549536707056,
+      "grad_norm": 5.25,
+      "learning_rate": 9.816750367586425e-05,
+      "loss": 0.8578,
+      "step": 890
+    },
+    {
+      "epoch": 0.6414825374198146,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.807269464225355e-05,
+      "loss": 0.9067,
+      "step": 900
+    },
+    {
+      "epoch": 0.6486101211689237,
+      "grad_norm": 6.8125,
+      "learning_rate": 9.797554261332645e-05,
+      "loss": 0.7862,
+      "step": 910
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 5.875,
+      "learning_rate": 9.787605232413574e-05,
+      "loss": 0.8275,
+      "step": 920
+    },
+    {
+      "epoch": 0.6628652886671419,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.777422862369783e-05,
+      "loss": 0.9155,
+      "step": 930
+    },
+    {
+      "epoch": 0.6699928724162509,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.767007647475617e-05,
+      "loss": 0.8526,
+      "step": 940
+    },
+    {
+      "epoch": 0.67712045616536,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.756360095353956e-05,
+      "loss": 0.8123,
+      "step": 950
+    },
+    {
+      "epoch": 0.684248039914469,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.7986,
+      "step": 960
+    },
+    {
+      "epoch": 0.6913756236635781,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.73437006651333e-05,
+      "loss": 0.9203,
+      "step": 970
+    },
+    {
+      "epoch": 0.6985032074126871,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.723028661557345e-05,
+      "loss": 0.7805,
+      "step": 980
+    },
+    {
+      "epoch": 0.7056307911617962,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.711457062847595e-05,
+      "loss": 0.8683,
+      "step": 990
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.699655834367478e-05,
+      "loss": 0.8515,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.arguana": 0.49904762134680597,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.climate_fever": 0.2710368630075979,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.dbpedia_entity": 0.32627448935866016,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fever": 0.6251056136949217,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fiqa": 0.4264627976119544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.hotpotqa": 0.67890984670666,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nfcorpus": 0.33575253271342886,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nq": 0.41529049449791683,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.quora": 0.8400060722791932,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scidocs": 0.25236583848912914,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scifact": 0.7248783076751154,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.webis_touche2020": 0.29804501841705877,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/avg": 0.4744312913165369,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7198859586600143,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.68762555129222e-05,
+      "loss": 0.8998,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7270135424091233,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.675366799960841e-05,
+      "loss": 0.8763,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7341411261582323,
+      "grad_norm": 4.875,
+      "learning_rate": 9.662880177847594e-05,
+      "loss": 0.8851,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7412687099073414,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.65016629353282e-05,
+      "loss": 0.8247,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7483962936564504,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.742,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7555238774055595,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.624059227972075e-05,
+      "loss": 0.8707,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7626514611546685,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.610667319147647e-05,
+      "loss": 0.7698,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7697790449037776,
+      "grad_norm": 6.1875,
+      "learning_rate": 9.597050692902765e-05,
+      "loss": 0.9318,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7769066286528866,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.583210012892582e-05,
+      "loss": 0.752,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7840342124019958,
+      "grad_norm": 6.0,
+      "learning_rate": 9.569145953692315e-05,
+      "loss": 0.9631,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7911617961511048,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.55485920076437e-05,
+      "loss": 0.8623,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7982893799002139,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.540350450424926e-05,
+      "loss": 0.8677,
+      "step": 1120
+    },
+    {
+      "epoch": 0.8054169636493229,
+      "grad_norm": 4.03125,
+      "learning_rate": 9.525620409810007e-05,
+      "loss": 0.7795,
+      "step": 1130
+    },
+    {
+      "epoch": 0.812544547398432,
+      "grad_norm": 4.125,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.8011,
+      "step": 1140
+    },
+    {
+      "epoch": 0.819672131147541,
+      "grad_norm": 4.875,
+      "learning_rate": 9.495499340189728e-05,
+      "loss": 0.7984,
+      "step": 1150
+    },
+    {
+      "epoch": 0.82679971489665,
+      "grad_norm": 4.84375,
+      "learning_rate": 9.480109779242805e-05,
+      "loss": 0.8049,
+      "step": 1160
+    },
+    {
+      "epoch": 0.8339272986457591,
+      "grad_norm": 5.15625,
+      "learning_rate": 9.464501864065735e-05,
+      "loss": 0.7839,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8410548823948681,
+      "grad_norm": 4.78125,
+      "learning_rate": 9.448676355366282e-05,
+      "loss": 0.7809,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8481824661439772,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.432634024457414e-05,
+      "loss": 0.7688,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8553100498930862,
+      "grad_norm": 7.75,
+      "learning_rate": 9.416375653219709e-05,
+      "loss": 0.8638,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8624376336421953,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.399902034063242e-05,
+      "loss": 0.8409,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 5.875,
+      "learning_rate": 9.383213969888972e-05,
+      "loss": 0.916,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8766928011404134,
+      "grad_norm": 5.40625,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.9114,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8838203848895224,
+      "grad_norm": 4.5,
+      "learning_rate": 9.34919777030994e-05,
+      "loss": 0.8744,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8909479686386315,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.331871292806759e-05,
+      "loss": 0.8471,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8980755523877405,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.314333686008125e-05,
+      "loss": 0.8014,
+      "step": 1260
+    },
+    {
+      "epoch": 0.9052031361368497,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.296585804672252e-05,
+      "loss": 0.8587,
+      "step": 1270
+    },
+    {
+      "epoch": 0.9123307198859587,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.278628513805838e-05,
+      "loss": 0.8003,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9194583036350678,
+      "grad_norm": 6.09375,
+      "learning_rate": 9.260462688621905e-05,
+      "loss": 0.7404,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9265858873841768,
+      "grad_norm": 5.5,
+      "learning_rate": 9.242089214497145e-05,
+      "loss": 0.722,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9337134711332858,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.223508986928766e-05,
+      "loss": 0.7005,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9408410548823949,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.7987,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9479686386315039,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.1857319037902e-05,
+      "loss": 0.7872,
+      "step": 1330
+    },
+    {
+      "epoch": 0.955096222380613,
+      "grad_norm": 4.875,
+      "learning_rate": 9.16653688942175e-05,
+      "loss": 0.8632,
+      "step": 1340
+    },
+    {
+      "epoch": 0.962223806129722,
+      "grad_norm": 6.96875,
+      "learning_rate": 9.147138803923416e-05,
+      "loss": 0.8165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9693513898788311,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.127538592730519e-05,
+      "loss": 0.7802,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9764789736279401,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.107737211129701e-05,
+      "loss": 0.8668,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9836065573770492,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.087735624212365e-05,
+      "loss": 0.7785,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9907341411261582,
+      "grad_norm": 6.25,
+      "learning_rate": 9.06753480682764e-05,
+      "loss": 0.8952,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9978617248752673,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.047135743534868e-05,
+      "loss": 0.826,
+      "step": 1400
+    },
+    {
+      "epoch": 1.0049893086243764,
+      "grad_norm": 6.625,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.9168,
+      "step": 1410
+    },
+    {
+      "epoch": 1.0121168923734853,
+      "grad_norm": 5.625,
+      "learning_rate": 9.005746865725207e-05,
+      "loss": 0.8506,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0192444761225945,
+      "grad_norm": 4.875,
+      "learning_rate": 8.984759068443831e-05,
+      "loss": 0.9257,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0263720598717034,
+      "grad_norm": 5.96875,
+      "learning_rate": 8.963577059627118e-05,
+      "loss": 0.8302,
+      "step": 1440
+    },
+    {
+      "epoch": 1.0334996436208126,
+      "grad_norm": 5.375,
+      "learning_rate": 8.942201871656291e-05,
+      "loss": 0.7898,
+      "step": 1450
+    },
+    {
+      "epoch": 1.0406272273699215,
+      "grad_norm": 5.75,
+      "learning_rate": 8.920634546327858e-05,
+      "loss": 0.7682,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0477548111190307,
+      "grad_norm": 4.9375,
+      "learning_rate": 8.898876134802826e-05,
+      "loss": 0.8378,
+      "step": 1470
+    },
+    {
+      "epoch": 1.0548823948681396,
+      "grad_norm": 4.15625,
+      "learning_rate": 8.87692769755548e-05,
+      "loss": 0.7113,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0620099786172488,
+      "grad_norm": 5.71875,
+      "learning_rate": 8.854790304321681e-05,
+      "loss": 0.7187,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0691375623663577,
+      "grad_norm": 5.21875,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.7843,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0762651461154669,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.80995297483286e-05,
+      "loss": 0.7253,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0833927298645758,
+      "grad_norm": 4.78125,
+      "learning_rate": 8.787255223886019e-05,
+      "loss": 0.8077,
+      "step": 1520
+    },
+    {
+      "epoch": 1.090520313613685,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.764372887462586e-05,
+      "loss": 0.8465,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0976478973627941,
+      "grad_norm": 4.875,
+      "learning_rate": 8.741307080815357e-05,
+      "loss": 0.7467,
+      "step": 1540
+    },
+    {
+      "epoch": 1.104775481111903,
+      "grad_norm": 5.0,
+      "learning_rate": 8.718058928139205e-05,
+      "loss": 0.7919,
+      "step": 1550
+    },
+    {
+      "epoch": 1.1119030648610122,
+      "grad_norm": 5.78125,
+      "learning_rate": 8.694629562516294e-05,
+      "loss": 0.7399,
+      "step": 1560
+    },
+    {
+      "epoch": 1.1190306486101211,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.671020125860851e-05,
+      "loss": 0.8305,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1261582323592303,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.647231768863512e-05,
+      "loss": 0.8986,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1332858161083392,
+      "grad_norm": 5.65625,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.8041,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1404133998574484,
+      "grad_norm": 5.5625,
+      "learning_rate": 8.599122940150795e-05,
+      "loss": 0.8434,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1475409836065573,
+      "grad_norm": 5.9375,
+      "learning_rate": 8.574804813191858e-05,
+      "loss": 0.7681,
+      "step": 1610
+    },
+    {
+      "epoch": 1.1546685673556665,
+      "grad_norm": 4.71875,
+      "learning_rate": 8.550312455289625e-05,
+      "loss": 0.6835,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1617961511047754,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.525647060167063e-05,
+      "loss": 0.8917,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1689237348538846,
+      "grad_norm": 5.25,
+      "learning_rate": 8.500809829980734e-05,
+      "loss": 0.8296,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1760513186029935,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.4758019752622e-05,
+      "loss": 0.8147,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1831789023521027,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.450624714859016e-05,
+      "loss": 0.7656,
+      "step": 1660
+    },
+    {
+      "epoch": 1.1903064861012118,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.425279275875336e-05,
+      "loss": 0.8931,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1974340698503207,
+      "grad_norm": 4.21875,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.7844,
+      "step": 1680
+    },
+    {
+      "epoch": 1.2045616535994297,
+      "grad_norm": 5.53125,
+      "learning_rate": 8.374088811506819e-05,
+      "loss": 0.7334,
+      "step": 1690
+    },
+    {
+      "epoch": 1.2116892373485388,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.348246281072998e-05,
+      "loss": 0.7509,
+      "step": 1700
+    },
+    {
+      "epoch": 1.218816821097648,
+      "grad_norm": 5.5,
+      "learning_rate": 8.322240561839109e-05,
+      "loss": 0.8287,
+      "step": 1710
+    },
+    {
+      "epoch": 1.225944404846757,
+      "grad_norm": 5.4375,
+      "learning_rate": 8.296072921287217e-05,
+      "loss": 0.8383,
+      "step": 1720
+    },
+    {
+      "epoch": 1.233071988595866,
+      "grad_norm": 6.6875,
+      "learning_rate": 8.269744634791208e-05,
+      "loss": 0.7443,
+      "step": 1730
+    },
+    {
+      "epoch": 1.240199572344975,
+      "grad_norm": 6.0625,
+      "learning_rate": 8.243256985554621e-05,
+      "loss": 0.7868,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2473271560940842,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.21661126454811e-05,
+      "loss": 0.7662,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2544547398431931,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.189808770446528e-05,
+      "loss": 0.7558,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2615823235923023,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.6894,
+      "step": 1770
+    },
+    {
+      "epoch": 1.2687099073414112,
+      "grad_norm": 5.15625,
+      "learning_rate": 8.135738695798376e-05,
+      "loss": 0.7782,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2758374910905204,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.108473750550965e-05,
+      "loss": 0.756,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2829650748396293,
+      "grad_norm": 4.125,
+      "learning_rate": 8.081057302678352e-05,
+      "loss": 0.763,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2900926585887385,
+      "grad_norm": 10.5,
+      "learning_rate": 8.053490688419533e-05,
+      "loss": 0.7255,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2972202423378474,
+      "grad_norm": 4.65625,
+      "learning_rate": 8.02577525133239e-05,
+      "loss": 0.7615,
+      "step": 1820
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 5.15625,
+      "learning_rate": 7.997912342228232e-05,
+      "loss": 0.8092,
+      "step": 1830
+    },
+    {
+      "epoch": 1.3114754098360657,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.969903319105935e-05,
+      "loss": 0.7503,
+      "step": 1840
+    },
+    {
+      "epoch": 1.3186029935851746,
+      "grad_norm": 4.5,
+      "learning_rate": 7.941749547085777e-05,
+      "loss": 0.8121,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3257305773342836,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.8773,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3328581610833927,
+      "grad_norm": 5.6875,
+      "learning_rate": 7.885013252040359e-05,
+      "loss": 0.8202,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3399857448325019,
+      "grad_norm": 6.15625,
+      "learning_rate": 7.856433494262077e-05,
+      "loss": 0.8359,
+      "step": 1880
+    },
+    {
+      "epoch": 1.3471133285816108,
+      "grad_norm": 5.875,
+      "learning_rate": 7.827714517945115e-05,
+      "loss": 0.9009,
+      "step": 1890
+    },
+    {
+      "epoch": 1.3542409123307197,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.798857722811856e-05,
+      "loss": 0.7515,
+      "step": 1900
+    },
+    {
+      "epoch": 1.361368496079829,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.769864515301787e-05,
+      "loss": 0.8321,
+      "step": 1910
+    },
+    {
+      "epoch": 1.368496079828938,
+      "grad_norm": 6.875,
+      "learning_rate": 7.740736308502938e-05,
+      "loss": 0.8211,
+      "step": 1920
+    },
+    {
+      "epoch": 1.375623663578047,
+      "grad_norm": 3.921875,
+      "learning_rate": 7.711474522083015e-05,
+      "loss": 0.7751,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3827512473271562,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.682080582220206e-05,
+      "loss": 0.8174,
+      "step": 1940
+    },
+    {
+      "epoch": 1.389878831076265,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.7058,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3970064148253742,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.622901979013717e-05,
+      "loss": 0.7263,
+      "step": 1960
+    },
+    {
+      "epoch": 1.4041339985744832,
+      "grad_norm": 5.5625,
+      "learning_rate": 7.593120199951671e-05,
+      "loss": 0.7503,
+      "step": 1970
+    },
+    {
+      "epoch": 1.4112615823235923,
+      "grad_norm": 5.34375,
+      "learning_rate": 7.563212035869425e-05,
+      "loss": 0.87,
+      "step": 1980
+    },
+    {
+      "epoch": 1.4183891660727013,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.533178944448705e-05,
+      "loss": 0.7866,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.503022389460014e-05,
+      "loss": 0.787,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.arguana": 0.45921477926341064,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.climate_fever": 0.26026043643827973,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.dbpedia_entity": 0.32492910818341564,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.fever": 0.6243268043774819,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.fiqa": 0.406434177003067,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.hotpotqa": 0.6761769995795371,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.nfcorpus": 0.3490707285646159,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.nq": 0.40743688778055903,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.quora": 0.8329214304594091,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.scidocs": 0.2444219693805352,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.scifact": 0.7389917242185631,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.webis_touche2020": 0.2912623613111957,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/avg": 0.4679539505466725,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4326443335709196,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.472743840691299e-05,
+      "loss": 0.873,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4397719173200285,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.44234477387631e-05,
+      "loss": 0.8482,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4468995010691375,
+      "grad_norm": 4.875,
+      "learning_rate": 7.411826670622676e-05,
+      "loss": 0.8179,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4540270848182466,
+      "grad_norm": 4.40625,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.7916,
+      "step": 2040
+    },
+    {
+      "epoch": 1.4611546685673558,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.350439310165842e-05,
+      "loss": 0.7528,
+      "step": 2050
+    },
+    {
+      "epoch": 1.4682822523164647,
+      "grad_norm": 5.75,
+      "learning_rate": 7.319573044895986e-05,
+      "loss": 0.7749,
+      "step": 2060
+    },
+    {
+      "epoch": 1.4754098360655736,
+      "grad_norm": 5.625,
+      "learning_rate": 7.28859372690835e-05,
+      "loss": 0.8334,
+      "step": 2070
+    },
+    {
+      "epoch": 1.4825374198146828,
+      "grad_norm": 7.03125,
+      "learning_rate": 7.257502866091192e-05,
+      "loss": 0.7086,
+      "step": 2080
+    },
+    {
+      "epoch": 1.489665003563792,
+      "grad_norm": 5.25,
+      "learning_rate": 7.226301977769199e-05,
+      "loss": 0.7891,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4967925873129009,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.194992582629654e-05,
+      "loss": 0.7186,
+      "step": 2100
+    },
+    {
+      "epoch": 1.5039201710620098,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.163576206648304e-05,
+      "loss": 0.778,
+      "step": 2110
+    },
+    {
+      "epoch": 1.511047754811119,
+      "grad_norm": 4.875,
+      "learning_rate": 7.132054381014995e-05,
+      "loss": 0.7799,
+      "step": 2120
+    },
+    {
+      "epoch": 1.5181753385602281,
+      "grad_norm": 6.09375,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.7962,
+      "step": 2130
+    },
+    {
+      "epoch": 1.5253029223093373,
+      "grad_norm": 5.84375,
+      "learning_rate": 7.068700531174319e-05,
+      "loss": 0.8263,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5324305060584462,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.036871594744217e-05,
+      "loss": 0.7541,
+      "step": 2150
+    },
+    {
+      "epoch": 1.5395580898075552,
+      "grad_norm": 5.1875,
+      "learning_rate": 7.004943384066187e-05,
+      "loss": 0.7617,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5466856735566643,
+      "grad_norm": 6.28125,
+      "learning_rate": 6.97291745527617e-05,
+      "loss": 0.8576,
+      "step": 2170
+    },
+    {
+      "epoch": 1.5538132573057735,
+      "grad_norm": 5.6875,
+      "learning_rate": 6.940795369272754e-05,
+      "loss": 0.7801,
+      "step": 2180
+    },
+    {
+      "epoch": 1.5609408410548824,
+      "grad_norm": 5.96875,
+      "learning_rate": 6.908578691641092e-05,
+      "loss": 0.7256,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5680684248039913,
+      "grad_norm": 4.03125,
+      "learning_rate": 6.876268992576604e-05,
+      "loss": 0.7574,
+      "step": 2200
+    },
+    {
+      "epoch": 1.5751960085531005,
+      "grad_norm": 6.65625,
+      "learning_rate": 6.843867846808438e-05,
+      "loss": 0.7539,
+      "step": 2210
+    },
+    {
+      "epoch": 1.5823235923022096,
+      "grad_norm": 6.25,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.7865,
+      "step": 2220
+    },
+    {
+      "epoch": 1.5894511760513186,
+      "grad_norm": 4.90625,
+      "learning_rate": 6.778797536285624e-05,
+      "loss": 0.781,
+      "step": 2230
+    },
+    {
+      "epoch": 1.5965787598004275,
+      "grad_norm": 6.5625,
+      "learning_rate": 6.746131542966112e-05,
+      "loss": 0.9054,
+      "step": 2240
+    },
+    {
+      "epoch": 1.6037063435495367,
+      "grad_norm": 4.0625,
+      "learning_rate": 6.713380445658618e-05,
+      "loss": 0.7958,
+      "step": 2250
+    },
+    {
+      "epoch": 1.6108339272986458,
+      "grad_norm": 5.8125,
+      "learning_rate": 6.680545840605423e-05,
+      "loss": 0.7319,
+      "step": 2260
+    },
+    {
+      "epoch": 1.6179615110477548,
+      "grad_norm": 5.5625,
+      "learning_rate": 6.647629328118851e-05,
+      "loss": 0.7538,
+      "step": 2270
+    },
+    {
+      "epoch": 1.6250890947968637,
+      "grad_norm": 4.625,
+      "learning_rate": 6.614632512503288e-05,
+      "loss": 0.7781,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6322166785459729,
+      "grad_norm": 6.21875,
+      "learning_rate": 6.58155700197697e-05,
+      "loss": 0.7636,
+      "step": 2290
+    },
+    {
+      "epoch": 1.639344262295082,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.548404408593621e-05,
+      "loss": 0.7041,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6464718460441912,
+      "grad_norm": 4.53125,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.8014,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6535994297933,
+      "grad_norm": 4.8125,
+      "learning_rate": 6.481874440176506e-05,
+      "loss": 0.7043,
+      "step": 2320
+    },
+    {
+      "epoch": 1.660727013542409,
+      "grad_norm": 4.78125,
+      "learning_rate": 6.448500307719538e-05,
+      "loss": 0.7739,
+      "step": 2330
+    },
+    {
+      "epoch": 1.6678545972915182,
+      "grad_norm": 5.0625,
+      "learning_rate": 6.415055577401102e-05,
+      "loss": 0.7807,
+      "step": 2340
+    },
+    {
+      "epoch": 1.6749821810406273,
+      "grad_norm": 4.84375,
+      "learning_rate": 6.381541879270169e-05,
+      "loss": 0.8122,
+      "step": 2350
+    },
+    {
+      "epoch": 1.6821097647897363,
+      "grad_norm": 6.25,
+      "learning_rate": 6.347960846737106e-05,
+      "loss": 0.8824,
+      "step": 2360
+    },
+    {
+      "epoch": 1.6892373485388452,
+      "grad_norm": 4.46875,
+      "learning_rate": 6.31431411649406e-05,
+      "loss": 0.6925,
+      "step": 2370
+    },
+    {
+      "epoch": 1.6963649322879544,
+      "grad_norm": 5.59375,
+      "learning_rate": 6.280603328435198e-05,
+      "loss": 0.7558,
+      "step": 2380
+    },
+    {
+      "epoch": 1.7034925160370635,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.24683012557677e-05,
+      "loss": 0.7045,
+      "step": 2390
+    },
+    {
+      "epoch": 1.7106200997861725,
+      "grad_norm": 5.0625,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.7654,
+      "step": 2400
+    },
+    {
+      "epoch": 1.7177476835352814,
+      "grad_norm": 4.90625,
+      "learning_rate": 6.179103062656042e-05,
+      "loss": 0.7696,
+      "step": 2410
+    },
+    {
+      "epoch": 1.7248752672843906,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.145152503515239e-05,
+      "loss": 0.8205,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7320028510334997,
+      "grad_norm": 6.1875,
+      "learning_rate": 6.111146131256983e-05,
+      "loss": 0.7752,
+      "step": 2430
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 4.4375,
+      "learning_rate": 6.077085603303882e-05,
+      "loss": 0.79,
+      "step": 2440
+    },
+    {
+      "epoch": 1.7462580185317176,
+      "grad_norm": 4.75,
+      "learning_rate": 6.042972579718019e-05,
+      "loss": 0.9032,
+      "step": 2450
+    },
+    {
+      "epoch": 1.7533856022808267,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.008808723120035e-05,
+      "loss": 0.7631,
+      "step": 2460
+    },
+    {
+      "epoch": 1.760513186029936,
+      "grad_norm": 4.9375,
+      "learning_rate": 5.974595698608103e-05,
+      "loss": 0.7612,
+      "step": 2470
+    },
+    {
+      "epoch": 1.767640769779045,
+      "grad_norm": 5.125,
+      "learning_rate": 5.9403351736767696e-05,
+      "loss": 0.7827,
+      "step": 2480
+    },
+    {
+      "epoch": 1.774768353528154,
+      "grad_norm": 4.8125,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.7827,
+      "step": 2490
+    },
+    {
+      "epoch": 1.781895937277263,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.8716783040282244e-05,
+      "loss": 0.7646,
+      "step": 2500
+    },
+    {
+      "epoch": 1.789023521026372,
+      "grad_norm": 7.3125,
+      "learning_rate": 5.837285305549978e-05,
+      "loss": 0.8215,
+      "step": 2510
+    },
+    {
+      "epoch": 1.7961511047754812,
+      "grad_norm": 4.84375,
+      "learning_rate": 5.8028514989671724e-05,
+      "loss": 0.7416,
+      "step": 2520
+    },
+    {
+      "epoch": 1.8032786885245902,
+      "grad_norm": 4.3125,
+      "learning_rate": 5.7683785625349616e-05,
+      "loss": 0.7577,
+      "step": 2530
+    },
+    {
+      "epoch": 1.810406272273699,
+      "grad_norm": 4.34375,
+      "learning_rate": 5.733868176415633e-05,
+      "loss": 0.6839,
+      "step": 2540
+    },
+    {
+      "epoch": 1.8175338560228083,
+      "grad_norm": 4.46875,
+      "learning_rate": 5.699322022596722e-05,
+      "loss": 0.719,
+      "step": 2550
+    },
+    {
+      "epoch": 1.8246614397719174,
+      "grad_norm": 5.875,
+      "learning_rate": 5.664741784809022e-05,
+      "loss": 0.7487,
+      "step": 2560
+    },
+    {
+      "epoch": 1.8317890235210263,
+      "grad_norm": 5.40625,
+      "learning_rate": 5.630129148444543e-05,
+      "loss": 0.7946,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8389166072701353,
+      "grad_norm": 5.59375,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.7822,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8460441910192444,
+      "grad_norm": 6.59375,
+      "learning_rate": 5.560813429366345e-05,
+      "loss": 0.7763,
+      "step": 2590
+    },
+    {
+      "epoch": 1.8531717747683536,
+      "grad_norm": 5.90625,
+      "learning_rate": 5.5261137250029835e-05,
+      "loss": 0.8152,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8602993585174625,
+      "grad_norm": 5.53125,
+      "learning_rate": 5.4913883785988993e-05,
+      "loss": 0.8451,
+      "step": 2610
+    },
+    {
+      "epoch": 1.8674269422665715,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.456639082618489e-05,
+      "loss": 0.7819,
+      "step": 2620
+    },
+    {
+      "epoch": 1.8745545260156806,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.421867530693414e-05,
+      "loss": 0.794,
+      "step": 2630
+    },
+    {
+      "epoch": 1.8816821097647898,
+      "grad_norm": 4.8125,
+      "learning_rate": 5.3870754175400595e-05,
+      "loss": 0.7435,
+      "step": 2640
+    },
+    {
+      "epoch": 1.888809693513899,
+      "grad_norm": 6.09375,
+      "learning_rate": 5.352264438876935e-05,
+      "loss": 0.8852,
+      "step": 2650
+    },
+    {
+      "epoch": 1.8959372772630079,
+      "grad_norm": 5.46875,
+      "learning_rate": 5.3174362913420306e-05,
+      "loss": 0.7538,
+      "step": 2660
+    },
+    {
+      "epoch": 1.9030648610121168,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.7795,
+      "step": 2670
+    },
+    {
+      "epoch": 1.910192444761226,
+      "grad_norm": 5.65625,
+      "learning_rate": 5.247735280310041e-05,
+      "loss": 0.7887,
+      "step": 2680
+    },
+    {
+      "epoch": 1.9173200285103351,
+      "grad_norm": 6.09375,
+      "learning_rate": 5.212865813941899e-05,
+      "loss": 0.6838,
+      "step": 2690
+    },
+    {
+      "epoch": 1.924447612259444,
+      "grad_norm": 6.4375,
+      "learning_rate": 5.1779859727942924e-05,
+      "loss": 0.7495,
+      "step": 2700
+    },
+    {
+      "epoch": 1.931575196008553,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.143097456861474e-05,
+      "loss": 0.7499,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9387027797576621,
+      "grad_norm": 5.21875,
+      "learning_rate": 5.1082019665604895e-05,
+      "loss": 0.8243,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9458303635067713,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.073301202648304e-05,
+      "loss": 0.8128,
+      "step": 2730
+    },
+    {
+      "epoch": 1.9529579472558802,
+      "grad_norm": 5.71875,
+      "learning_rate": 5.0383968661389146e-05,
+      "loss": 0.7626,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9600855310049892,
+      "grad_norm": 4.96875,
+      "learning_rate": 5.003490658220438e-05,
+      "loss": 0.7724,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9672131147540983,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.7741,
+      "step": 2760
+    },
+    {
+      "epoch": 1.9743406985032075,
+      "grad_norm": 6.3125,
+      "learning_rate": 4.933679433281836e-05,
+      "loss": 0.7482,
+      "step": 2770
+    },
+    {
+      "epoch": 1.9814682822523164,
+      "grad_norm": 5.6875,
+      "learning_rate": 4.8987778187623245e-05,
+      "loss": 0.7981,
+      "step": 2780
+    },
+    {
+      "epoch": 1.9885958660014256,
+      "grad_norm": 4.875,
+      "learning_rate": 4.863881137669123e-05,
+      "loss": 0.7901,
+      "step": 2790
+    },
+    {
+      "epoch": 1.9957234497505345,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.8289910908172376e-05,
+      "loss": 0.7576,
+      "step": 2800
+    },
+    {
+      "epoch": 2.0028510334996437,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.794109378698327e-05,
+      "loss": 0.7789,
+      "step": 2810
+    },
+    {
+      "epoch": 2.009978617248753,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.7592377013978306e-05,
+      "loss": 0.8136,
+      "step": 2820
+    },
+    {
+      "epoch": 2.0171062009978615,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.724377758512103e-05,
+      "loss": 0.813,
+      "step": 2830
+    },
+    {
+      "epoch": 2.0242337847469707,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.68953124906558e-05,
+      "loss": 0.7948,
+      "step": 2840
+    },
+    {
+      "epoch": 2.03136136849608,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.6773,
+      "step": 2850
+    },
+    {
+      "epoch": 2.038488952245189,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.619885323231484e-05,
+      "loss": 0.6645,
+      "step": 2860
+    },
+    {
+      "epoch": 2.0456165359942977,
+      "grad_norm": 5.34375,
+      "learning_rate": 4.58508930128808e-05,
+      "loss": 0.7572,
+      "step": 2870
+    },
+    {
+      "epoch": 2.052744119743407,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.550313501506781e-05,
+      "loss": 0.7588,
+      "step": 2880
+    },
+    {
+      "epoch": 2.059871703492516,
+      "grad_norm": 5.625,
+      "learning_rate": 4.515559618811005e-05,
+      "loss": 0.7294,
+      "step": 2890
+    },
+    {
+      "epoch": 2.066999287241625,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.4808293470559643e-05,
+      "loss": 0.7621,
+      "step": 2900
+    },
+    {
+      "epoch": 2.0741268709907343,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.446124378946107e-05,
+      "loss": 0.8967,
+      "step": 2910
+    },
+    {
+      "epoch": 2.081254454739843,
+      "grad_norm": 5.3125,
+      "learning_rate": 4.411446405952618e-05,
+      "loss": 0.84,
+      "step": 2920
+    },
+    {
+      "epoch": 2.088382038488952,
+      "grad_norm": 4.375,
+      "learning_rate": 4.376797118230979e-05,
+      "loss": 0.7584,
+      "step": 2930
+    },
+    {
+      "epoch": 2.0955096222380614,
+      "grad_norm": 5.21875,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.8475,
+      "step": 2940
+    },
+    {
+      "epoch": 2.1026372059871705,
+      "grad_norm": 6.40625,
+      "learning_rate": 4.307591352152459e-05,
+      "loss": 0.7067,
+      "step": 2950
+    },
+    {
+      "epoch": 2.1097647897362792,
+      "grad_norm": 6.1875,
+      "learning_rate": 4.273038246786986e-05,
+      "loss": 0.8023,
+      "step": 2960
+    },
+    {
+      "epoch": 2.1168923734853884,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.238520572511773e-05,
+      "loss": 0.7074,
+      "step": 2970
+    },
+    {
+      "epoch": 2.1240199572344975,
+      "grad_norm": 7.5,
+      "learning_rate": 4.2040400116695666e-05,
+      "loss": 0.7498,
+      "step": 2980
+    },
+    {
+      "epoch": 2.1311475409836067,
+      "grad_norm": 7.0625,
+      "learning_rate": 4.169598244794261e-05,
+      "loss": 0.7249,
+      "step": 2990
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "grad_norm": 5.53125,
+      "learning_rate": 4.135196950528982e-05,
+      "loss": 0.7907,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-3000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1afec67e7d18e7478623434fcf8c1bcc7719c01c4f7874089687d7138953afff
+size 6097

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/config.json

@@ -0,0 +1,45 @@
+{
+  "architectures": [
+    "ModernBertModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 50281,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "embedding_dropout": 0.0,
+  "eos_token_id": 50282,
+  "global_attn_every_n_layers": 3,
+  "global_rope_theta": 160000.0,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "layer_norm_eps": 1e-05,
+  "local_attention": 128,
+  "local_rope_theta": 10000.0,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "repad_logits_with_grad": false,
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.0",
+  "vocab_size": 50368
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b72e747ec081562370d86737cb071fdb265789168f91e384f6b53c8d0ab05200
+size 298041696

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bf67c8d687a414eecf2c66a278f63643cc7e1dbaa26d63ca88af5b5188693ff
+size 596170443

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac0e770c61e3acef7bd144efee65a450ed487b244ed293a8dd801394ffcae775
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:21354c9980d6ba769d430eb9962ffba457eb04a66cb90e33b0934d3157fae7cf
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d574a8567b1636a399633d2b20273fdb3be4b888e37f0d331cd695b973f10463
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cca2b80df90b78b3ba9d9d2951c5b7aa80fc0a44b4512b495de6de40396fcb6e
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9218a8f161ba45a13a24f24cd020f32b0abec4bcc53a2764e48b480568dd2d13
+size 1465

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/trainer_state.json

@@ -0,0 +1,3094 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.851033499643621,
+  "eval_steps": 1000.0,
+  "global_step": 4000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval/nano_beir.arguana": 0.5201586889046041,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.climate_fever": 0.23997409265084552,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.dbpedia_entity": 0.34286887314902237,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fever": 0.6213496755517833,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fiqa": 0.4542903465876373,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.hotpotqa": 0.6560051657630788,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nfcorpus": 0.3328835143785305,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nq": 0.4570725655218753,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.quora": 0.832098944676056,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scidocs": 0.25908236715595256,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scifact": 0.7391066781087818,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.webis_touche2020": 0.30431141846717047,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/avg": 0.4799335275762782,
+      "step": 0
+    },
+    {
+      "epoch": 0.007127583749109052,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8e-06,
+      "loss": 1.5314,
+      "step": 10
+    },
+    {
+      "epoch": 0.014255167498218105,
+      "grad_norm": 11.125,
+      "learning_rate": 3.8e-06,
+      "loss": 1.4333,
+      "step": 20
+    },
+    {
+      "epoch": 0.021382751247327157,
+      "grad_norm": 22.75,
+      "learning_rate": 5.8e-06,
+      "loss": 1.6002,
+      "step": 30
+    },
+    {
+      "epoch": 0.02851033499643621,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5692,
+      "step": 40
+    },
+    {
+      "epoch": 0.03563791874554526,
+      "grad_norm": 14.125,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.2221,
+      "step": 50
+    },
+    {
+      "epoch": 0.042765502494654314,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3981,
+      "step": 60
+    },
+    {
+      "epoch": 0.04989308624376337,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3844,
+      "step": 70
+    },
+    {
+      "epoch": 0.05702066999287242,
+      "grad_norm": 12.375,
+      "learning_rate": 1.58e-05,
+      "loss": 1.6756,
+      "step": 80
+    },
+    {
+      "epoch": 0.06414825374198146,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.78e-05,
+      "loss": 1.3874,
+      "step": 90
+    },
+    {
+      "epoch": 0.07127583749109052,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.3705,
+      "step": 100
+    },
+    {
+      "epoch": 0.07840342124019957,
+      "grad_norm": 10.0,
+      "learning_rate": 2.18e-05,
+      "loss": 1.3897,
+      "step": 110
+    },
+    {
+      "epoch": 0.08553100498930863,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.38e-05,
+      "loss": 1.4046,
+      "step": 120
+    },
+    {
+      "epoch": 0.09265858873841767,
+      "grad_norm": 8.25,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1084,
+      "step": 130
+    },
+    {
+      "epoch": 0.09978617248752673,
+      "grad_norm": 7.96875,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.1872,
+      "step": 140
+    },
+    {
+      "epoch": 0.10691375623663578,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.98e-05,
+      "loss": 1.4332,
+      "step": 150
+    },
+    {
+      "epoch": 0.11404133998574484,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.18e-05,
+      "loss": 1.2686,
+      "step": 160
+    },
+    {
+      "epoch": 0.12116892373485388,
+      "grad_norm": 12.625,
+      "learning_rate": 3.38e-05,
+      "loss": 0.9751,
+      "step": 170
+    },
+    {
+      "epoch": 0.12829650748396293,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.58e-05,
+      "loss": 1.1097,
+      "step": 180
+    },
+    {
+      "epoch": 0.13542409123307197,
+      "grad_norm": 10.6875,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.0446,
+      "step": 190
+    },
+    {
+      "epoch": 0.14255167498218105,
+      "grad_norm": 8.625,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.2052,
+      "step": 200
+    },
+    {
+      "epoch": 0.1496792587312901,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.18e-05,
+      "loss": 1.0325,
+      "step": 210
+    },
+    {
+      "epoch": 0.15680684248039914,
+      "grad_norm": 9.1875,
+      "learning_rate": 4.38e-05,
+      "loss": 0.9825,
+      "step": 220
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.58e-05,
+      "loss": 1.0758,
+      "step": 230
+    },
+    {
+      "epoch": 0.17106200997861726,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.78e-05,
+      "loss": 1.076,
+      "step": 240
+    },
+    {
+      "epoch": 0.1781895937277263,
+      "grad_norm": 8.25,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.0498,
+      "step": 250
+    },
+    {
+      "epoch": 0.18531717747683535,
+      "grad_norm": 7.9375,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.9616,
+      "step": 260
+    },
+    {
+      "epoch": 0.1924447612259444,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 1.0782,
+      "step": 270
+    },
+    {
+      "epoch": 0.19957234497505347,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.9648,
+      "step": 280
+    },
+    {
+      "epoch": 0.2066999287241625,
+      "grad_norm": 9.375,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.9675,
+      "step": 290
+    },
+    {
+      "epoch": 0.21382751247327156,
+      "grad_norm": 7.53125,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.9063,
+      "step": 300
+    },
+    {
+      "epoch": 0.2209550962223806,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.18e-05,
+      "loss": 0.8631,
+      "step": 310
+    },
+    {
+      "epoch": 0.22808267997148968,
+      "grad_norm": 7.28125,
+      "learning_rate": 6.38e-05,
+      "loss": 0.9451,
+      "step": 320
+    },
+    {
+      "epoch": 0.23521026372059872,
+      "grad_norm": 8.9375,
+      "learning_rate": 6.58e-05,
+      "loss": 1.0807,
+      "step": 330
+    },
+    {
+      "epoch": 0.24233784746970777,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.9637,
+      "step": 340
+    },
+    {
+      "epoch": 0.2494654312188168,
+      "grad_norm": 6.25,
+      "learning_rate": 6.98e-05,
+      "loss": 0.9216,
+      "step": 350
+    },
+    {
+      "epoch": 0.25659301496792586,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.18e-05,
+      "loss": 0.9588,
+      "step": 360
+    },
+    {
+      "epoch": 0.2637205987170349,
+      "grad_norm": 6.65625,
+      "learning_rate": 7.38e-05,
+      "loss": 0.8696,
+      "step": 370
+    },
+    {
+      "epoch": 0.27084818246614395,
+      "grad_norm": 6.0625,
+      "learning_rate": 7.58e-05,
+      "loss": 0.951,
+      "step": 380
+    },
+    {
+      "epoch": 0.27797576621525305,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.9381,
+      "step": 390
+    },
+    {
+      "epoch": 0.2851033499643621,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.98e-05,
+      "loss": 0.9314,
+      "step": 400
+    },
+    {
+      "epoch": 0.29223093371347114,
+      "grad_norm": 5.84375,
+      "learning_rate": 8.18e-05,
+      "loss": 0.9212,
+      "step": 410
+    },
+    {
+      "epoch": 0.2993585174625802,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.38e-05,
+      "loss": 0.8897,
+      "step": 420
+    },
+    {
+      "epoch": 0.30648610121168923,
+      "grad_norm": 5.46875,
+      "learning_rate": 8.58e-05,
+      "loss": 0.9813,
+      "step": 430
+    },
+    {
+      "epoch": 0.3136136849607983,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.78e-05,
+      "loss": 0.9023,
+      "step": 440
+    },
+    {
+      "epoch": 0.3207412687099073,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.98e-05,
+      "loss": 0.882,
+      "step": 450
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.9943,
+      "step": 460
+    },
+    {
+      "epoch": 0.33499643620812547,
+      "grad_norm": 12.0,
+      "learning_rate": 9.38e-05,
+      "loss": 0.8606,
+      "step": 470
+    },
+    {
+      "epoch": 0.3421240199572345,
+      "grad_norm": 6.125,
+      "learning_rate": 9.58e-05,
+      "loss": 0.9137,
+      "step": 480
+    },
+    {
+      "epoch": 0.34925160370634356,
+      "grad_norm": 6.25,
+      "learning_rate": 9.78e-05,
+      "loss": 0.8677,
+      "step": 490
+    },
+    {
+      "epoch": 0.3563791874554526,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.98e-05,
+      "loss": 0.9161,
+      "step": 500
+    },
+    {
+      "epoch": 0.36350677120456165,
+      "grad_norm": 6.125,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.9176,
+      "step": 510
+    },
+    {
+      "epoch": 0.3706343549536707,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.999560138895238e-05,
+      "loss": 0.871,
+      "step": 520
+    },
+    {
+      "epoch": 0.37776193870277974,
+      "grad_norm": 6.28125,
+      "learning_rate": 9.99897530200195e-05,
+      "loss": 0.9097,
+      "step": 530
+    },
+    {
+      "epoch": 0.3848895224518888,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.998146822104943e-05,
+      "loss": 0.8721,
+      "step": 540
+    },
+    {
+      "epoch": 0.3920171062009979,
+      "grad_norm": 7.15625,
+      "learning_rate": 9.997074739583162e-05,
+      "loss": 0.9282,
+      "step": 550
+    },
+    {
+      "epoch": 0.39914468995010693,
+      "grad_norm": 5.375,
+      "learning_rate": 9.995759106688393e-05,
+      "loss": 0.8213,
+      "step": 560
+    },
+    {
+      "epoch": 0.406272273699216,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.99419998754273e-05,
+      "loss": 0.8527,
+      "step": 570
+    },
+    {
+      "epoch": 0.413399857448325,
+      "grad_norm": 5.75,
+      "learning_rate": 9.992397458135438e-05,
+      "loss": 0.944,
+      "step": 580
+    },
+    {
+      "epoch": 0.42052744119743407,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.990351606319261e-05,
+      "loss": 0.9269,
+      "step": 590
+    },
+    {
+      "epoch": 0.4276550249465431,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.8261,
+      "step": 600
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 6.4375,
+      "learning_rate": 9.985530346162299e-05,
+      "loss": 1.0049,
+      "step": 610
+    },
+    {
+      "epoch": 0.4419101924447612,
+      "grad_norm": 6.65625,
+      "learning_rate": 9.982755172802933e-05,
+      "loss": 0.9438,
+      "step": 620
+    },
+    {
+      "epoch": 0.44903777619387025,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.979737146986064e-05,
+      "loss": 0.9527,
+      "step": 630
+    },
+    {
+      "epoch": 0.45616535994297935,
+      "grad_norm": 6.375,
+      "learning_rate": 9.976476415806013e-05,
+      "loss": 0.8917,
+      "step": 640
+    },
+    {
+      "epoch": 0.4632929436920884,
+      "grad_norm": 4.625,
+      "learning_rate": 9.972973138186217e-05,
+      "loss": 0.8554,
+      "step": 650
+    },
+    {
+      "epoch": 0.47042052744119744,
+      "grad_norm": 5.59375,
+      "learning_rate": 9.969227484871484e-05,
+      "loss": 0.978,
+      "step": 660
+    },
+    {
+      "epoch": 0.4775481111903065,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.965239638419672e-05,
+      "loss": 0.774,
+      "step": 670
+    },
+    {
+      "epoch": 0.48467569493941554,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.961009793192792e-05,
+      "loss": 0.8388,
+      "step": 680
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.8888,
+      "step": 690
+    },
+    {
+      "epoch": 0.4989308624376336,
+      "grad_norm": 6.03125,
+      "learning_rate": 9.951824942825215e-05,
+      "loss": 0.7606,
+      "step": 700
+    },
+    {
+      "epoch": 0.5060584461867427,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.946870385341167e-05,
+      "loss": 0.9804,
+      "step": 710
+    },
+    {
+      "epoch": 0.5131860299358517,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.941674724373531e-05,
+      "loss": 0.8832,
+      "step": 720
+    },
+    {
+      "epoch": 0.5203136136849608,
+      "grad_norm": 5.0,
+      "learning_rate": 9.936238213151491e-05,
+      "loss": 0.8385,
+      "step": 730
+    },
+    {
+      "epoch": 0.5274411974340698,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.930561116642935e-05,
+      "loss": 0.9106,
+      "step": 740
+    },
+    {
+      "epoch": 0.5345687811831789,
+      "grad_norm": 6.3125,
+      "learning_rate": 9.924643711541539e-05,
+      "loss": 0.807,
+      "step": 750
+    },
+    {
+      "epoch": 0.5416963649322879,
+      "grad_norm": 4.0,
+      "learning_rate": 9.918486286253279e-05,
+      "loss": 0.7932,
+      "step": 760
+    },
+    {
+      "epoch": 0.5488239486813971,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.912089140882376e-05,
+      "loss": 0.7533,
+      "step": 770
+    },
+    {
+      "epoch": 0.5559515324305061,
+      "grad_norm": 5.125,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.8676,
+      "step": 780
+    },
+    {
+      "epoch": 0.5630791161796151,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.898576948712426e-05,
+      "loss": 0.8783,
+      "step": 790
+    },
+    {
+      "epoch": 0.5702066999287242,
+      "grad_norm": 5.5625,
+      "learning_rate": 9.891462560478562e-05,
+      "loss": 0.9352,
+      "step": 800
+    },
+    {
+      "epoch": 0.5773342836778332,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.884109769260325e-05,
+      "loss": 0.8814,
+      "step": 810
+    },
+    {
+      "epoch": 0.5844618674269423,
+      "grad_norm": 5.96875,
+      "learning_rate": 9.876518933422386e-05,
+      "loss": 0.8252,
+      "step": 820
+    },
+    {
+      "epoch": 0.5915894511760513,
+      "grad_norm": 4.96875,
+      "learning_rate": 9.868690422931372e-05,
+      "loss": 0.8695,
+      "step": 830
+    },
+    {
+      "epoch": 0.5987170349251604,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.860624619337844e-05,
+      "loss": 0.8821,
+      "step": 840
+    },
+    {
+      "epoch": 0.6058446186742694,
+      "grad_norm": 6.375,
+      "learning_rate": 9.852321915757687e-05,
+      "loss": 0.8968,
+      "step": 850
+    },
+    {
+      "epoch": 0.6129722024233785,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.843782716852963e-05,
+      "loss": 0.9109,
+      "step": 860
+    },
+    {
+      "epoch": 0.6200997861724875,
+      "grad_norm": 6.625,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.8701,
+      "step": 870
+    },
+    {
+      "epoch": 0.6272273699215966,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.82599650933e-05,
+      "loss": 0.8649,
+      "step": 880
+    },
+    {
+      "epoch": 0.6343549536707056,
+      "grad_norm": 5.25,
+      "learning_rate": 9.816750367586425e-05,
+      "loss": 0.8578,
+      "step": 890
+    },
+    {
+      "epoch": 0.6414825374198146,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.807269464225355e-05,
+      "loss": 0.9067,
+      "step": 900
+    },
+    {
+      "epoch": 0.6486101211689237,
+      "grad_norm": 6.8125,
+      "learning_rate": 9.797554261332645e-05,
+      "loss": 0.7862,
+      "step": 910
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 5.875,
+      "learning_rate": 9.787605232413574e-05,
+      "loss": 0.8275,
+      "step": 920
+    },
+    {
+      "epoch": 0.6628652886671419,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.777422862369783e-05,
+      "loss": 0.9155,
+      "step": 930
+    },
+    {
+      "epoch": 0.6699928724162509,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.767007647475617e-05,
+      "loss": 0.8526,
+      "step": 940
+    },
+    {
+      "epoch": 0.67712045616536,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.756360095353956e-05,
+      "loss": 0.8123,
+      "step": 950
+    },
+    {
+      "epoch": 0.684248039914469,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.7986,
+      "step": 960
+    },
+    {
+      "epoch": 0.6913756236635781,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.73437006651333e-05,
+      "loss": 0.9203,
+      "step": 970
+    },
+    {
+      "epoch": 0.6985032074126871,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.723028661557345e-05,
+      "loss": 0.7805,
+      "step": 980
+    },
+    {
+      "epoch": 0.7056307911617962,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.711457062847595e-05,
+      "loss": 0.8683,
+      "step": 990
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.699655834367478e-05,
+      "loss": 0.8515,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.arguana": 0.49904762134680597,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.climate_fever": 0.2710368630075979,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.dbpedia_entity": 0.32627448935866016,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fever": 0.6251056136949217,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fiqa": 0.4264627976119544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.hotpotqa": 0.67890984670666,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nfcorpus": 0.33575253271342886,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nq": 0.41529049449791683,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.quora": 0.8400060722791932,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scidocs": 0.25236583848912914,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scifact": 0.7248783076751154,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.webis_touche2020": 0.29804501841705877,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/avg": 0.4744312913165369,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7198859586600143,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.68762555129222e-05,
+      "loss": 0.8998,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7270135424091233,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.675366799960841e-05,
+      "loss": 0.8763,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7341411261582323,
+      "grad_norm": 4.875,
+      "learning_rate": 9.662880177847594e-05,
+      "loss": 0.8851,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7412687099073414,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.65016629353282e-05,
+      "loss": 0.8247,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7483962936564504,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.742,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7555238774055595,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.624059227972075e-05,
+      "loss": 0.8707,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7626514611546685,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.610667319147647e-05,
+      "loss": 0.7698,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7697790449037776,
+      "grad_norm": 6.1875,
+      "learning_rate": 9.597050692902765e-05,
+      "loss": 0.9318,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7769066286528866,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.583210012892582e-05,
+      "loss": 0.752,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7840342124019958,
+      "grad_norm": 6.0,
+      "learning_rate": 9.569145953692315e-05,
+      "loss": 0.9631,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7911617961511048,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.55485920076437e-05,
+      "loss": 0.8623,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7982893799002139,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.540350450424926e-05,
+      "loss": 0.8677,
+      "step": 1120
+    },
+    {
+      "epoch": 0.8054169636493229,
+      "grad_norm": 4.03125,
+      "learning_rate": 9.525620409810007e-05,
+      "loss": 0.7795,
+      "step": 1130
+    },
+    {
+      "epoch": 0.812544547398432,
+      "grad_norm": 4.125,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.8011,
+      "step": 1140
+    },
+    {
+      "epoch": 0.819672131147541,
+      "grad_norm": 4.875,
+      "learning_rate": 9.495499340189728e-05,
+      "loss": 0.7984,
+      "step": 1150
+    },
+    {
+      "epoch": 0.82679971489665,
+      "grad_norm": 4.84375,
+      "learning_rate": 9.480109779242805e-05,
+      "loss": 0.8049,
+      "step": 1160
+    },
+    {
+      "epoch": 0.8339272986457591,
+      "grad_norm": 5.15625,
+      "learning_rate": 9.464501864065735e-05,
+      "loss": 0.7839,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8410548823948681,
+      "grad_norm": 4.78125,
+      "learning_rate": 9.448676355366282e-05,
+      "loss": 0.7809,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8481824661439772,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.432634024457414e-05,
+      "loss": 0.7688,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8553100498930862,
+      "grad_norm": 7.75,
+      "learning_rate": 9.416375653219709e-05,
+      "loss": 0.8638,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8624376336421953,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.399902034063242e-05,
+      "loss": 0.8409,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 5.875,
+      "learning_rate": 9.383213969888972e-05,
+      "loss": 0.916,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8766928011404134,
+      "grad_norm": 5.40625,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.9114,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8838203848895224,
+      "grad_norm": 4.5,
+      "learning_rate": 9.34919777030994e-05,
+      "loss": 0.8744,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8909479686386315,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.331871292806759e-05,
+      "loss": 0.8471,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8980755523877405,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.314333686008125e-05,
+      "loss": 0.8014,
+      "step": 1260
+    },
+    {
+      "epoch": 0.9052031361368497,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.296585804672252e-05,
+      "loss": 0.8587,
+      "step": 1270
+    },
+    {
+      "epoch": 0.9123307198859587,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.278628513805838e-05,
+      "loss": 0.8003,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9194583036350678,
+      "grad_norm": 6.09375,
+      "learning_rate": 9.260462688621905e-05,
+      "loss": 0.7404,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9265858873841768,
+      "grad_norm": 5.5,
+      "learning_rate": 9.242089214497145e-05,
+      "loss": 0.722,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9337134711332858,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.223508986928766e-05,
+      "loss": 0.7005,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9408410548823949,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.7987,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9479686386315039,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.1857319037902e-05,
+      "loss": 0.7872,
+      "step": 1330
+    },
+    {
+      "epoch": 0.955096222380613,
+      "grad_norm": 4.875,
+      "learning_rate": 9.16653688942175e-05,
+      "loss": 0.8632,
+      "step": 1340
+    },
+    {
+      "epoch": 0.962223806129722,
+      "grad_norm": 6.96875,
+      "learning_rate": 9.147138803923416e-05,
+      "loss": 0.8165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9693513898788311,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.127538592730519e-05,
+      "loss": 0.7802,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9764789736279401,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.107737211129701e-05,
+      "loss": 0.8668,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9836065573770492,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.087735624212365e-05,
+      "loss": 0.7785,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9907341411261582,
+      "grad_norm": 6.25,
+      "learning_rate": 9.06753480682764e-05,
+      "loss": 0.8952,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9978617248752673,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.047135743534868e-05,
+      "loss": 0.826,
+      "step": 1400
+    },
+    {
+      "epoch": 1.0049893086243764,
+      "grad_norm": 6.625,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.9168,
+      "step": 1410
+    },
+    {
+      "epoch": 1.0121168923734853,
+      "grad_norm": 5.625,
+      "learning_rate": 9.005746865725207e-05,
+      "loss": 0.8506,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0192444761225945,
+      "grad_norm": 4.875,
+      "learning_rate": 8.984759068443831e-05,
+      "loss": 0.9257,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0263720598717034,
+      "grad_norm": 5.96875,
+      "learning_rate": 8.963577059627118e-05,
+      "loss": 0.8302,
+      "step": 1440
+    },
+    {
+      "epoch": 1.0334996436208126,
+      "grad_norm": 5.375,
+      "learning_rate": 8.942201871656291e-05,
+      "loss": 0.7898,
+      "step": 1450
+    },
+    {
+      "epoch": 1.0406272273699215,
+      "grad_norm": 5.75,
+      "learning_rate": 8.920634546327858e-05,
+      "loss": 0.7682,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0477548111190307,
+      "grad_norm": 4.9375,
+      "learning_rate": 8.898876134802826e-05,
+      "loss": 0.8378,
+      "step": 1470
+    },
+    {
+      "epoch": 1.0548823948681396,
+      "grad_norm": 4.15625,
+      "learning_rate": 8.87692769755548e-05,
+      "loss": 0.7113,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0620099786172488,
+      "grad_norm": 5.71875,
+      "learning_rate": 8.854790304321681e-05,
+      "loss": 0.7187,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0691375623663577,
+      "grad_norm": 5.21875,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.7843,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0762651461154669,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.80995297483286e-05,
+      "loss": 0.7253,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0833927298645758,
+      "grad_norm": 4.78125,
+      "learning_rate": 8.787255223886019e-05,
+      "loss": 0.8077,
+      "step": 1520
+    },
+    {
+      "epoch": 1.090520313613685,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.764372887462586e-05,
+      "loss": 0.8465,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0976478973627941,
+      "grad_norm": 4.875,
+      "learning_rate": 8.741307080815357e-05,
+      "loss": 0.7467,
+      "step": 1540
+    },
+    {
+      "epoch": 1.104775481111903,
+      "grad_norm": 5.0,
+      "learning_rate": 8.718058928139205e-05,
+      "loss": 0.7919,
+      "step": 1550
+    },
+    {
+      "epoch": 1.1119030648610122,
+      "grad_norm": 5.78125,
+      "learning_rate": 8.694629562516294e-05,
+      "loss": 0.7399,
+      "step": 1560
+    },
+    {
+      "epoch": 1.1190306486101211,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.671020125860851e-05,
+      "loss": 0.8305,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1261582323592303,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.647231768863512e-05,
+      "loss": 0.8986,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1332858161083392,
+      "grad_norm": 5.65625,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.8041,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1404133998574484,
+      "grad_norm": 5.5625,
+      "learning_rate": 8.599122940150795e-05,
+      "loss": 0.8434,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1475409836065573,
+      "grad_norm": 5.9375,
+      "learning_rate": 8.574804813191858e-05,
+      "loss": 0.7681,
+      "step": 1610
+    },
+    {
+      "epoch": 1.1546685673556665,
+      "grad_norm": 4.71875,
+      "learning_rate": 8.550312455289625e-05,
+      "loss": 0.6835,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1617961511047754,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.525647060167063e-05,
+      "loss": 0.8917,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1689237348538846,
+      "grad_norm": 5.25,
+      "learning_rate": 8.500809829980734e-05,
+      "loss": 0.8296,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1760513186029935,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.4758019752622e-05,
+      "loss": 0.8147,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1831789023521027,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.450624714859016e-05,
+      "loss": 0.7656,
+      "step": 1660
+    },
+    {
+      "epoch": 1.1903064861012118,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.425279275875336e-05,
+      "loss": 0.8931,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1974340698503207,
+      "grad_norm": 4.21875,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.7844,
+      "step": 1680
+    },
+    {
+      "epoch": 1.2045616535994297,
+      "grad_norm": 5.53125,
+      "learning_rate": 8.374088811506819e-05,
+      "loss": 0.7334,
+      "step": 1690
+    },
+    {
+      "epoch": 1.2116892373485388,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.348246281072998e-05,
+      "loss": 0.7509,
+      "step": 1700
+    },
+    {
+      "epoch": 1.218816821097648,
+      "grad_norm": 5.5,
+      "learning_rate": 8.322240561839109e-05,
+      "loss": 0.8287,
+      "step": 1710
+    },
+    {
+      "epoch": 1.225944404846757,
+      "grad_norm": 5.4375,
+      "learning_rate": 8.296072921287217e-05,
+      "loss": 0.8383,
+      "step": 1720
+    },
+    {
+      "epoch": 1.233071988595866,
+      "grad_norm": 6.6875,
+      "learning_rate": 8.269744634791208e-05,
+      "loss": 0.7443,
+      "step": 1730
+    },
+    {
+      "epoch": 1.240199572344975,
+      "grad_norm": 6.0625,
+      "learning_rate": 8.243256985554621e-05,
+      "loss": 0.7868,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2473271560940842,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.21661126454811e-05,
+      "loss": 0.7662,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2544547398431931,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.189808770446528e-05,
+      "loss": 0.7558,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2615823235923023,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.6894,
+      "step": 1770
+    },
+    {
+      "epoch": 1.2687099073414112,
+      "grad_norm": 5.15625,
+      "learning_rate": 8.135738695798376e-05,
+      "loss": 0.7782,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2758374910905204,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.108473750550965e-05,
+      "loss": 0.756,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2829650748396293,
+      "grad_norm": 4.125,
+      "learning_rate": 8.081057302678352e-05,
+      "loss": 0.763,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2900926585887385,
+      "grad_norm": 10.5,
+      "learning_rate": 8.053490688419533e-05,
+      "loss": 0.7255,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2972202423378474,
+      "grad_norm": 4.65625,
+      "learning_rate": 8.02577525133239e-05,
+      "loss": 0.7615,
+      "step": 1820
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 5.15625,
+      "learning_rate": 7.997912342228232e-05,
+      "loss": 0.8092,
+      "step": 1830
+    },
+    {
+      "epoch": 1.3114754098360657,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.969903319105935e-05,
+      "loss": 0.7503,
+      "step": 1840
+    },
+    {
+      "epoch": 1.3186029935851746,
+      "grad_norm": 4.5,
+      "learning_rate": 7.941749547085777e-05,
+      "loss": 0.8121,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3257305773342836,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.8773,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3328581610833927,
+      "grad_norm": 5.6875,
+      "learning_rate": 7.885013252040359e-05,
+      "loss": 0.8202,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3399857448325019,
+      "grad_norm": 6.15625,
+      "learning_rate": 7.856433494262077e-05,
+      "loss": 0.8359,
+      "step": 1880
+    },
+    {
+      "epoch": 1.3471133285816108,
+      "grad_norm": 5.875,
+      "learning_rate": 7.827714517945115e-05,
+      "loss": 0.9009,
+      "step": 1890
+    },
+    {
+      "epoch": 1.3542409123307197,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.798857722811856e-05,
+      "loss": 0.7515,
+      "step": 1900
+    },
+    {
+      "epoch": 1.361368496079829,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.769864515301787e-05,
+      "loss": 0.8321,
+      "step": 1910
+    },
+    {
+      "epoch": 1.368496079828938,
+      "grad_norm": 6.875,
+      "learning_rate": 7.740736308502938e-05,
+      "loss": 0.8211,
+      "step": 1920
+    },
+    {
+      "epoch": 1.375623663578047,
+      "grad_norm": 3.921875,
+      "learning_rate": 7.711474522083015e-05,
+      "loss": 0.7751,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3827512473271562,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.682080582220206e-05,
+      "loss": 0.8174,
+      "step": 1940
+    },
+    {
+      "epoch": 1.389878831076265,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.7058,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3970064148253742,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.622901979013717e-05,
+      "loss": 0.7263,
+      "step": 1960
+    },
+    {
+      "epoch": 1.4041339985744832,
+      "grad_norm": 5.5625,
+      "learning_rate": 7.593120199951671e-05,
+      "loss": 0.7503,
+      "step": 1970
+    },
+    {
+      "epoch": 1.4112615823235923,
+      "grad_norm": 5.34375,
+      "learning_rate": 7.563212035869425e-05,
+      "loss": 0.87,
+      "step": 1980
+    },
+    {
+      "epoch": 1.4183891660727013,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.533178944448705e-05,
+      "loss": 0.7866,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.503022389460014e-05,
+      "loss": 0.787,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.arguana": 0.45921477926341064,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.climate_fever": 0.26026043643827973,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.dbpedia_entity": 0.32492910818341564,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.fever": 0.6243268043774819,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.fiqa": 0.406434177003067,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.hotpotqa": 0.6761769995795371,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.nfcorpus": 0.3490707285646159,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.nq": 0.40743688778055903,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.quora": 0.8329214304594091,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.scidocs": 0.2444219693805352,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.scifact": 0.7389917242185631,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.webis_touche2020": 0.2912623613111957,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/avg": 0.4679539505466725,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4326443335709196,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.472743840691299e-05,
+      "loss": 0.873,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4397719173200285,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.44234477387631e-05,
+      "loss": 0.8482,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4468995010691375,
+      "grad_norm": 4.875,
+      "learning_rate": 7.411826670622676e-05,
+      "loss": 0.8179,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4540270848182466,
+      "grad_norm": 4.40625,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.7916,
+      "step": 2040
+    },
+    {
+      "epoch": 1.4611546685673558,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.350439310165842e-05,
+      "loss": 0.7528,
+      "step": 2050
+    },
+    {
+      "epoch": 1.4682822523164647,
+      "grad_norm": 5.75,
+      "learning_rate": 7.319573044895986e-05,
+      "loss": 0.7749,
+      "step": 2060
+    },
+    {
+      "epoch": 1.4754098360655736,
+      "grad_norm": 5.625,
+      "learning_rate": 7.28859372690835e-05,
+      "loss": 0.8334,
+      "step": 2070
+    },
+    {
+      "epoch": 1.4825374198146828,
+      "grad_norm": 7.03125,
+      "learning_rate": 7.257502866091192e-05,
+      "loss": 0.7086,
+      "step": 2080
+    },
+    {
+      "epoch": 1.489665003563792,
+      "grad_norm": 5.25,
+      "learning_rate": 7.226301977769199e-05,
+      "loss": 0.7891,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4967925873129009,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.194992582629654e-05,
+      "loss": 0.7186,
+      "step": 2100
+    },
+    {
+      "epoch": 1.5039201710620098,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.163576206648304e-05,
+      "loss": 0.778,
+      "step": 2110
+    },
+    {
+      "epoch": 1.511047754811119,
+      "grad_norm": 4.875,
+      "learning_rate": 7.132054381014995e-05,
+      "loss": 0.7799,
+      "step": 2120
+    },
+    {
+      "epoch": 1.5181753385602281,
+      "grad_norm": 6.09375,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.7962,
+      "step": 2130
+    },
+    {
+      "epoch": 1.5253029223093373,
+      "grad_norm": 5.84375,
+      "learning_rate": 7.068700531174319e-05,
+      "loss": 0.8263,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5324305060584462,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.036871594744217e-05,
+      "loss": 0.7541,
+      "step": 2150
+    },
+    {
+      "epoch": 1.5395580898075552,
+      "grad_norm": 5.1875,
+      "learning_rate": 7.004943384066187e-05,
+      "loss": 0.7617,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5466856735566643,
+      "grad_norm": 6.28125,
+      "learning_rate": 6.97291745527617e-05,
+      "loss": 0.8576,
+      "step": 2170
+    },
+    {
+      "epoch": 1.5538132573057735,
+      "grad_norm": 5.6875,
+      "learning_rate": 6.940795369272754e-05,
+      "loss": 0.7801,
+      "step": 2180
+    },
+    {
+      "epoch": 1.5609408410548824,
+      "grad_norm": 5.96875,
+      "learning_rate": 6.908578691641092e-05,
+      "loss": 0.7256,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5680684248039913,
+      "grad_norm": 4.03125,
+      "learning_rate": 6.876268992576604e-05,
+      "loss": 0.7574,
+      "step": 2200
+    },
+    {
+      "epoch": 1.5751960085531005,
+      "grad_norm": 6.65625,
+      "learning_rate": 6.843867846808438e-05,
+      "loss": 0.7539,
+      "step": 2210
+    },
+    {
+      "epoch": 1.5823235923022096,
+      "grad_norm": 6.25,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.7865,
+      "step": 2220
+    },
+    {
+      "epoch": 1.5894511760513186,
+      "grad_norm": 4.90625,
+      "learning_rate": 6.778797536285624e-05,
+      "loss": 0.781,
+      "step": 2230
+    },
+    {
+      "epoch": 1.5965787598004275,
+      "grad_norm": 6.5625,
+      "learning_rate": 6.746131542966112e-05,
+      "loss": 0.9054,
+      "step": 2240
+    },
+    {
+      "epoch": 1.6037063435495367,
+      "grad_norm": 4.0625,
+      "learning_rate": 6.713380445658618e-05,
+      "loss": 0.7958,
+      "step": 2250
+    },
+    {
+      "epoch": 1.6108339272986458,
+      "grad_norm": 5.8125,
+      "learning_rate": 6.680545840605423e-05,
+      "loss": 0.7319,
+      "step": 2260
+    },
+    {
+      "epoch": 1.6179615110477548,
+      "grad_norm": 5.5625,
+      "learning_rate": 6.647629328118851e-05,
+      "loss": 0.7538,
+      "step": 2270
+    },
+    {
+      "epoch": 1.6250890947968637,
+      "grad_norm": 4.625,
+      "learning_rate": 6.614632512503288e-05,
+      "loss": 0.7781,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6322166785459729,
+      "grad_norm": 6.21875,
+      "learning_rate": 6.58155700197697e-05,
+      "loss": 0.7636,
+      "step": 2290
+    },
+    {
+      "epoch": 1.639344262295082,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.548404408593621e-05,
+      "loss": 0.7041,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6464718460441912,
+      "grad_norm": 4.53125,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.8014,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6535994297933,
+      "grad_norm": 4.8125,
+      "learning_rate": 6.481874440176506e-05,
+      "loss": 0.7043,
+      "step": 2320
+    },
+    {
+      "epoch": 1.660727013542409,
+      "grad_norm": 4.78125,
+      "learning_rate": 6.448500307719538e-05,
+      "loss": 0.7739,
+      "step": 2330
+    },
+    {
+      "epoch": 1.6678545972915182,
+      "grad_norm": 5.0625,
+      "learning_rate": 6.415055577401102e-05,
+      "loss": 0.7807,
+      "step": 2340
+    },
+    {
+      "epoch": 1.6749821810406273,
+      "grad_norm": 4.84375,
+      "learning_rate": 6.381541879270169e-05,
+      "loss": 0.8122,
+      "step": 2350
+    },
+    {
+      "epoch": 1.6821097647897363,
+      "grad_norm": 6.25,
+      "learning_rate": 6.347960846737106e-05,
+      "loss": 0.8824,
+      "step": 2360
+    },
+    {
+      "epoch": 1.6892373485388452,
+      "grad_norm": 4.46875,
+      "learning_rate": 6.31431411649406e-05,
+      "loss": 0.6925,
+      "step": 2370
+    },
+    {
+      "epoch": 1.6963649322879544,
+      "grad_norm": 5.59375,
+      "learning_rate": 6.280603328435198e-05,
+      "loss": 0.7558,
+      "step": 2380
+    },
+    {
+      "epoch": 1.7034925160370635,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.24683012557677e-05,
+      "loss": 0.7045,
+      "step": 2390
+    },
+    {
+      "epoch": 1.7106200997861725,
+      "grad_norm": 5.0625,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.7654,
+      "step": 2400
+    },
+    {
+      "epoch": 1.7177476835352814,
+      "grad_norm": 4.90625,
+      "learning_rate": 6.179103062656042e-05,
+      "loss": 0.7696,
+      "step": 2410
+    },
+    {
+      "epoch": 1.7248752672843906,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.145152503515239e-05,
+      "loss": 0.8205,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7320028510334997,
+      "grad_norm": 6.1875,
+      "learning_rate": 6.111146131256983e-05,
+      "loss": 0.7752,
+      "step": 2430
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 4.4375,
+      "learning_rate": 6.077085603303882e-05,
+      "loss": 0.79,
+      "step": 2440
+    },
+    {
+      "epoch": 1.7462580185317176,
+      "grad_norm": 4.75,
+      "learning_rate": 6.042972579718019e-05,
+      "loss": 0.9032,
+      "step": 2450
+    },
+    {
+      "epoch": 1.7533856022808267,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.008808723120035e-05,
+      "loss": 0.7631,
+      "step": 2460
+    },
+    {
+      "epoch": 1.760513186029936,
+      "grad_norm": 4.9375,
+      "learning_rate": 5.974595698608103e-05,
+      "loss": 0.7612,
+      "step": 2470
+    },
+    {
+      "epoch": 1.767640769779045,
+      "grad_norm": 5.125,
+      "learning_rate": 5.9403351736767696e-05,
+      "loss": 0.7827,
+      "step": 2480
+    },
+    {
+      "epoch": 1.774768353528154,
+      "grad_norm": 4.8125,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.7827,
+      "step": 2490
+    },
+    {
+      "epoch": 1.781895937277263,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.8716783040282244e-05,
+      "loss": 0.7646,
+      "step": 2500
+    },
+    {
+      "epoch": 1.789023521026372,
+      "grad_norm": 7.3125,
+      "learning_rate": 5.837285305549978e-05,
+      "loss": 0.8215,
+      "step": 2510
+    },
+    {
+      "epoch": 1.7961511047754812,
+      "grad_norm": 4.84375,
+      "learning_rate": 5.8028514989671724e-05,
+      "loss": 0.7416,
+      "step": 2520
+    },
+    {
+      "epoch": 1.8032786885245902,
+      "grad_norm": 4.3125,
+      "learning_rate": 5.7683785625349616e-05,
+      "loss": 0.7577,
+      "step": 2530
+    },
+    {
+      "epoch": 1.810406272273699,
+      "grad_norm": 4.34375,
+      "learning_rate": 5.733868176415633e-05,
+      "loss": 0.6839,
+      "step": 2540
+    },
+    {
+      "epoch": 1.8175338560228083,
+      "grad_norm": 4.46875,
+      "learning_rate": 5.699322022596722e-05,
+      "loss": 0.719,
+      "step": 2550
+    },
+    {
+      "epoch": 1.8246614397719174,
+      "grad_norm": 5.875,
+      "learning_rate": 5.664741784809022e-05,
+      "loss": 0.7487,
+      "step": 2560
+    },
+    {
+      "epoch": 1.8317890235210263,
+      "grad_norm": 5.40625,
+      "learning_rate": 5.630129148444543e-05,
+      "loss": 0.7946,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8389166072701353,
+      "grad_norm": 5.59375,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.7822,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8460441910192444,
+      "grad_norm": 6.59375,
+      "learning_rate": 5.560813429366345e-05,
+      "loss": 0.7763,
+      "step": 2590
+    },
+    {
+      "epoch": 1.8531717747683536,
+      "grad_norm": 5.90625,
+      "learning_rate": 5.5261137250029835e-05,
+      "loss": 0.8152,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8602993585174625,
+      "grad_norm": 5.53125,
+      "learning_rate": 5.4913883785988993e-05,
+      "loss": 0.8451,
+      "step": 2610
+    },
+    {
+      "epoch": 1.8674269422665715,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.456639082618489e-05,
+      "loss": 0.7819,
+      "step": 2620
+    },
+    {
+      "epoch": 1.8745545260156806,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.421867530693414e-05,
+      "loss": 0.794,
+      "step": 2630
+    },
+    {
+      "epoch": 1.8816821097647898,
+      "grad_norm": 4.8125,
+      "learning_rate": 5.3870754175400595e-05,
+      "loss": 0.7435,
+      "step": 2640
+    },
+    {
+      "epoch": 1.888809693513899,
+      "grad_norm": 6.09375,
+      "learning_rate": 5.352264438876935e-05,
+      "loss": 0.8852,
+      "step": 2650
+    },
+    {
+      "epoch": 1.8959372772630079,
+      "grad_norm": 5.46875,
+      "learning_rate": 5.3174362913420306e-05,
+      "loss": 0.7538,
+      "step": 2660
+    },
+    {
+      "epoch": 1.9030648610121168,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.7795,
+      "step": 2670
+    },
+    {
+      "epoch": 1.910192444761226,
+      "grad_norm": 5.65625,
+      "learning_rate": 5.247735280310041e-05,
+      "loss": 0.7887,
+      "step": 2680
+    },
+    {
+      "epoch": 1.9173200285103351,
+      "grad_norm": 6.09375,
+      "learning_rate": 5.212865813941899e-05,
+      "loss": 0.6838,
+      "step": 2690
+    },
+    {
+      "epoch": 1.924447612259444,
+      "grad_norm": 6.4375,
+      "learning_rate": 5.1779859727942924e-05,
+      "loss": 0.7495,
+      "step": 2700
+    },
+    {
+      "epoch": 1.931575196008553,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.143097456861474e-05,
+      "loss": 0.7499,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9387027797576621,
+      "grad_norm": 5.21875,
+      "learning_rate": 5.1082019665604895e-05,
+      "loss": 0.8243,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9458303635067713,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.073301202648304e-05,
+      "loss": 0.8128,
+      "step": 2730
+    },
+    {
+      "epoch": 1.9529579472558802,
+      "grad_norm": 5.71875,
+      "learning_rate": 5.0383968661389146e-05,
+      "loss": 0.7626,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9600855310049892,
+      "grad_norm": 4.96875,
+      "learning_rate": 5.003490658220438e-05,
+      "loss": 0.7724,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9672131147540983,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.7741,
+      "step": 2760
+    },
+    {
+      "epoch": 1.9743406985032075,
+      "grad_norm": 6.3125,
+      "learning_rate": 4.933679433281836e-05,
+      "loss": 0.7482,
+      "step": 2770
+    },
+    {
+      "epoch": 1.9814682822523164,
+      "grad_norm": 5.6875,
+      "learning_rate": 4.8987778187623245e-05,
+      "loss": 0.7981,
+      "step": 2780
+    },
+    {
+      "epoch": 1.9885958660014256,
+      "grad_norm": 4.875,
+      "learning_rate": 4.863881137669123e-05,
+      "loss": 0.7901,
+      "step": 2790
+    },
+    {
+      "epoch": 1.9957234497505345,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.8289910908172376e-05,
+      "loss": 0.7576,
+      "step": 2800
+    },
+    {
+      "epoch": 2.0028510334996437,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.794109378698327e-05,
+      "loss": 0.7789,
+      "step": 2810
+    },
+    {
+      "epoch": 2.009978617248753,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.7592377013978306e-05,
+      "loss": 0.8136,
+      "step": 2820
+    },
+    {
+      "epoch": 2.0171062009978615,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.724377758512103e-05,
+      "loss": 0.813,
+      "step": 2830
+    },
+    {
+      "epoch": 2.0242337847469707,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.68953124906558e-05,
+      "loss": 0.7948,
+      "step": 2840
+    },
+    {
+      "epoch": 2.03136136849608,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.6773,
+      "step": 2850
+    },
+    {
+      "epoch": 2.038488952245189,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.619885323231484e-05,
+      "loss": 0.6645,
+      "step": 2860
+    },
+    {
+      "epoch": 2.0456165359942977,
+      "grad_norm": 5.34375,
+      "learning_rate": 4.58508930128808e-05,
+      "loss": 0.7572,
+      "step": 2870
+    },
+    {
+      "epoch": 2.052744119743407,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.550313501506781e-05,
+      "loss": 0.7588,
+      "step": 2880
+    },
+    {
+      "epoch": 2.059871703492516,
+      "grad_norm": 5.625,
+      "learning_rate": 4.515559618811005e-05,
+      "loss": 0.7294,
+      "step": 2890
+    },
+    {
+      "epoch": 2.066999287241625,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.4808293470559643e-05,
+      "loss": 0.7621,
+      "step": 2900
+    },
+    {
+      "epoch": 2.0741268709907343,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.446124378946107e-05,
+      "loss": 0.8967,
+      "step": 2910
+    },
+    {
+      "epoch": 2.081254454739843,
+      "grad_norm": 5.3125,
+      "learning_rate": 4.411446405952618e-05,
+      "loss": 0.84,
+      "step": 2920
+    },
+    {
+      "epoch": 2.088382038488952,
+      "grad_norm": 4.375,
+      "learning_rate": 4.376797118230979e-05,
+      "loss": 0.7584,
+      "step": 2930
+    },
+    {
+      "epoch": 2.0955096222380614,
+      "grad_norm": 5.21875,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.8475,
+      "step": 2940
+    },
+    {
+      "epoch": 2.1026372059871705,
+      "grad_norm": 6.40625,
+      "learning_rate": 4.307591352152459e-05,
+      "loss": 0.7067,
+      "step": 2950
+    },
+    {
+      "epoch": 2.1097647897362792,
+      "grad_norm": 6.1875,
+      "learning_rate": 4.273038246786986e-05,
+      "loss": 0.8023,
+      "step": 2960
+    },
+    {
+      "epoch": 2.1168923734853884,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.238520572511773e-05,
+      "loss": 0.7074,
+      "step": 2970
+    },
+    {
+      "epoch": 2.1240199572344975,
+      "grad_norm": 7.5,
+      "learning_rate": 4.2040400116695666e-05,
+      "loss": 0.7498,
+      "step": 2980
+    },
+    {
+      "epoch": 2.1311475409836067,
+      "grad_norm": 7.0625,
+      "learning_rate": 4.169598244794261e-05,
+      "loss": 0.7249,
+      "step": 2990
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "grad_norm": 5.53125,
+      "learning_rate": 4.135196950528982e-05,
+      "loss": 0.7907,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.arguana": 0.4886028250717679,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.climate_fever": 0.2710241560666136,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.dbpedia_entity": 0.3285043023237046,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.fever": 0.6331373355635795,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.fiqa": 0.4150357832110943,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.hotpotqa": 0.68158322360907,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.nfcorpus": 0.3531375464357257,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.nq": 0.41568885251388027,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.quora": 0.8349377083144377,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.scidocs": 0.24315533084930935,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.scifact": 0.7290861351959168,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.webis_touche2020": 0.28783156872457877,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/avg": 0.47347706398997325,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1454027084818246,
+      "grad_norm": 5.75,
+      "learning_rate": 4.100837805544279e-05,
+      "loss": 0.7733,
+      "step": 3010
+    },
+    {
+      "epoch": 2.1525302922309337,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.0665224844564053e-05,
+      "loss": 0.7141,
+      "step": 3020
+    },
+    {
+      "epoch": 2.159657875980043,
+      "grad_norm": 6.8125,
+      "learning_rate": 4.032252659745699e-05,
+      "loss": 0.6352,
+      "step": 3030
+    },
+    {
+      "epoch": 2.1667854597291516,
+      "grad_norm": 4.75,
+      "learning_rate": 3.9980300016750694e-05,
+      "loss": 0.8184,
+      "step": 3040
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 5.0625,
+      "learning_rate": 3.963856178208588e-05,
+      "loss": 0.7785,
+      "step": 3050
+    },
+    {
+      "epoch": 2.18104062722737,
+      "grad_norm": 4.875,
+      "learning_rate": 3.9297328549302e-05,
+      "loss": 0.7565,
+      "step": 3060
+    },
+    {
+      "epoch": 2.188168210976479,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.895661694962542e-05,
+      "loss": 0.7608,
+      "step": 3070
+    },
+    {
+      "epoch": 2.1952957947255882,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.8616443588858805e-05,
+      "loss": 0.7354,
+      "step": 3080
+    },
+    {
+      "epoch": 2.202423378474697,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.827682504657187e-05,
+      "loss": 0.7828,
+      "step": 3090
+    },
+    {
+      "epoch": 2.209550962223806,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.7937777875293244e-05,
+      "loss": 0.7934,
+      "step": 3100
+    },
+    {
+      "epoch": 2.2166785459729152,
+      "grad_norm": 3.859375,
+      "learning_rate": 3.759931859970374e-05,
+      "loss": 0.6331,
+      "step": 3110
+    },
+    {
+      "epoch": 2.2238061297220244,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.72614637158309e-05,
+      "loss": 0.7351,
+      "step": 3120
+    },
+    {
+      "epoch": 2.230933713471133,
+      "grad_norm": 6.15625,
+      "learning_rate": 3.692422969024516e-05,
+      "loss": 0.7856,
+      "step": 3130
+    },
+    {
+      "epoch": 2.2380612972202423,
+      "grad_norm": 5.5,
+      "learning_rate": 3.6587632959257165e-05,
+      "loss": 0.7819,
+      "step": 3140
+    },
+    {
+      "epoch": 2.2451888809693514,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.625168992811671e-05,
+      "loss": 0.7397,
+      "step": 3150
+    },
+    {
+      "epoch": 2.2523164647184606,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.591641697021317e-05,
+      "loss": 0.7588,
+      "step": 3160
+    },
+    {
+      "epoch": 2.2594440484675697,
+      "grad_norm": 5.5625,
+      "learning_rate": 3.558183042627755e-05,
+      "loss": 0.751,
+      "step": 3170
+    },
+    {
+      "epoch": 2.2665716322166785,
+      "grad_norm": 5.46875,
+      "learning_rate": 3.524794660358593e-05,
+      "loss": 0.8684,
+      "step": 3180
+    },
+    {
+      "epoch": 2.2736992159657876,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.491478177516484e-05,
+      "loss": 0.7863,
+      "step": 3190
+    },
+    {
+      "epoch": 2.2808267997148968,
+      "grad_norm": 5.25,
+      "learning_rate": 3.4582352178997935e-05,
+      "loss": 0.7262,
+      "step": 3200
+    },
+    {
+      "epoch": 2.2879543834640055,
+      "grad_norm": 4.0,
+      "learning_rate": 3.425067401723477e-05,
+      "loss": 0.7284,
+      "step": 3210
+    },
+    {
+      "epoch": 2.2950819672131146,
+      "grad_norm": 5.3125,
+      "learning_rate": 3.3919763455401015e-05,
+      "loss": 0.7184,
+      "step": 3220
+    },
+    {
+      "epoch": 2.302209550962224,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.358963662161062e-05,
+      "loss": 0.8039,
+      "step": 3230
+    },
+    {
+      "epoch": 2.309337134711333,
+      "grad_norm": 4.28125,
+      "learning_rate": 3.326030960577972e-05,
+      "loss": 0.7646,
+      "step": 3240
+    },
+    {
+      "epoch": 2.316464718460442,
+      "grad_norm": 5.96875,
+      "learning_rate": 3.293179845884245e-05,
+      "loss": 0.7086,
+      "step": 3250
+    },
+    {
+      "epoch": 2.323592302209551,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.2604119191968654e-05,
+      "loss": 0.8153,
+      "step": 3260
+    },
+    {
+      "epoch": 2.33071988595866,
+      "grad_norm": 6.125,
+      "learning_rate": 3.2277287775783525e-05,
+      "loss": 0.7863,
+      "step": 3270
+    },
+    {
+      "epoch": 2.337847469707769,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.195132013958918e-05,
+      "loss": 0.7569,
+      "step": 3280
+    },
+    {
+      "epoch": 2.3449750534568783,
+      "grad_norm": 4.34375,
+      "learning_rate": 3.162623217058834e-05,
+      "loss": 0.7193,
+      "step": 3290
+    },
+    {
+      "epoch": 2.352102637205987,
+      "grad_norm": 7.15625,
+      "learning_rate": 3.130203971310999e-05,
+      "loss": 0.847,
+      "step": 3300
+    },
+    {
+      "epoch": 2.359230220955096,
+      "grad_norm": 4.6875,
+      "learning_rate": 3.097875856783713e-05,
+      "loss": 0.7884,
+      "step": 3310
+    },
+    {
+      "epoch": 2.3663578047042053,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.065640449103669e-05,
+      "loss": 0.7849,
+      "step": 3320
+    },
+    {
+      "epoch": 2.3734853884533145,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.033499319379163e-05,
+      "loss": 0.7069,
+      "step": 3330
+    },
+    {
+      "epoch": 2.3806129722024236,
+      "grad_norm": 4.28125,
+      "learning_rate": 3.001454034123512e-05,
+      "loss": 0.6801,
+      "step": 3340
+    },
+    {
+      "epoch": 2.3877405559515323,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.9695061551787108e-05,
+      "loss": 0.7755,
+      "step": 3350
+    },
+    {
+      "epoch": 2.3948681397006415,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.9376572396393048e-05,
+      "loss": 0.8031,
+      "step": 3360
+    },
+    {
+      "epoch": 2.4019957234497507,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.9059088397765088e-05,
+      "loss": 0.8501,
+      "step": 3370
+    },
+    {
+      "epoch": 2.4091233071988594,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.874262502962537e-05,
+      "loss": 0.8451,
+      "step": 3380
+    },
+    {
+      "epoch": 2.4162508909479685,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.8427197715952047e-05,
+      "loss": 0.7452,
+      "step": 3390
+    },
+    {
+      "epoch": 2.4233784746970777,
+      "grad_norm": 4.8125,
+      "learning_rate": 2.811282183022736e-05,
+      "loss": 0.6863,
+      "step": 3400
+    },
+    {
+      "epoch": 2.430506058446187,
+      "grad_norm": 5.5,
+      "learning_rate": 2.7799512694688468e-05,
+      "loss": 0.811,
+      "step": 3410
+    },
+    {
+      "epoch": 2.437633642195296,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7487285579580637e-05,
+      "loss": 0.7234,
+      "step": 3420
+    },
+    {
+      "epoch": 2.4447612259444047,
+      "grad_norm": 5.0625,
+      "learning_rate": 2.7176155702412937e-05,
+      "loss": 0.8367,
+      "step": 3430
+    },
+    {
+      "epoch": 2.451888809693514,
+      "grad_norm": 5.90625,
+      "learning_rate": 2.686613822721666e-05,
+      "loss": 0.6122,
+      "step": 3440
+    },
+    {
+      "epoch": 2.459016393442623,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.6557248263806174e-05,
+      "loss": 0.7393,
+      "step": 3450
+    },
+    {
+      "epoch": 2.466143977191732,
+      "grad_norm": 5.0,
+      "learning_rate": 2.6249500867042522e-05,
+      "loss": 0.7783,
+      "step": 3460
+    },
+    {
+      "epoch": 2.473271560940841,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.5942911036099655e-05,
+      "loss": 0.6654,
+      "step": 3470
+    },
+    {
+      "epoch": 2.48039914468995,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.5637493713733374e-05,
+      "loss": 0.752,
+      "step": 3480
+    },
+    {
+      "epoch": 2.487526728439059,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.533326378555314e-05,
+      "loss": 0.7761,
+      "step": 3490
+    },
+    {
+      "epoch": 2.4946543121881684,
+      "grad_norm": 5.53125,
+      "learning_rate": 2.5030236079296444e-05,
+      "loss": 0.7594,
+      "step": 3500
+    },
+    {
+      "epoch": 2.5017818959372775,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.4728425364106135e-05,
+      "loss": 0.7519,
+      "step": 3510
+    },
+    {
+      "epoch": 2.5089094796863862,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.442784634981071e-05,
+      "loss": 0.7527,
+      "step": 3520
+    },
+    {
+      "epoch": 2.5160370634354954,
+      "grad_norm": 4.28125,
+      "learning_rate": 2.412851368620726e-05,
+      "loss": 0.6595,
+      "step": 3530
+    },
+    {
+      "epoch": 2.5231646471846045,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.383044196234753e-05,
+      "loss": 0.7621,
+      "step": 3540
+    },
+    {
+      "epoch": 2.5302922309337132,
+      "grad_norm": 4.28125,
+      "learning_rate": 2.3533645705826806e-05,
+      "loss": 0.7766,
+      "step": 3550
+    },
+    {
+      "epoch": 2.5374198146828224,
+      "grad_norm": 4.625,
+      "learning_rate": 2.323813938207593e-05,
+      "loss": 0.7717,
+      "step": 3560
+    },
+    {
+      "epoch": 2.5445473984319316,
+      "grad_norm": 5.625,
+      "learning_rate": 2.294393739365621e-05,
+      "loss": 0.6604,
+      "step": 3570
+    },
+    {
+      "epoch": 2.5516749821810407,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.265105407955752e-05,
+      "loss": 0.7614,
+      "step": 3580
+    },
+    {
+      "epoch": 2.55880256593015,
+      "grad_norm": 5.59375,
+      "learning_rate": 2.235950371449938e-05,
+      "loss": 0.7702,
+      "step": 3590
+    },
+    {
+      "epoch": 2.5659301496792586,
+      "grad_norm": 5.96875,
+      "learning_rate": 2.2069300508235275e-05,
+      "loss": 0.8866,
+      "step": 3600
+    },
+    {
+      "epoch": 2.5730577334283677,
+      "grad_norm": 6.21875,
+      "learning_rate": 2.1780458604860056e-05,
+      "loss": 0.8359,
+      "step": 3610
+    },
+    {
+      "epoch": 2.580185317177477,
+      "grad_norm": 4.75,
+      "learning_rate": 2.14929920821206e-05,
+      "loss": 0.7036,
+      "step": 3620
+    },
+    {
+      "epoch": 2.5873129009265856,
+      "grad_norm": 6.25,
+      "learning_rate": 2.1206914950729672e-05,
+      "loss": 0.7946,
+      "step": 3630
+    },
+    {
+      "epoch": 2.5944404846756948,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.0922241153683066e-05,
+      "loss": 0.7035,
+      "step": 3640
+    },
+    {
+      "epoch": 2.601568068424804,
+      "grad_norm": 5.0625,
+      "learning_rate": 2.0638984565580023e-05,
+      "loss": 0.7456,
+      "step": 3650
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.035715899194704e-05,
+      "loss": 0.7462,
+      "step": 3660
+    },
+    {
+      "epoch": 2.6158232359230222,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.007677816856498e-05,
+      "loss": 0.7088,
+      "step": 3670
+    },
+    {
+      "epoch": 2.6229508196721314,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9797855760799612e-05,
+      "loss": 0.721,
+      "step": 3680
+    },
+    {
+      "epoch": 2.63007840342124,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.9520405362935594e-05,
+      "loss": 0.7292,
+      "step": 3690
+    },
+    {
+      "epoch": 2.6372059871703493,
+      "grad_norm": 5.0,
+      "learning_rate": 1.9244440497513893e-05,
+      "loss": 0.7031,
+      "step": 3700
+    },
+    {
+      "epoch": 2.6443335709194584,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.896997461467272e-05,
+      "loss": 0.696,
+      "step": 3710
+    },
+    {
+      "epoch": 2.651461154668567,
+      "grad_norm": 4.875,
+      "learning_rate": 1.869702109149199e-05,
+      "loss": 0.749,
+      "step": 3720
+    },
+    {
+      "epoch": 2.6585887384176763,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.842559323134136e-05,
+      "loss": 0.7282,
+      "step": 3730
+    },
+    {
+      "epoch": 2.6657163221667854,
+      "grad_norm": 5.0625,
+      "learning_rate": 1.8155704263231776e-05,
+      "loss": 0.6514,
+      "step": 3740
+    },
+    {
+      "epoch": 2.6728439059158946,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.788736734117078e-05,
+      "loss": 0.7496,
+      "step": 3750
+    },
+    {
+      "epoch": 2.6799714896650038,
+      "grad_norm": 4.5,
+      "learning_rate": 1.762059554352143e-05,
+      "loss": 0.8261,
+      "step": 3760
+    },
+    {
+      "epoch": 2.6870990734141125,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.7355401872364758e-05,
+      "loss": 0.8161,
+      "step": 3770
+    },
+    {
+      "epoch": 2.6942266571632216,
+      "grad_norm": 5.125,
+      "learning_rate": 1.709179925286617e-05,
+      "loss": 0.682,
+      "step": 3780
+    },
+    {
+      "epoch": 2.701354240912331,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.6829800532645447e-05,
+      "loss": 0.7542,
+      "step": 3790
+    },
+    {
+      "epoch": 2.7084818246614395,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.6569418481150595e-05,
+      "loss": 0.8126,
+      "step": 3800
+    },
+    {
+      "epoch": 2.7156094084105487,
+      "grad_norm": 4.6875,
+      "learning_rate": 1.6310665789035467e-05,
+      "loss": 0.7445,
+      "step": 3810
+    },
+    {
+      "epoch": 2.722736992159658,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.605355506754121e-05,
+      "loss": 0.8296,
+      "step": 3820
+    },
+    {
+      "epoch": 2.729864575908767,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.5798098847881666e-05,
+      "loss": 0.6684,
+      "step": 3830
+    },
+    {
+      "epoch": 2.736992159657876,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.554430958063259e-05,
+      "loss": 0.7063,
+      "step": 3840
+    },
+    {
+      "epoch": 2.7441197434069853,
+      "grad_norm": 4.625,
+      "learning_rate": 1.529219963512481e-05,
+      "loss": 0.6668,
+      "step": 3850
+    },
+    {
+      "epoch": 2.751247327156094,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.5041781298841423e-05,
+      "loss": 0.7464,
+      "step": 3860
+    },
+    {
+      "epoch": 2.758374910905203,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.4793066776818842e-05,
+      "loss": 0.7625,
+      "step": 3870
+    },
+    {
+      "epoch": 2.7655024946543123,
+      "grad_norm": 5.625,
+      "learning_rate": 1.4546068191051987e-05,
+      "loss": 0.738,
+      "step": 3880
+    },
+    {
+      "epoch": 2.772630078403421,
+      "grad_norm": 6.125,
+      "learning_rate": 1.4300797579903474e-05,
+      "loss": 0.8185,
+      "step": 3890
+    },
+    {
+      "epoch": 2.77975766215253,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.4057266897516841e-05,
+      "loss": 0.7021,
+      "step": 3900
+    },
+    {
+      "epoch": 2.7868852459016393,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.3815488013233984e-05,
+      "loss": 0.8398,
+      "step": 3910
+    },
+    {
+      "epoch": 2.7940128296507485,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.3575472711016635e-05,
+      "loss": 0.8821,
+      "step": 3920
+    },
+    {
+      "epoch": 2.8011404133998576,
+      "grad_norm": 4.75,
+      "learning_rate": 1.3337232688872009e-05,
+      "loss": 0.7238,
+      "step": 3930
+    },
+    {
+      "epoch": 2.8082679971489664,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.3100779558282672e-05,
+      "loss": 0.6887,
+      "step": 3940
+    },
+    {
+      "epoch": 2.8153955808980755,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.2866124843640615e-05,
+      "loss": 0.7804,
+      "step": 3950
+    },
+    {
+      "epoch": 2.8225231646471847,
+      "grad_norm": 5.78125,
+      "learning_rate": 1.2633279981685608e-05,
+      "loss": 0.871,
+      "step": 3960
+    },
+    {
+      "epoch": 2.8296507483962934,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.240225632094773e-05,
+      "loss": 0.764,
+      "step": 3970
+    },
+    {
+      "epoch": 2.8367783321454025,
+      "grad_norm": 4.65625,
+      "learning_rate": 1.217306512119425e-05,
+      "loss": 0.7754,
+      "step": 3980
+    },
+    {
+      "epoch": 2.8439059158945117,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.194571755288092e-05,
+      "loss": 0.7185,
+      "step": 3990
+    },
+    {
+      "epoch": 2.851033499643621,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.1720224696607474e-05,
+      "loss": 0.6823,
+      "step": 4000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-4000/training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1afec67e7d18e7478623434fcf8c1bcc7719c01c4f7874089687d7138953afff
+size 6097

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/config.json

@@ -0,0 +1,45 @@
+{
+  "architectures": [
+    "ModernBertModel"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 50281,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "embedding_dropout": 0.0,
+  "eos_token_id": 50282,
+  "global_attn_every_n_layers": 3,
+  "global_rope_theta": 160000.0,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "layer_norm_eps": 1e-05,
+  "local_attention": 128,
+  "local_rope_theta": 10000.0,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "repad_logits_with_grad": false,
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.53.0",
+  "vocab_size": 50368
+}

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e1a5059a50ad8539351fafc2820f9d48123be1a23a2be6777263c5223144834
+size 298041696

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:46afb89b2e44d638dea07819f23ad7f4dcae4b7942aab09479a61a4fa5373638
+size 596170443

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/rng_state_0.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:24c12cceaffe8acec04189ba71dbb81f100dd5cd03b86390253117b3fab2911d
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/rng_state_1.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c24e46cf0599262f5ee2ba9ec6bddad7d3e8feb965b424cc9c4dc4563771b1e2
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/rng_state_2.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5af6ed530327c4af91aa04cc19a7523ed8295478ca34bb5f3124ff88cd33f77
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/rng_state_3.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74c82a4770595a28e603ecf5353ab454c0fb9b0e184187b20ebdc4685217f3a5
+size 15365

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23766ee8d5dbc082836e1a591a2e0b269bc9bd5027b36062367e8f61a621f007
+size 1465

modernbert-crux-researchy-pos_20.neg_51.filtered.b64_n512.1e-4/checkpoint-5000/trainer_state.json

@@ -0,0 +1,3859 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.563791874554526,
+  "eval_steps": 1000.0,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0,
+      "eval/nano_beir.arguana": 0.5201586889046041,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.climate_fever": 0.23997409265084552,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.dbpedia_entity": 0.34286887314902237,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fever": 0.6213496755517833,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.fiqa": 0.4542903465876373,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.hotpotqa": 0.6560051657630788,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nfcorpus": 0.3328835143785305,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.nq": 0.4570725655218753,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.quora": 0.832098944676056,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scidocs": 0.25908236715595256,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.scifact": 0.7391066781087818,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/nano_beir.webis_touche2020": 0.30431141846717047,
+      "step": 0
+    },
+    {
+      "epoch": 0,
+      "eval/avg": 0.4799335275762782,
+      "step": 0
+    },
+    {
+      "epoch": 0.007127583749109052,
+      "grad_norm": 11.9375,
+      "learning_rate": 1.8e-06,
+      "loss": 1.5314,
+      "step": 10
+    },
+    {
+      "epoch": 0.014255167498218105,
+      "grad_norm": 11.125,
+      "learning_rate": 3.8e-06,
+      "loss": 1.4333,
+      "step": 20
+    },
+    {
+      "epoch": 0.021382751247327157,
+      "grad_norm": 22.75,
+      "learning_rate": 5.8e-06,
+      "loss": 1.6002,
+      "step": 30
+    },
+    {
+      "epoch": 0.02851033499643621,
+      "grad_norm": 13.3125,
+      "learning_rate": 7.8e-06,
+      "loss": 1.5692,
+      "step": 40
+    },
+    {
+      "epoch": 0.03563791874554526,
+      "grad_norm": 14.125,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 1.2221,
+      "step": 50
+    },
+    {
+      "epoch": 0.042765502494654314,
+      "grad_norm": 12.4375,
+      "learning_rate": 1.18e-05,
+      "loss": 1.3981,
+      "step": 60
+    },
+    {
+      "epoch": 0.04989308624376337,
+      "grad_norm": 11.375,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 1.3844,
+      "step": 70
+    },
+    {
+      "epoch": 0.05702066999287242,
+      "grad_norm": 12.375,
+      "learning_rate": 1.58e-05,
+      "loss": 1.6756,
+      "step": 80
+    },
+    {
+      "epoch": 0.06414825374198146,
+      "grad_norm": 9.8125,
+      "learning_rate": 1.78e-05,
+      "loss": 1.3874,
+      "step": 90
+    },
+    {
+      "epoch": 0.07127583749109052,
+      "grad_norm": 11.5,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 1.3705,
+      "step": 100
+    },
+    {
+      "epoch": 0.07840342124019957,
+      "grad_norm": 10.0,
+      "learning_rate": 2.18e-05,
+      "loss": 1.3897,
+      "step": 110
+    },
+    {
+      "epoch": 0.08553100498930863,
+      "grad_norm": 11.5625,
+      "learning_rate": 2.38e-05,
+      "loss": 1.4046,
+      "step": 120
+    },
+    {
+      "epoch": 0.09265858873841767,
+      "grad_norm": 8.25,
+      "learning_rate": 2.58e-05,
+      "loss": 1.1084,
+      "step": 130
+    },
+    {
+      "epoch": 0.09978617248752673,
+      "grad_norm": 7.96875,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 1.1872,
+      "step": 140
+    },
+    {
+      "epoch": 0.10691375623663578,
+      "grad_norm": 10.8125,
+      "learning_rate": 2.98e-05,
+      "loss": 1.4332,
+      "step": 150
+    },
+    {
+      "epoch": 0.11404133998574484,
+      "grad_norm": 12.8125,
+      "learning_rate": 3.18e-05,
+      "loss": 1.2686,
+      "step": 160
+    },
+    {
+      "epoch": 0.12116892373485388,
+      "grad_norm": 12.625,
+      "learning_rate": 3.38e-05,
+      "loss": 0.9751,
+      "step": 170
+    },
+    {
+      "epoch": 0.12829650748396293,
+      "grad_norm": 9.4375,
+      "learning_rate": 3.58e-05,
+      "loss": 1.1097,
+      "step": 180
+    },
+    {
+      "epoch": 0.13542409123307197,
+      "grad_norm": 10.6875,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 1.0446,
+      "step": 190
+    },
+    {
+      "epoch": 0.14255167498218105,
+      "grad_norm": 8.625,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.2052,
+      "step": 200
+    },
+    {
+      "epoch": 0.1496792587312901,
+      "grad_norm": 9.5625,
+      "learning_rate": 4.18e-05,
+      "loss": 1.0325,
+      "step": 210
+    },
+    {
+      "epoch": 0.15680684248039914,
+      "grad_norm": 9.1875,
+      "learning_rate": 4.38e-05,
+      "loss": 0.9825,
+      "step": 220
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 8.6875,
+      "learning_rate": 4.58e-05,
+      "loss": 1.0758,
+      "step": 230
+    },
+    {
+      "epoch": 0.17106200997861726,
+      "grad_norm": 7.9375,
+      "learning_rate": 4.78e-05,
+      "loss": 1.076,
+      "step": 240
+    },
+    {
+      "epoch": 0.1781895937277263,
+      "grad_norm": 8.25,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.0498,
+      "step": 250
+    },
+    {
+      "epoch": 0.18531717747683535,
+      "grad_norm": 7.9375,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 0.9616,
+      "step": 260
+    },
+    {
+      "epoch": 0.1924447612259444,
+      "grad_norm": 7.71875,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 1.0782,
+      "step": 270
+    },
+    {
+      "epoch": 0.19957234497505347,
+      "grad_norm": 8.6875,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 0.9648,
+      "step": 280
+    },
+    {
+      "epoch": 0.2066999287241625,
+      "grad_norm": 9.375,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 0.9675,
+      "step": 290
+    },
+    {
+      "epoch": 0.21382751247327156,
+      "grad_norm": 7.53125,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 0.9063,
+      "step": 300
+    },
+    {
+      "epoch": 0.2209550962223806,
+      "grad_norm": 6.96875,
+      "learning_rate": 6.18e-05,
+      "loss": 0.8631,
+      "step": 310
+    },
+    {
+      "epoch": 0.22808267997148968,
+      "grad_norm": 7.28125,
+      "learning_rate": 6.38e-05,
+      "loss": 0.9451,
+      "step": 320
+    },
+    {
+      "epoch": 0.23521026372059872,
+      "grad_norm": 8.9375,
+      "learning_rate": 6.58e-05,
+      "loss": 1.0807,
+      "step": 330
+    },
+    {
+      "epoch": 0.24233784746970777,
+      "grad_norm": 7.09375,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 0.9637,
+      "step": 340
+    },
+    {
+      "epoch": 0.2494654312188168,
+      "grad_norm": 6.25,
+      "learning_rate": 6.98e-05,
+      "loss": 0.9216,
+      "step": 350
+    },
+    {
+      "epoch": 0.25659301496792586,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.18e-05,
+      "loss": 0.9588,
+      "step": 360
+    },
+    {
+      "epoch": 0.2637205987170349,
+      "grad_norm": 6.65625,
+      "learning_rate": 7.38e-05,
+      "loss": 0.8696,
+      "step": 370
+    },
+    {
+      "epoch": 0.27084818246614395,
+      "grad_norm": 6.0625,
+      "learning_rate": 7.58e-05,
+      "loss": 0.951,
+      "step": 380
+    },
+    {
+      "epoch": 0.27797576621525305,
+      "grad_norm": 6.6875,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 0.9381,
+      "step": 390
+    },
+    {
+      "epoch": 0.2851033499643621,
+      "grad_norm": 7.3125,
+      "learning_rate": 7.98e-05,
+      "loss": 0.9314,
+      "step": 400
+    },
+    {
+      "epoch": 0.29223093371347114,
+      "grad_norm": 5.84375,
+      "learning_rate": 8.18e-05,
+      "loss": 0.9212,
+      "step": 410
+    },
+    {
+      "epoch": 0.2993585174625802,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.38e-05,
+      "loss": 0.8897,
+      "step": 420
+    },
+    {
+      "epoch": 0.30648610121168923,
+      "grad_norm": 5.46875,
+      "learning_rate": 8.58e-05,
+      "loss": 0.9813,
+      "step": 430
+    },
+    {
+      "epoch": 0.3136136849607983,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.78e-05,
+      "loss": 0.9023,
+      "step": 440
+    },
+    {
+      "epoch": 0.3207412687099073,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.98e-05,
+      "loss": 0.882,
+      "step": 450
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 6.71875,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 0.9943,
+      "step": 460
+    },
+    {
+      "epoch": 0.33499643620812547,
+      "grad_norm": 12.0,
+      "learning_rate": 9.38e-05,
+      "loss": 0.8606,
+      "step": 470
+    },
+    {
+      "epoch": 0.3421240199572345,
+      "grad_norm": 6.125,
+      "learning_rate": 9.58e-05,
+      "loss": 0.9137,
+      "step": 480
+    },
+    {
+      "epoch": 0.34925160370634356,
+      "grad_norm": 6.25,
+      "learning_rate": 9.78e-05,
+      "loss": 0.8677,
+      "step": 490
+    },
+    {
+      "epoch": 0.3563791874554526,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.98e-05,
+      "loss": 0.9161,
+      "step": 500
+    },
+    {
+      "epoch": 0.36350677120456165,
+      "grad_norm": 6.125,
+      "learning_rate": 9.999901304280685e-05,
+      "loss": 0.9176,
+      "step": 510
+    },
+    {
+      "epoch": 0.3706343549536707,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.999560138895238e-05,
+      "loss": 0.871,
+      "step": 520
+    },
+    {
+      "epoch": 0.37776193870277974,
+      "grad_norm": 6.28125,
+      "learning_rate": 9.99897530200195e-05,
+      "loss": 0.9097,
+      "step": 530
+    },
+    {
+      "epoch": 0.3848895224518888,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.998146822104943e-05,
+      "loss": 0.8721,
+      "step": 540
+    },
+    {
+      "epoch": 0.3920171062009979,
+      "grad_norm": 7.15625,
+      "learning_rate": 9.997074739583162e-05,
+      "loss": 0.9282,
+      "step": 550
+    },
+    {
+      "epoch": 0.39914468995010693,
+      "grad_norm": 5.375,
+      "learning_rate": 9.995759106688393e-05,
+      "loss": 0.8213,
+      "step": 560
+    },
+    {
+      "epoch": 0.406272273699216,
+      "grad_norm": 7.28125,
+      "learning_rate": 9.99419998754273e-05,
+      "loss": 0.8527,
+      "step": 570
+    },
+    {
+      "epoch": 0.413399857448325,
+      "grad_norm": 5.75,
+      "learning_rate": 9.992397458135438e-05,
+      "loss": 0.944,
+      "step": 580
+    },
+    {
+      "epoch": 0.42052744119743407,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.990351606319261e-05,
+      "loss": 0.9269,
+      "step": 590
+    },
+    {
+      "epoch": 0.4276550249465431,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.988062531806126e-05,
+      "loss": 0.8261,
+      "step": 600
+    },
+    {
+      "epoch": 0.43478260869565216,
+      "grad_norm": 6.4375,
+      "learning_rate": 9.985530346162299e-05,
+      "loss": 1.0049,
+      "step": 610
+    },
+    {
+      "epoch": 0.4419101924447612,
+      "grad_norm": 6.65625,
+      "learning_rate": 9.982755172802933e-05,
+      "loss": 0.9438,
+      "step": 620
+    },
+    {
+      "epoch": 0.44903777619387025,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.979737146986064e-05,
+      "loss": 0.9527,
+      "step": 630
+    },
+    {
+      "epoch": 0.45616535994297935,
+      "grad_norm": 6.375,
+      "learning_rate": 9.976476415806013e-05,
+      "loss": 0.8917,
+      "step": 640
+    },
+    {
+      "epoch": 0.4632929436920884,
+      "grad_norm": 4.625,
+      "learning_rate": 9.972973138186217e-05,
+      "loss": 0.8554,
+      "step": 650
+    },
+    {
+      "epoch": 0.47042052744119744,
+      "grad_norm": 5.59375,
+      "learning_rate": 9.969227484871484e-05,
+      "loss": 0.978,
+      "step": 660
+    },
+    {
+      "epoch": 0.4775481111903065,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.965239638419672e-05,
+      "loss": 0.774,
+      "step": 670
+    },
+    {
+      "epoch": 0.48467569493941554,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.961009793192792e-05,
+      "loss": 0.8388,
+      "step": 680
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.956538155347534e-05,
+      "loss": 0.8888,
+      "step": 690
+    },
+    {
+      "epoch": 0.4989308624376336,
+      "grad_norm": 6.03125,
+      "learning_rate": 9.951824942825215e-05,
+      "loss": 0.7606,
+      "step": 700
+    },
+    {
+      "epoch": 0.5060584461867427,
+      "grad_norm": 5.1875,
+      "learning_rate": 9.946870385341167e-05,
+      "loss": 0.9804,
+      "step": 710
+    },
+    {
+      "epoch": 0.5131860299358517,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.941674724373531e-05,
+      "loss": 0.8832,
+      "step": 720
+    },
+    {
+      "epoch": 0.5203136136849608,
+      "grad_norm": 5.0,
+      "learning_rate": 9.936238213151491e-05,
+      "loss": 0.8385,
+      "step": 730
+    },
+    {
+      "epoch": 0.5274411974340698,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.930561116642935e-05,
+      "loss": 0.9106,
+      "step": 740
+    },
+    {
+      "epoch": 0.5345687811831789,
+      "grad_norm": 6.3125,
+      "learning_rate": 9.924643711541539e-05,
+      "loss": 0.807,
+      "step": 750
+    },
+    {
+      "epoch": 0.5416963649322879,
+      "grad_norm": 4.0,
+      "learning_rate": 9.918486286253279e-05,
+      "loss": 0.7932,
+      "step": 760
+    },
+    {
+      "epoch": 0.5488239486813971,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.912089140882376e-05,
+      "loss": 0.7533,
+      "step": 770
+    },
+    {
+      "epoch": 0.5559515324305061,
+      "grad_norm": 5.125,
+      "learning_rate": 9.90545258721667e-05,
+      "loss": 0.8676,
+      "step": 780
+    },
+    {
+      "epoch": 0.5630791161796151,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.898576948712426e-05,
+      "loss": 0.8783,
+      "step": 790
+    },
+    {
+      "epoch": 0.5702066999287242,
+      "grad_norm": 5.5625,
+      "learning_rate": 9.891462560478562e-05,
+      "loss": 0.9352,
+      "step": 800
+    },
+    {
+      "epoch": 0.5773342836778332,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.884109769260325e-05,
+      "loss": 0.8814,
+      "step": 810
+    },
+    {
+      "epoch": 0.5844618674269423,
+      "grad_norm": 5.96875,
+      "learning_rate": 9.876518933422386e-05,
+      "loss": 0.8252,
+      "step": 820
+    },
+    {
+      "epoch": 0.5915894511760513,
+      "grad_norm": 4.96875,
+      "learning_rate": 9.868690422931372e-05,
+      "loss": 0.8695,
+      "step": 830
+    },
+    {
+      "epoch": 0.5987170349251604,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.860624619337844e-05,
+      "loss": 0.8821,
+      "step": 840
+    },
+    {
+      "epoch": 0.6058446186742694,
+      "grad_norm": 6.375,
+      "learning_rate": 9.852321915757687e-05,
+      "loss": 0.8968,
+      "step": 850
+    },
+    {
+      "epoch": 0.6129722024233785,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.843782716852963e-05,
+      "loss": 0.9109,
+      "step": 860
+    },
+    {
+      "epoch": 0.6200997861724875,
+      "grad_norm": 6.625,
+      "learning_rate": 9.835007438812177e-05,
+      "loss": 0.8701,
+      "step": 870
+    },
+    {
+      "epoch": 0.6272273699215966,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.82599650933e-05,
+      "loss": 0.8649,
+      "step": 880
+    },
+    {
+      "epoch": 0.6343549536707056,
+      "grad_norm": 5.25,
+      "learning_rate": 9.816750367586425e-05,
+      "loss": 0.8578,
+      "step": 890
+    },
+    {
+      "epoch": 0.6414825374198146,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.807269464225355e-05,
+      "loss": 0.9067,
+      "step": 900
+    },
+    {
+      "epoch": 0.6486101211689237,
+      "grad_norm": 6.8125,
+      "learning_rate": 9.797554261332645e-05,
+      "loss": 0.7862,
+      "step": 910
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 5.875,
+      "learning_rate": 9.787605232413574e-05,
+      "loss": 0.8275,
+      "step": 920
+    },
+    {
+      "epoch": 0.6628652886671419,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.777422862369783e-05,
+      "loss": 0.9155,
+      "step": 930
+    },
+    {
+      "epoch": 0.6699928724162509,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.767007647475617e-05,
+      "loss": 0.8526,
+      "step": 940
+    },
+    {
+      "epoch": 0.67712045616536,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.756360095353956e-05,
+      "loss": 0.8123,
+      "step": 950
+    },
+    {
+      "epoch": 0.684248039914469,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.745480724951473e-05,
+      "loss": 0.7986,
+      "step": 960
+    },
+    {
+      "epoch": 0.6913756236635781,
+      "grad_norm": 5.6875,
+      "learning_rate": 9.73437006651333e-05,
+      "loss": 0.9203,
+      "step": 970
+    },
+    {
+      "epoch": 0.6985032074126871,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.723028661557345e-05,
+      "loss": 0.7805,
+      "step": 980
+    },
+    {
+      "epoch": 0.7056307911617962,
+      "grad_norm": 5.4375,
+      "learning_rate": 9.711457062847595e-05,
+      "loss": 0.8683,
+      "step": 990
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "grad_norm": 5.46875,
+      "learning_rate": 9.699655834367478e-05,
+      "loss": 0.8515,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.arguana": 0.49904762134680597,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.climate_fever": 0.2710368630075979,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.dbpedia_entity": 0.32627448935866016,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fever": 0.6251056136949217,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.fiqa": 0.4264627976119544,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.hotpotqa": 0.67890984670666,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nfcorpus": 0.33575253271342886,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.nq": 0.41529049449791683,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.quora": 0.8400060722791932,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scidocs": 0.25236583848912914,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.scifact": 0.7248783076751154,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/nano_beir.webis_touche2020": 0.29804501841705877,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7127583749109052,
+      "eval/avg": 0.4744312913165369,
+      "step": 1000
+    },
+    {
+      "epoch": 0.7198859586600143,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.68762555129222e-05,
+      "loss": 0.8998,
+      "step": 1010
+    },
+    {
+      "epoch": 0.7270135424091233,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.675366799960841e-05,
+      "loss": 0.8763,
+      "step": 1020
+    },
+    {
+      "epoch": 0.7341411261582323,
+      "grad_norm": 4.875,
+      "learning_rate": 9.662880177847594e-05,
+      "loss": 0.8851,
+      "step": 1030
+    },
+    {
+      "epoch": 0.7412687099073414,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.65016629353282e-05,
+      "loss": 0.8247,
+      "step": 1040
+    },
+    {
+      "epoch": 0.7483962936564504,
+      "grad_norm": 6.5625,
+      "learning_rate": 9.637225766673307e-05,
+      "loss": 0.742,
+      "step": 1050
+    },
+    {
+      "epoch": 0.7555238774055595,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.624059227972075e-05,
+      "loss": 0.8707,
+      "step": 1060
+    },
+    {
+      "epoch": 0.7626514611546685,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.610667319147647e-05,
+      "loss": 0.7698,
+      "step": 1070
+    },
+    {
+      "epoch": 0.7697790449037776,
+      "grad_norm": 6.1875,
+      "learning_rate": 9.597050692902765e-05,
+      "loss": 0.9318,
+      "step": 1080
+    },
+    {
+      "epoch": 0.7769066286528866,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.583210012892582e-05,
+      "loss": 0.752,
+      "step": 1090
+    },
+    {
+      "epoch": 0.7840342124019958,
+      "grad_norm": 6.0,
+      "learning_rate": 9.569145953692315e-05,
+      "loss": 0.9631,
+      "step": 1100
+    },
+    {
+      "epoch": 0.7911617961511048,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.55485920076437e-05,
+      "loss": 0.8623,
+      "step": 1110
+    },
+    {
+      "epoch": 0.7982893799002139,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.540350450424926e-05,
+      "loss": 0.8677,
+      "step": 1120
+    },
+    {
+      "epoch": 0.8054169636493229,
+      "grad_norm": 4.03125,
+      "learning_rate": 9.525620409810007e-05,
+      "loss": 0.7795,
+      "step": 1130
+    },
+    {
+      "epoch": 0.812544547398432,
+      "grad_norm": 4.125,
+      "learning_rate": 9.510669796841014e-05,
+      "loss": 0.8011,
+      "step": 1140
+    },
+    {
+      "epoch": 0.819672131147541,
+      "grad_norm": 4.875,
+      "learning_rate": 9.495499340189728e-05,
+      "loss": 0.7984,
+      "step": 1150
+    },
+    {
+      "epoch": 0.82679971489665,
+      "grad_norm": 4.84375,
+      "learning_rate": 9.480109779242805e-05,
+      "loss": 0.8049,
+      "step": 1160
+    },
+    {
+      "epoch": 0.8339272986457591,
+      "grad_norm": 5.15625,
+      "learning_rate": 9.464501864065735e-05,
+      "loss": 0.7839,
+      "step": 1170
+    },
+    {
+      "epoch": 0.8410548823948681,
+      "grad_norm": 4.78125,
+      "learning_rate": 9.448676355366282e-05,
+      "loss": 0.7809,
+      "step": 1180
+    },
+    {
+      "epoch": 0.8481824661439772,
+      "grad_norm": 4.71875,
+      "learning_rate": 9.432634024457414e-05,
+      "loss": 0.7688,
+      "step": 1190
+    },
+    {
+      "epoch": 0.8553100498930862,
+      "grad_norm": 7.75,
+      "learning_rate": 9.416375653219709e-05,
+      "loss": 0.8638,
+      "step": 1200
+    },
+    {
+      "epoch": 0.8624376336421953,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.399902034063242e-05,
+      "loss": 0.8409,
+      "step": 1210
+    },
+    {
+      "epoch": 0.8695652173913043,
+      "grad_norm": 5.875,
+      "learning_rate": 9.383213969888972e-05,
+      "loss": 0.916,
+      "step": 1220
+    },
+    {
+      "epoch": 0.8766928011404134,
+      "grad_norm": 5.40625,
+      "learning_rate": 9.366312274049602e-05,
+      "loss": 0.9114,
+      "step": 1230
+    },
+    {
+      "epoch": 0.8838203848895224,
+      "grad_norm": 4.5,
+      "learning_rate": 9.34919777030994e-05,
+      "loss": 0.8744,
+      "step": 1240
+    },
+    {
+      "epoch": 0.8909479686386315,
+      "grad_norm": 4.9375,
+      "learning_rate": 9.331871292806759e-05,
+      "loss": 0.8471,
+      "step": 1250
+    },
+    {
+      "epoch": 0.8980755523877405,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.314333686008125e-05,
+      "loss": 0.8014,
+      "step": 1260
+    },
+    {
+      "epoch": 0.9052031361368497,
+      "grad_norm": 5.71875,
+      "learning_rate": 9.296585804672252e-05,
+      "loss": 0.8587,
+      "step": 1270
+    },
+    {
+      "epoch": 0.9123307198859587,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.278628513805838e-05,
+      "loss": 0.8003,
+      "step": 1280
+    },
+    {
+      "epoch": 0.9194583036350678,
+      "grad_norm": 6.09375,
+      "learning_rate": 9.260462688621905e-05,
+      "loss": 0.7404,
+      "step": 1290
+    },
+    {
+      "epoch": 0.9265858873841768,
+      "grad_norm": 5.5,
+      "learning_rate": 9.242089214497145e-05,
+      "loss": 0.722,
+      "step": 1300
+    },
+    {
+      "epoch": 0.9337134711332858,
+      "grad_norm": 5.21875,
+      "learning_rate": 9.223508986928766e-05,
+      "loss": 0.7005,
+      "step": 1310
+    },
+    {
+      "epoch": 0.9408410548823949,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.204722911490846e-05,
+      "loss": 0.7987,
+      "step": 1320
+    },
+    {
+      "epoch": 0.9479686386315039,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.1857319037902e-05,
+      "loss": 0.7872,
+      "step": 1330
+    },
+    {
+      "epoch": 0.955096222380613,
+      "grad_norm": 4.875,
+      "learning_rate": 9.16653688942175e-05,
+      "loss": 0.8632,
+      "step": 1340
+    },
+    {
+      "epoch": 0.962223806129722,
+      "grad_norm": 6.96875,
+      "learning_rate": 9.147138803923416e-05,
+      "loss": 0.8165,
+      "step": 1350
+    },
+    {
+      "epoch": 0.9693513898788311,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.127538592730519e-05,
+      "loss": 0.7802,
+      "step": 1360
+    },
+    {
+      "epoch": 0.9764789736279401,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.107737211129701e-05,
+      "loss": 0.8668,
+      "step": 1370
+    },
+    {
+      "epoch": 0.9836065573770492,
+      "grad_norm": 5.84375,
+      "learning_rate": 9.087735624212365e-05,
+      "loss": 0.7785,
+      "step": 1380
+    },
+    {
+      "epoch": 0.9907341411261582,
+      "grad_norm": 6.25,
+      "learning_rate": 9.06753480682764e-05,
+      "loss": 0.8952,
+      "step": 1390
+    },
+    {
+      "epoch": 0.9978617248752673,
+      "grad_norm": 6.34375,
+      "learning_rate": 9.047135743534868e-05,
+      "loss": 0.826,
+      "step": 1400
+    },
+    {
+      "epoch": 1.0049893086243764,
+      "grad_norm": 6.625,
+      "learning_rate": 9.02653942855561e-05,
+      "loss": 0.9168,
+      "step": 1410
+    },
+    {
+      "epoch": 1.0121168923734853,
+      "grad_norm": 5.625,
+      "learning_rate": 9.005746865725207e-05,
+      "loss": 0.8506,
+      "step": 1420
+    },
+    {
+      "epoch": 1.0192444761225945,
+      "grad_norm": 4.875,
+      "learning_rate": 8.984759068443831e-05,
+      "loss": 0.9257,
+      "step": 1430
+    },
+    {
+      "epoch": 1.0263720598717034,
+      "grad_norm": 5.96875,
+      "learning_rate": 8.963577059627118e-05,
+      "loss": 0.8302,
+      "step": 1440
+    },
+    {
+      "epoch": 1.0334996436208126,
+      "grad_norm": 5.375,
+      "learning_rate": 8.942201871656291e-05,
+      "loss": 0.7898,
+      "step": 1450
+    },
+    {
+      "epoch": 1.0406272273699215,
+      "grad_norm": 5.75,
+      "learning_rate": 8.920634546327858e-05,
+      "loss": 0.7682,
+      "step": 1460
+    },
+    {
+      "epoch": 1.0477548111190307,
+      "grad_norm": 4.9375,
+      "learning_rate": 8.898876134802826e-05,
+      "loss": 0.8378,
+      "step": 1470
+    },
+    {
+      "epoch": 1.0548823948681396,
+      "grad_norm": 4.15625,
+      "learning_rate": 8.87692769755548e-05,
+      "loss": 0.7113,
+      "step": 1480
+    },
+    {
+      "epoch": 1.0620099786172488,
+      "grad_norm": 5.71875,
+      "learning_rate": 8.854790304321681e-05,
+      "loss": 0.7187,
+      "step": 1490
+    },
+    {
+      "epoch": 1.0691375623663577,
+      "grad_norm": 5.21875,
+      "learning_rate": 8.832465034046749e-05,
+      "loss": 0.7843,
+      "step": 1500
+    },
+    {
+      "epoch": 1.0762651461154669,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.80995297483286e-05,
+      "loss": 0.7253,
+      "step": 1510
+    },
+    {
+      "epoch": 1.0833927298645758,
+      "grad_norm": 4.78125,
+      "learning_rate": 8.787255223886019e-05,
+      "loss": 0.8077,
+      "step": 1520
+    },
+    {
+      "epoch": 1.090520313613685,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.764372887462586e-05,
+      "loss": 0.8465,
+      "step": 1530
+    },
+    {
+      "epoch": 1.0976478973627941,
+      "grad_norm": 4.875,
+      "learning_rate": 8.741307080815357e-05,
+      "loss": 0.7467,
+      "step": 1540
+    },
+    {
+      "epoch": 1.104775481111903,
+      "grad_norm": 5.0,
+      "learning_rate": 8.718058928139205e-05,
+      "loss": 0.7919,
+      "step": 1550
+    },
+    {
+      "epoch": 1.1119030648610122,
+      "grad_norm": 5.78125,
+      "learning_rate": 8.694629562516294e-05,
+      "loss": 0.7399,
+      "step": 1560
+    },
+    {
+      "epoch": 1.1190306486101211,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.671020125860851e-05,
+      "loss": 0.8305,
+      "step": 1570
+    },
+    {
+      "epoch": 1.1261582323592303,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.647231768863512e-05,
+      "loss": 0.8986,
+      "step": 1580
+    },
+    {
+      "epoch": 1.1332858161083392,
+      "grad_norm": 5.65625,
+      "learning_rate": 8.623265650935234e-05,
+      "loss": 0.8041,
+      "step": 1590
+    },
+    {
+      "epoch": 1.1404133998574484,
+      "grad_norm": 5.5625,
+      "learning_rate": 8.599122940150795e-05,
+      "loss": 0.8434,
+      "step": 1600
+    },
+    {
+      "epoch": 1.1475409836065573,
+      "grad_norm": 5.9375,
+      "learning_rate": 8.574804813191858e-05,
+      "loss": 0.7681,
+      "step": 1610
+    },
+    {
+      "epoch": 1.1546685673556665,
+      "grad_norm": 4.71875,
+      "learning_rate": 8.550312455289625e-05,
+      "loss": 0.6835,
+      "step": 1620
+    },
+    {
+      "epoch": 1.1617961511047754,
+      "grad_norm": 4.8125,
+      "learning_rate": 8.525647060167063e-05,
+      "loss": 0.8917,
+      "step": 1630
+    },
+    {
+      "epoch": 1.1689237348538846,
+      "grad_norm": 5.25,
+      "learning_rate": 8.500809829980734e-05,
+      "loss": 0.8296,
+      "step": 1640
+    },
+    {
+      "epoch": 1.1760513186029935,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.4758019752622e-05,
+      "loss": 0.8147,
+      "step": 1650
+    },
+    {
+      "epoch": 1.1831789023521027,
+      "grad_norm": 6.90625,
+      "learning_rate": 8.450624714859016e-05,
+      "loss": 0.7656,
+      "step": 1660
+    },
+    {
+      "epoch": 1.1903064861012118,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.425279275875336e-05,
+      "loss": 0.8931,
+      "step": 1670
+    },
+    {
+      "epoch": 1.1974340698503207,
+      "grad_norm": 4.21875,
+      "learning_rate": 8.399766893612096e-05,
+      "loss": 0.7844,
+      "step": 1680
+    },
+    {
+      "epoch": 1.2045616535994297,
+      "grad_norm": 5.53125,
+      "learning_rate": 8.374088811506819e-05,
+      "loss": 0.7334,
+      "step": 1690
+    },
+    {
+      "epoch": 1.2116892373485388,
+      "grad_norm": 5.28125,
+      "learning_rate": 8.348246281072998e-05,
+      "loss": 0.7509,
+      "step": 1700
+    },
+    {
+      "epoch": 1.218816821097648,
+      "grad_norm": 5.5,
+      "learning_rate": 8.322240561839109e-05,
+      "loss": 0.8287,
+      "step": 1710
+    },
+    {
+      "epoch": 1.225944404846757,
+      "grad_norm": 5.4375,
+      "learning_rate": 8.296072921287217e-05,
+      "loss": 0.8383,
+      "step": 1720
+    },
+    {
+      "epoch": 1.233071988595866,
+      "grad_norm": 6.6875,
+      "learning_rate": 8.269744634791208e-05,
+      "loss": 0.7443,
+      "step": 1730
+    },
+    {
+      "epoch": 1.240199572344975,
+      "grad_norm": 6.0625,
+      "learning_rate": 8.243256985554621e-05,
+      "loss": 0.7868,
+      "step": 1740
+    },
+    {
+      "epoch": 1.2473271560940842,
+      "grad_norm": 5.09375,
+      "learning_rate": 8.21661126454811e-05,
+      "loss": 0.7662,
+      "step": 1750
+    },
+    {
+      "epoch": 1.2544547398431931,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.189808770446528e-05,
+      "loss": 0.7558,
+      "step": 1760
+    },
+    {
+      "epoch": 1.2615823235923023,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.162850809565623e-05,
+      "loss": 0.6894,
+      "step": 1770
+    },
+    {
+      "epoch": 1.2687099073414112,
+      "grad_norm": 5.15625,
+      "learning_rate": 8.135738695798376e-05,
+      "loss": 0.7782,
+      "step": 1780
+    },
+    {
+      "epoch": 1.2758374910905204,
+      "grad_norm": 5.34375,
+      "learning_rate": 8.108473750550965e-05,
+      "loss": 0.756,
+      "step": 1790
+    },
+    {
+      "epoch": 1.2829650748396293,
+      "grad_norm": 4.125,
+      "learning_rate": 8.081057302678352e-05,
+      "loss": 0.763,
+      "step": 1800
+    },
+    {
+      "epoch": 1.2900926585887385,
+      "grad_norm": 10.5,
+      "learning_rate": 8.053490688419533e-05,
+      "loss": 0.7255,
+      "step": 1810
+    },
+    {
+      "epoch": 1.2972202423378474,
+      "grad_norm": 4.65625,
+      "learning_rate": 8.02577525133239e-05,
+      "loss": 0.7615,
+      "step": 1820
+    },
+    {
+      "epoch": 1.3043478260869565,
+      "grad_norm": 5.15625,
+      "learning_rate": 7.997912342228232e-05,
+      "loss": 0.8092,
+      "step": 1830
+    },
+    {
+      "epoch": 1.3114754098360657,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.969903319105935e-05,
+      "loss": 0.7503,
+      "step": 1840
+    },
+    {
+      "epoch": 1.3186029935851746,
+      "grad_norm": 4.5,
+      "learning_rate": 7.941749547085777e-05,
+      "loss": 0.8121,
+      "step": 1850
+    },
+    {
+      "epoch": 1.3257305773342836,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.913452398342881e-05,
+      "loss": 0.8773,
+      "step": 1860
+    },
+    {
+      "epoch": 1.3328581610833927,
+      "grad_norm": 5.6875,
+      "learning_rate": 7.885013252040359e-05,
+      "loss": 0.8202,
+      "step": 1870
+    },
+    {
+      "epoch": 1.3399857448325019,
+      "grad_norm": 6.15625,
+      "learning_rate": 7.856433494262077e-05,
+      "loss": 0.8359,
+      "step": 1880
+    },
+    {
+      "epoch": 1.3471133285816108,
+      "grad_norm": 5.875,
+      "learning_rate": 7.827714517945115e-05,
+      "loss": 0.9009,
+      "step": 1890
+    },
+    {
+      "epoch": 1.3542409123307197,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.798857722811856e-05,
+      "loss": 0.7515,
+      "step": 1900
+    },
+    {
+      "epoch": 1.361368496079829,
+      "grad_norm": 4.96875,
+      "learning_rate": 7.769864515301787e-05,
+      "loss": 0.8321,
+      "step": 1910
+    },
+    {
+      "epoch": 1.368496079828938,
+      "grad_norm": 6.875,
+      "learning_rate": 7.740736308502938e-05,
+      "loss": 0.8211,
+      "step": 1920
+    },
+    {
+      "epoch": 1.375623663578047,
+      "grad_norm": 3.921875,
+      "learning_rate": 7.711474522083015e-05,
+      "loss": 0.7751,
+      "step": 1930
+    },
+    {
+      "epoch": 1.3827512473271562,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.682080582220206e-05,
+      "loss": 0.8174,
+      "step": 1940
+    },
+    {
+      "epoch": 1.389878831076265,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.65255592153367e-05,
+      "loss": 0.7058,
+      "step": 1950
+    },
+    {
+      "epoch": 1.3970064148253742,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.622901979013717e-05,
+      "loss": 0.7263,
+      "step": 1960
+    },
+    {
+      "epoch": 1.4041339985744832,
+      "grad_norm": 5.5625,
+      "learning_rate": 7.593120199951671e-05,
+      "loss": 0.7503,
+      "step": 1970
+    },
+    {
+      "epoch": 1.4112615823235923,
+      "grad_norm": 5.34375,
+      "learning_rate": 7.563212035869425e-05,
+      "loss": 0.87,
+      "step": 1980
+    },
+    {
+      "epoch": 1.4183891660727013,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.533178944448705e-05,
+      "loss": 0.7866,
+      "step": 1990
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.503022389460014e-05,
+      "loss": 0.787,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.arguana": 0.45921477926341064,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.climate_fever": 0.26026043643827973,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.dbpedia_entity": 0.32492910818341564,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.fever": 0.6243268043774819,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.fiqa": 0.406434177003067,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.hotpotqa": 0.6761769995795371,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.nfcorpus": 0.3490707285646159,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.nq": 0.40743688778055903,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.quora": 0.8329214304594091,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.scidocs": 0.2444219693805352,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.scifact": 0.7389917242185631,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/nano_beir.webis_touche2020": 0.2912623613111957,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4255167498218104,
+      "eval/avg": 0.4679539505466725,
+      "step": 2000
+    },
+    {
+      "epoch": 1.4326443335709196,
+      "grad_norm": 5.53125,
+      "learning_rate": 7.472743840691299e-05,
+      "loss": 0.873,
+      "step": 2010
+    },
+    {
+      "epoch": 1.4397719173200285,
+      "grad_norm": 6.34375,
+      "learning_rate": 7.44234477387631e-05,
+      "loss": 0.8482,
+      "step": 2020
+    },
+    {
+      "epoch": 1.4468995010691375,
+      "grad_norm": 4.875,
+      "learning_rate": 7.411826670622676e-05,
+      "loss": 0.8179,
+      "step": 2030
+    },
+    {
+      "epoch": 1.4540270848182466,
+      "grad_norm": 4.40625,
+      "learning_rate": 7.381191018339696e-05,
+      "loss": 0.7916,
+      "step": 2040
+    },
+    {
+      "epoch": 1.4611546685673558,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.350439310165842e-05,
+      "loss": 0.7528,
+      "step": 2050
+    },
+    {
+      "epoch": 1.4682822523164647,
+      "grad_norm": 5.75,
+      "learning_rate": 7.319573044895986e-05,
+      "loss": 0.7749,
+      "step": 2060
+    },
+    {
+      "epoch": 1.4754098360655736,
+      "grad_norm": 5.625,
+      "learning_rate": 7.28859372690835e-05,
+      "loss": 0.8334,
+      "step": 2070
+    },
+    {
+      "epoch": 1.4825374198146828,
+      "grad_norm": 7.03125,
+      "learning_rate": 7.257502866091192e-05,
+      "loss": 0.7086,
+      "step": 2080
+    },
+    {
+      "epoch": 1.489665003563792,
+      "grad_norm": 5.25,
+      "learning_rate": 7.226301977769199e-05,
+      "loss": 0.7891,
+      "step": 2090
+    },
+    {
+      "epoch": 1.4967925873129009,
+      "grad_norm": 5.0625,
+      "learning_rate": 7.194992582629654e-05,
+      "loss": 0.7186,
+      "step": 2100
+    },
+    {
+      "epoch": 1.5039201710620098,
+      "grad_norm": 5.28125,
+      "learning_rate": 7.163576206648304e-05,
+      "loss": 0.778,
+      "step": 2110
+    },
+    {
+      "epoch": 1.511047754811119,
+      "grad_norm": 4.875,
+      "learning_rate": 7.132054381014995e-05,
+      "loss": 0.7799,
+      "step": 2120
+    },
+    {
+      "epoch": 1.5181753385602281,
+      "grad_norm": 6.09375,
+      "learning_rate": 7.100428642059033e-05,
+      "loss": 0.7962,
+      "step": 2130
+    },
+    {
+      "epoch": 1.5253029223093373,
+      "grad_norm": 5.84375,
+      "learning_rate": 7.068700531174319e-05,
+      "loss": 0.8263,
+      "step": 2140
+    },
+    {
+      "epoch": 1.5324305060584462,
+      "grad_norm": 4.78125,
+      "learning_rate": 7.036871594744217e-05,
+      "loss": 0.7541,
+      "step": 2150
+    },
+    {
+      "epoch": 1.5395580898075552,
+      "grad_norm": 5.1875,
+      "learning_rate": 7.004943384066187e-05,
+      "loss": 0.7617,
+      "step": 2160
+    },
+    {
+      "epoch": 1.5466856735566643,
+      "grad_norm": 6.28125,
+      "learning_rate": 6.97291745527617e-05,
+      "loss": 0.8576,
+      "step": 2170
+    },
+    {
+      "epoch": 1.5538132573057735,
+      "grad_norm": 5.6875,
+      "learning_rate": 6.940795369272754e-05,
+      "loss": 0.7801,
+      "step": 2180
+    },
+    {
+      "epoch": 1.5609408410548824,
+      "grad_norm": 5.96875,
+      "learning_rate": 6.908578691641092e-05,
+      "loss": 0.7256,
+      "step": 2190
+    },
+    {
+      "epoch": 1.5680684248039913,
+      "grad_norm": 4.03125,
+      "learning_rate": 6.876268992576604e-05,
+      "loss": 0.7574,
+      "step": 2200
+    },
+    {
+      "epoch": 1.5751960085531005,
+      "grad_norm": 6.65625,
+      "learning_rate": 6.843867846808438e-05,
+      "loss": 0.7539,
+      "step": 2210
+    },
+    {
+      "epoch": 1.5823235923022096,
+      "grad_norm": 6.25,
+      "learning_rate": 6.811376833522729e-05,
+      "loss": 0.7865,
+      "step": 2220
+    },
+    {
+      "epoch": 1.5894511760513186,
+      "grad_norm": 4.90625,
+      "learning_rate": 6.778797536285624e-05,
+      "loss": 0.781,
+      "step": 2230
+    },
+    {
+      "epoch": 1.5965787598004275,
+      "grad_norm": 6.5625,
+      "learning_rate": 6.746131542966112e-05,
+      "loss": 0.9054,
+      "step": 2240
+    },
+    {
+      "epoch": 1.6037063435495367,
+      "grad_norm": 4.0625,
+      "learning_rate": 6.713380445658618e-05,
+      "loss": 0.7958,
+      "step": 2250
+    },
+    {
+      "epoch": 1.6108339272986458,
+      "grad_norm": 5.8125,
+      "learning_rate": 6.680545840605423e-05,
+      "loss": 0.7319,
+      "step": 2260
+    },
+    {
+      "epoch": 1.6179615110477548,
+      "grad_norm": 5.5625,
+      "learning_rate": 6.647629328118851e-05,
+      "loss": 0.7538,
+      "step": 2270
+    },
+    {
+      "epoch": 1.6250890947968637,
+      "grad_norm": 4.625,
+      "learning_rate": 6.614632512503288e-05,
+      "loss": 0.7781,
+      "step": 2280
+    },
+    {
+      "epoch": 1.6322166785459729,
+      "grad_norm": 6.21875,
+      "learning_rate": 6.58155700197697e-05,
+      "loss": 0.7636,
+      "step": 2290
+    },
+    {
+      "epoch": 1.639344262295082,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.548404408593621e-05,
+      "loss": 0.7041,
+      "step": 2300
+    },
+    {
+      "epoch": 1.6464718460441912,
+      "grad_norm": 4.53125,
+      "learning_rate": 6.515176348163871e-05,
+      "loss": 0.8014,
+      "step": 2310
+    },
+    {
+      "epoch": 1.6535994297933,
+      "grad_norm": 4.8125,
+      "learning_rate": 6.481874440176506e-05,
+      "loss": 0.7043,
+      "step": 2320
+    },
+    {
+      "epoch": 1.660727013542409,
+      "grad_norm": 4.78125,
+      "learning_rate": 6.448500307719538e-05,
+      "loss": 0.7739,
+      "step": 2330
+    },
+    {
+      "epoch": 1.6678545972915182,
+      "grad_norm": 5.0625,
+      "learning_rate": 6.415055577401102e-05,
+      "loss": 0.7807,
+      "step": 2340
+    },
+    {
+      "epoch": 1.6749821810406273,
+      "grad_norm": 4.84375,
+      "learning_rate": 6.381541879270169e-05,
+      "loss": 0.8122,
+      "step": 2350
+    },
+    {
+      "epoch": 1.6821097647897363,
+      "grad_norm": 6.25,
+      "learning_rate": 6.347960846737106e-05,
+      "loss": 0.8824,
+      "step": 2360
+    },
+    {
+      "epoch": 1.6892373485388452,
+      "grad_norm": 4.46875,
+      "learning_rate": 6.31431411649406e-05,
+      "loss": 0.6925,
+      "step": 2370
+    },
+    {
+      "epoch": 1.6963649322879544,
+      "grad_norm": 5.59375,
+      "learning_rate": 6.280603328435198e-05,
+      "loss": 0.7558,
+      "step": 2380
+    },
+    {
+      "epoch": 1.7034925160370635,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.24683012557677e-05,
+      "loss": 0.7045,
+      "step": 2390
+    },
+    {
+      "epoch": 1.7106200997861725,
+      "grad_norm": 5.0625,
+      "learning_rate": 6.212996153977037e-05,
+      "loss": 0.7654,
+      "step": 2400
+    },
+    {
+      "epoch": 1.7177476835352814,
+      "grad_norm": 4.90625,
+      "learning_rate": 6.179103062656042e-05,
+      "loss": 0.7696,
+      "step": 2410
+    },
+    {
+      "epoch": 1.7248752672843906,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.145152503515239e-05,
+      "loss": 0.8205,
+      "step": 2420
+    },
+    {
+      "epoch": 1.7320028510334997,
+      "grad_norm": 6.1875,
+      "learning_rate": 6.111146131256983e-05,
+      "loss": 0.7752,
+      "step": 2430
+    },
+    {
+      "epoch": 1.7391304347826086,
+      "grad_norm": 4.4375,
+      "learning_rate": 6.077085603303882e-05,
+      "loss": 0.79,
+      "step": 2440
+    },
+    {
+      "epoch": 1.7462580185317176,
+      "grad_norm": 4.75,
+      "learning_rate": 6.042972579718019e-05,
+      "loss": 0.9032,
+      "step": 2450
+    },
+    {
+      "epoch": 1.7533856022808267,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.008808723120035e-05,
+      "loss": 0.7631,
+      "step": 2460
+    },
+    {
+      "epoch": 1.760513186029936,
+      "grad_norm": 4.9375,
+      "learning_rate": 5.974595698608103e-05,
+      "loss": 0.7612,
+      "step": 2470
+    },
+    {
+      "epoch": 1.767640769779045,
+      "grad_norm": 5.125,
+      "learning_rate": 5.9403351736767696e-05,
+      "loss": 0.7827,
+      "step": 2480
+    },
+    {
+      "epoch": 1.774768353528154,
+      "grad_norm": 4.8125,
+      "learning_rate": 5.906028818135687e-05,
+      "loss": 0.7827,
+      "step": 2490
+    },
+    {
+      "epoch": 1.781895937277263,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.8716783040282244e-05,
+      "loss": 0.7646,
+      "step": 2500
+    },
+    {
+      "epoch": 1.789023521026372,
+      "grad_norm": 7.3125,
+      "learning_rate": 5.837285305549978e-05,
+      "loss": 0.8215,
+      "step": 2510
+    },
+    {
+      "epoch": 1.7961511047754812,
+      "grad_norm": 4.84375,
+      "learning_rate": 5.8028514989671724e-05,
+      "loss": 0.7416,
+      "step": 2520
+    },
+    {
+      "epoch": 1.8032786885245902,
+      "grad_norm": 4.3125,
+      "learning_rate": 5.7683785625349616e-05,
+      "loss": 0.7577,
+      "step": 2530
+    },
+    {
+      "epoch": 1.810406272273699,
+      "grad_norm": 4.34375,
+      "learning_rate": 5.733868176415633e-05,
+      "loss": 0.6839,
+      "step": 2540
+    },
+    {
+      "epoch": 1.8175338560228083,
+      "grad_norm": 4.46875,
+      "learning_rate": 5.699322022596722e-05,
+      "loss": 0.719,
+      "step": 2550
+    },
+    {
+      "epoch": 1.8246614397719174,
+      "grad_norm": 5.875,
+      "learning_rate": 5.664741784809022e-05,
+      "loss": 0.7487,
+      "step": 2560
+    },
+    {
+      "epoch": 1.8317890235210263,
+      "grad_norm": 5.40625,
+      "learning_rate": 5.630129148444543e-05,
+      "loss": 0.7946,
+      "step": 2570
+    },
+    {
+      "epoch": 1.8389166072701353,
+      "grad_norm": 5.59375,
+      "learning_rate": 5.595485800474349e-05,
+      "loss": 0.7822,
+      "step": 2580
+    },
+    {
+      "epoch": 1.8460441910192444,
+      "grad_norm": 6.59375,
+      "learning_rate": 5.560813429366345e-05,
+      "loss": 0.7763,
+      "step": 2590
+    },
+    {
+      "epoch": 1.8531717747683536,
+      "grad_norm": 5.90625,
+      "learning_rate": 5.5261137250029835e-05,
+      "loss": 0.8152,
+      "step": 2600
+    },
+    {
+      "epoch": 1.8602993585174625,
+      "grad_norm": 5.53125,
+      "learning_rate": 5.4913883785988993e-05,
+      "loss": 0.8451,
+      "step": 2610
+    },
+    {
+      "epoch": 1.8674269422665715,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.456639082618489e-05,
+      "loss": 0.7819,
+      "step": 2620
+    },
+    {
+      "epoch": 1.8745545260156806,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.421867530693414e-05,
+      "loss": 0.794,
+      "step": 2630
+    },
+    {
+      "epoch": 1.8816821097647898,
+      "grad_norm": 4.8125,
+      "learning_rate": 5.3870754175400595e-05,
+      "loss": 0.7435,
+      "step": 2640
+    },
+    {
+      "epoch": 1.888809693513899,
+      "grad_norm": 6.09375,
+      "learning_rate": 5.352264438876935e-05,
+      "loss": 0.8852,
+      "step": 2650
+    },
+    {
+      "epoch": 1.8959372772630079,
+      "grad_norm": 5.46875,
+      "learning_rate": 5.3174362913420306e-05,
+      "loss": 0.7538,
+      "step": 2660
+    },
+    {
+      "epoch": 1.9030648610121168,
+      "grad_norm": 5.3125,
+      "learning_rate": 5.2825926724101236e-05,
+      "loss": 0.7795,
+      "step": 2670
+    },
+    {
+      "epoch": 1.910192444761226,
+      "grad_norm": 5.65625,
+      "learning_rate": 5.247735280310041e-05,
+      "loss": 0.7887,
+      "step": 2680
+    },
+    {
+      "epoch": 1.9173200285103351,
+      "grad_norm": 6.09375,
+      "learning_rate": 5.212865813941899e-05,
+      "loss": 0.6838,
+      "step": 2690
+    },
+    {
+      "epoch": 1.924447612259444,
+      "grad_norm": 6.4375,
+      "learning_rate": 5.1779859727942924e-05,
+      "loss": 0.7495,
+      "step": 2700
+    },
+    {
+      "epoch": 1.931575196008553,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.143097456861474e-05,
+      "loss": 0.7499,
+      "step": 2710
+    },
+    {
+      "epoch": 1.9387027797576621,
+      "grad_norm": 5.21875,
+      "learning_rate": 5.1082019665604895e-05,
+      "loss": 0.8243,
+      "step": 2720
+    },
+    {
+      "epoch": 1.9458303635067713,
+      "grad_norm": 5.34375,
+      "learning_rate": 5.073301202648304e-05,
+      "loss": 0.8128,
+      "step": 2730
+    },
+    {
+      "epoch": 1.9529579472558802,
+      "grad_norm": 5.71875,
+      "learning_rate": 5.0383968661389146e-05,
+      "loss": 0.7626,
+      "step": 2740
+    },
+    {
+      "epoch": 1.9600855310049892,
+      "grad_norm": 4.96875,
+      "learning_rate": 5.003490658220438e-05,
+      "loss": 0.7724,
+      "step": 2750
+    },
+    {
+      "epoch": 1.9672131147540983,
+      "grad_norm": 5.59375,
+      "learning_rate": 4.968584280172206e-05,
+      "loss": 0.7741,
+      "step": 2760
+    },
+    {
+      "epoch": 1.9743406985032075,
+      "grad_norm": 6.3125,
+      "learning_rate": 4.933679433281836e-05,
+      "loss": 0.7482,
+      "step": 2770
+    },
+    {
+      "epoch": 1.9814682822523164,
+      "grad_norm": 5.6875,
+      "learning_rate": 4.8987778187623245e-05,
+      "loss": 0.7981,
+      "step": 2780
+    },
+    {
+      "epoch": 1.9885958660014256,
+      "grad_norm": 4.875,
+      "learning_rate": 4.863881137669123e-05,
+      "loss": 0.7901,
+      "step": 2790
+    },
+    {
+      "epoch": 1.9957234497505345,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.8289910908172376e-05,
+      "loss": 0.7576,
+      "step": 2800
+    },
+    {
+      "epoch": 2.0028510334996437,
+      "grad_norm": 5.4375,
+      "learning_rate": 4.794109378698327e-05,
+      "loss": 0.7789,
+      "step": 2810
+    },
+    {
+      "epoch": 2.009978617248753,
+      "grad_norm": 4.96875,
+      "learning_rate": 4.7592377013978306e-05,
+      "loss": 0.8136,
+      "step": 2820
+    },
+    {
+      "epoch": 2.0171062009978615,
+      "grad_norm": 5.1875,
+      "learning_rate": 4.724377758512103e-05,
+      "loss": 0.813,
+      "step": 2830
+    },
+    {
+      "epoch": 2.0242337847469707,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.68953124906558e-05,
+      "loss": 0.7948,
+      "step": 2840
+    },
+    {
+      "epoch": 2.03136136849608,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.654699871427971e-05,
+      "loss": 0.6773,
+      "step": 2850
+    },
+    {
+      "epoch": 2.038488952245189,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.619885323231484e-05,
+      "loss": 0.6645,
+      "step": 2860
+    },
+    {
+      "epoch": 2.0456165359942977,
+      "grad_norm": 5.34375,
+      "learning_rate": 4.58508930128808e-05,
+      "loss": 0.7572,
+      "step": 2870
+    },
+    {
+      "epoch": 2.052744119743407,
+      "grad_norm": 4.59375,
+      "learning_rate": 4.550313501506781e-05,
+      "loss": 0.7588,
+      "step": 2880
+    },
+    {
+      "epoch": 2.059871703492516,
+      "grad_norm": 5.625,
+      "learning_rate": 4.515559618811005e-05,
+      "loss": 0.7294,
+      "step": 2890
+    },
+    {
+      "epoch": 2.066999287241625,
+      "grad_norm": 5.09375,
+      "learning_rate": 4.4808293470559643e-05,
+      "loss": 0.7621,
+      "step": 2900
+    },
+    {
+      "epoch": 2.0741268709907343,
+      "grad_norm": 6.65625,
+      "learning_rate": 4.446124378946107e-05,
+      "loss": 0.8967,
+      "step": 2910
+    },
+    {
+      "epoch": 2.081254454739843,
+      "grad_norm": 5.3125,
+      "learning_rate": 4.411446405952618e-05,
+      "loss": 0.84,
+      "step": 2920
+    },
+    {
+      "epoch": 2.088382038488952,
+      "grad_norm": 4.375,
+      "learning_rate": 4.376797118230979e-05,
+      "loss": 0.7584,
+      "step": 2930
+    },
+    {
+      "epoch": 2.0955096222380614,
+      "grad_norm": 5.21875,
+      "learning_rate": 4.342178204538588e-05,
+      "loss": 0.8475,
+      "step": 2940
+    },
+    {
+      "epoch": 2.1026372059871705,
+      "grad_norm": 6.40625,
+      "learning_rate": 4.307591352152459e-05,
+      "loss": 0.7067,
+      "step": 2950
+    },
+    {
+      "epoch": 2.1097647897362792,
+      "grad_norm": 6.1875,
+      "learning_rate": 4.273038246786986e-05,
+      "loss": 0.8023,
+      "step": 2960
+    },
+    {
+      "epoch": 2.1168923734853884,
+      "grad_norm": 5.03125,
+      "learning_rate": 4.238520572511773e-05,
+      "loss": 0.7074,
+      "step": 2970
+    },
+    {
+      "epoch": 2.1240199572344975,
+      "grad_norm": 7.5,
+      "learning_rate": 4.2040400116695666e-05,
+      "loss": 0.7498,
+      "step": 2980
+    },
+    {
+      "epoch": 2.1311475409836067,
+      "grad_norm": 7.0625,
+      "learning_rate": 4.169598244794261e-05,
+      "loss": 0.7249,
+      "step": 2990
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "grad_norm": 5.53125,
+      "learning_rate": 4.135196950528982e-05,
+      "loss": 0.7907,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.arguana": 0.4886028250717679,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.climate_fever": 0.2710241560666136,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.dbpedia_entity": 0.3285043023237046,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.fever": 0.6331373355635795,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.fiqa": 0.4150357832110943,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.hotpotqa": 0.68158322360907,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.nfcorpus": 0.3531375464357257,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.nq": 0.41568885251388027,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.quora": 0.8349377083144377,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.scidocs": 0.24315533084930935,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.scifact": 0.7290861351959168,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/nano_beir.webis_touche2020": 0.28783156872457877,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1382751247327154,
+      "eval/avg": 0.47347706398997325,
+      "step": 3000
+    },
+    {
+      "epoch": 2.1454027084818246,
+      "grad_norm": 5.75,
+      "learning_rate": 4.100837805544279e-05,
+      "loss": 0.7733,
+      "step": 3010
+    },
+    {
+      "epoch": 2.1525302922309337,
+      "grad_norm": 4.28125,
+      "learning_rate": 4.0665224844564053e-05,
+      "loss": 0.7141,
+      "step": 3020
+    },
+    {
+      "epoch": 2.159657875980043,
+      "grad_norm": 6.8125,
+      "learning_rate": 4.032252659745699e-05,
+      "loss": 0.6352,
+      "step": 3030
+    },
+    {
+      "epoch": 2.1667854597291516,
+      "grad_norm": 4.75,
+      "learning_rate": 3.9980300016750694e-05,
+      "loss": 0.8184,
+      "step": 3040
+    },
+    {
+      "epoch": 2.1739130434782608,
+      "grad_norm": 5.0625,
+      "learning_rate": 3.963856178208588e-05,
+      "loss": 0.7785,
+      "step": 3050
+    },
+    {
+      "epoch": 2.18104062722737,
+      "grad_norm": 4.875,
+      "learning_rate": 3.9297328549302e-05,
+      "loss": 0.7565,
+      "step": 3060
+    },
+    {
+      "epoch": 2.188168210976479,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.895661694962542e-05,
+      "loss": 0.7608,
+      "step": 3070
+    },
+    {
+      "epoch": 2.1952957947255882,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.8616443588858805e-05,
+      "loss": 0.7354,
+      "step": 3080
+    },
+    {
+      "epoch": 2.202423378474697,
+      "grad_norm": 5.28125,
+      "learning_rate": 3.827682504657187e-05,
+      "loss": 0.7828,
+      "step": 3090
+    },
+    {
+      "epoch": 2.209550962223806,
+      "grad_norm": 7.1875,
+      "learning_rate": 3.7937777875293244e-05,
+      "loss": 0.7934,
+      "step": 3100
+    },
+    {
+      "epoch": 2.2166785459729152,
+      "grad_norm": 3.859375,
+      "learning_rate": 3.759931859970374e-05,
+      "loss": 0.6331,
+      "step": 3110
+    },
+    {
+      "epoch": 2.2238061297220244,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.72614637158309e-05,
+      "loss": 0.7351,
+      "step": 3120
+    },
+    {
+      "epoch": 2.230933713471133,
+      "grad_norm": 6.15625,
+      "learning_rate": 3.692422969024516e-05,
+      "loss": 0.7856,
+      "step": 3130
+    },
+    {
+      "epoch": 2.2380612972202423,
+      "grad_norm": 5.5,
+      "learning_rate": 3.6587632959257165e-05,
+      "loss": 0.7819,
+      "step": 3140
+    },
+    {
+      "epoch": 2.2451888809693514,
+      "grad_norm": 5.09375,
+      "learning_rate": 3.625168992811671e-05,
+      "loss": 0.7397,
+      "step": 3150
+    },
+    {
+      "epoch": 2.2523164647184606,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.591641697021317e-05,
+      "loss": 0.7588,
+      "step": 3160
+    },
+    {
+      "epoch": 2.2594440484675697,
+      "grad_norm": 5.5625,
+      "learning_rate": 3.558183042627755e-05,
+      "loss": 0.751,
+      "step": 3170
+    },
+    {
+      "epoch": 2.2665716322166785,
+      "grad_norm": 5.46875,
+      "learning_rate": 3.524794660358593e-05,
+      "loss": 0.8684,
+      "step": 3180
+    },
+    {
+      "epoch": 2.2736992159657876,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.491478177516484e-05,
+      "loss": 0.7863,
+      "step": 3190
+    },
+    {
+      "epoch": 2.2808267997148968,
+      "grad_norm": 5.25,
+      "learning_rate": 3.4582352178997935e-05,
+      "loss": 0.7262,
+      "step": 3200
+    },
+    {
+      "epoch": 2.2879543834640055,
+      "grad_norm": 4.0,
+      "learning_rate": 3.425067401723477e-05,
+      "loss": 0.7284,
+      "step": 3210
+    },
+    {
+      "epoch": 2.2950819672131146,
+      "grad_norm": 5.3125,
+      "learning_rate": 3.3919763455401015e-05,
+      "loss": 0.7184,
+      "step": 3220
+    },
+    {
+      "epoch": 2.302209550962224,
+      "grad_norm": 5.21875,
+      "learning_rate": 3.358963662161062e-05,
+      "loss": 0.8039,
+      "step": 3230
+    },
+    {
+      "epoch": 2.309337134711333,
+      "grad_norm": 4.28125,
+      "learning_rate": 3.326030960577972e-05,
+      "loss": 0.7646,
+      "step": 3240
+    },
+    {
+      "epoch": 2.316464718460442,
+      "grad_norm": 5.96875,
+      "learning_rate": 3.293179845884245e-05,
+      "loss": 0.7086,
+      "step": 3250
+    },
+    {
+      "epoch": 2.323592302209551,
+      "grad_norm": 5.6875,
+      "learning_rate": 3.2604119191968654e-05,
+      "loss": 0.8153,
+      "step": 3260
+    },
+    {
+      "epoch": 2.33071988595866,
+      "grad_norm": 6.125,
+      "learning_rate": 3.2277287775783525e-05,
+      "loss": 0.7863,
+      "step": 3270
+    },
+    {
+      "epoch": 2.337847469707769,
+      "grad_norm": 5.90625,
+      "learning_rate": 3.195132013958918e-05,
+      "loss": 0.7569,
+      "step": 3280
+    },
+    {
+      "epoch": 2.3449750534568783,
+      "grad_norm": 4.34375,
+      "learning_rate": 3.162623217058834e-05,
+      "loss": 0.7193,
+      "step": 3290
+    },
+    {
+      "epoch": 2.352102637205987,
+      "grad_norm": 7.15625,
+      "learning_rate": 3.130203971310999e-05,
+      "loss": 0.847,
+      "step": 3300
+    },
+    {
+      "epoch": 2.359230220955096,
+      "grad_norm": 4.6875,
+      "learning_rate": 3.097875856783713e-05,
+      "loss": 0.7884,
+      "step": 3310
+    },
+    {
+      "epoch": 2.3663578047042053,
+      "grad_norm": 4.46875,
+      "learning_rate": 3.065640449103669e-05,
+      "loss": 0.7849,
+      "step": 3320
+    },
+    {
+      "epoch": 2.3734853884533145,
+      "grad_norm": 4.84375,
+      "learning_rate": 3.033499319379163e-05,
+      "loss": 0.7069,
+      "step": 3330
+    },
+    {
+      "epoch": 2.3806129722024236,
+      "grad_norm": 4.28125,
+      "learning_rate": 3.001454034123512e-05,
+      "loss": 0.6801,
+      "step": 3340
+    },
+    {
+      "epoch": 2.3877405559515323,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.9695061551787108e-05,
+      "loss": 0.7755,
+      "step": 3350
+    },
+    {
+      "epoch": 2.3948681397006415,
+      "grad_norm": 4.21875,
+      "learning_rate": 2.9376572396393048e-05,
+      "loss": 0.8031,
+      "step": 3360
+    },
+    {
+      "epoch": 2.4019957234497507,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.9059088397765088e-05,
+      "loss": 0.8501,
+      "step": 3370
+    },
+    {
+      "epoch": 2.4091233071988594,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.874262502962537e-05,
+      "loss": 0.8451,
+      "step": 3380
+    },
+    {
+      "epoch": 2.4162508909479685,
+      "grad_norm": 4.3125,
+      "learning_rate": 2.8427197715952047e-05,
+      "loss": 0.7452,
+      "step": 3390
+    },
+    {
+      "epoch": 2.4233784746970777,
+      "grad_norm": 4.8125,
+      "learning_rate": 2.811282183022736e-05,
+      "loss": 0.6863,
+      "step": 3400
+    },
+    {
+      "epoch": 2.430506058446187,
+      "grad_norm": 5.5,
+      "learning_rate": 2.7799512694688468e-05,
+      "loss": 0.811,
+      "step": 3410
+    },
+    {
+      "epoch": 2.437633642195296,
+      "grad_norm": 4.375,
+      "learning_rate": 2.7487285579580637e-05,
+      "loss": 0.7234,
+      "step": 3420
+    },
+    {
+      "epoch": 2.4447612259444047,
+      "grad_norm": 5.0625,
+      "learning_rate": 2.7176155702412937e-05,
+      "loss": 0.8367,
+      "step": 3430
+    },
+    {
+      "epoch": 2.451888809693514,
+      "grad_norm": 5.90625,
+      "learning_rate": 2.686613822721666e-05,
+      "loss": 0.6122,
+      "step": 3440
+    },
+    {
+      "epoch": 2.459016393442623,
+      "grad_norm": 4.4375,
+      "learning_rate": 2.6557248263806174e-05,
+      "loss": 0.7393,
+      "step": 3450
+    },
+    {
+      "epoch": 2.466143977191732,
+      "grad_norm": 5.0,
+      "learning_rate": 2.6249500867042522e-05,
+      "loss": 0.7783,
+      "step": 3460
+    },
+    {
+      "epoch": 2.473271560940841,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.5942911036099655e-05,
+      "loss": 0.6654,
+      "step": 3470
+    },
+    {
+      "epoch": 2.48039914468995,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.5637493713733374e-05,
+      "loss": 0.752,
+      "step": 3480
+    },
+    {
+      "epoch": 2.487526728439059,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.533326378555314e-05,
+      "loss": 0.7761,
+      "step": 3490
+    },
+    {
+      "epoch": 2.4946543121881684,
+      "grad_norm": 5.53125,
+      "learning_rate": 2.5030236079296444e-05,
+      "loss": 0.7594,
+      "step": 3500
+    },
+    {
+      "epoch": 2.5017818959372775,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.4728425364106135e-05,
+      "loss": 0.7519,
+      "step": 3510
+    },
+    {
+      "epoch": 2.5089094796863862,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.442784634981071e-05,
+      "loss": 0.7527,
+      "step": 3520
+    },
+    {
+      "epoch": 2.5160370634354954,
+      "grad_norm": 4.28125,
+      "learning_rate": 2.412851368620726e-05,
+      "loss": 0.6595,
+      "step": 3530
+    },
+    {
+      "epoch": 2.5231646471846045,
+      "grad_norm": 4.53125,
+      "learning_rate": 2.383044196234753e-05,
+      "loss": 0.7621,
+      "step": 3540
+    },
+    {
+      "epoch": 2.5302922309337132,
+      "grad_norm": 4.28125,
+      "learning_rate": 2.3533645705826806e-05,
+      "loss": 0.7766,
+      "step": 3550
+    },
+    {
+      "epoch": 2.5374198146828224,
+      "grad_norm": 4.625,
+      "learning_rate": 2.323813938207593e-05,
+      "loss": 0.7717,
+      "step": 3560
+    },
+    {
+      "epoch": 2.5445473984319316,
+      "grad_norm": 5.625,
+      "learning_rate": 2.294393739365621e-05,
+      "loss": 0.6604,
+      "step": 3570
+    },
+    {
+      "epoch": 2.5516749821810407,
+      "grad_norm": 4.03125,
+      "learning_rate": 2.265105407955752e-05,
+      "loss": 0.7614,
+      "step": 3580
+    },
+    {
+      "epoch": 2.55880256593015,
+      "grad_norm": 5.59375,
+      "learning_rate": 2.235950371449938e-05,
+      "loss": 0.7702,
+      "step": 3590
+    },
+    {
+      "epoch": 2.5659301496792586,
+      "grad_norm": 5.96875,
+      "learning_rate": 2.2069300508235275e-05,
+      "loss": 0.8866,
+      "step": 3600
+    },
+    {
+      "epoch": 2.5730577334283677,
+      "grad_norm": 6.21875,
+      "learning_rate": 2.1780458604860056e-05,
+      "loss": 0.8359,
+      "step": 3610
+    },
+    {
+      "epoch": 2.580185317177477,
+      "grad_norm": 4.75,
+      "learning_rate": 2.14929920821206e-05,
+      "loss": 0.7036,
+      "step": 3620
+    },
+    {
+      "epoch": 2.5873129009265856,
+      "grad_norm": 6.25,
+      "learning_rate": 2.1206914950729672e-05,
+      "loss": 0.7946,
+      "step": 3630
+    },
+    {
+      "epoch": 2.5944404846756948,
+      "grad_norm": 8.1875,
+      "learning_rate": 2.0922241153683066e-05,
+      "loss": 0.7035,
+      "step": 3640
+    },
+    {
+      "epoch": 2.601568068424804,
+      "grad_norm": 5.0625,
+      "learning_rate": 2.0638984565580023e-05,
+      "loss": 0.7456,
+      "step": 3650
+    },
+    {
+      "epoch": 2.608695652173913,
+      "grad_norm": 5.6875,
+      "learning_rate": 2.035715899194704e-05,
+      "loss": 0.7462,
+      "step": 3660
+    },
+    {
+      "epoch": 2.6158232359230222,
+      "grad_norm": 4.65625,
+      "learning_rate": 2.007677816856498e-05,
+      "loss": 0.7088,
+      "step": 3670
+    },
+    {
+      "epoch": 2.6229508196721314,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.9797855760799612e-05,
+      "loss": 0.721,
+      "step": 3680
+    },
+    {
+      "epoch": 2.63007840342124,
+      "grad_norm": 5.84375,
+      "learning_rate": 1.9520405362935594e-05,
+      "loss": 0.7292,
+      "step": 3690
+    },
+    {
+      "epoch": 2.6372059871703493,
+      "grad_norm": 5.0,
+      "learning_rate": 1.9244440497513893e-05,
+      "loss": 0.7031,
+      "step": 3700
+    },
+    {
+      "epoch": 2.6443335709194584,
+      "grad_norm": 4.3125,
+      "learning_rate": 1.896997461467272e-05,
+      "loss": 0.696,
+      "step": 3710
+    },
+    {
+      "epoch": 2.651461154668567,
+      "grad_norm": 4.875,
+      "learning_rate": 1.869702109149199e-05,
+      "loss": 0.749,
+      "step": 3720
+    },
+    {
+      "epoch": 2.6585887384176763,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.842559323134136e-05,
+      "loss": 0.7282,
+      "step": 3730
+    },
+    {
+      "epoch": 2.6657163221667854,
+      "grad_norm": 5.0625,
+      "learning_rate": 1.8155704263231776e-05,
+      "loss": 0.6514,
+      "step": 3740
+    },
+    {
+      "epoch": 2.6728439059158946,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.788736734117078e-05,
+      "loss": 0.7496,
+      "step": 3750
+    },
+    {
+      "epoch": 2.6799714896650038,
+      "grad_norm": 4.5,
+      "learning_rate": 1.762059554352143e-05,
+      "loss": 0.8261,
+      "step": 3760
+    },
+    {
+      "epoch": 2.6870990734141125,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.7355401872364758e-05,
+      "loss": 0.8161,
+      "step": 3770
+    },
+    {
+      "epoch": 2.6942266571632216,
+      "grad_norm": 5.125,
+      "learning_rate": 1.709179925286617e-05,
+      "loss": 0.682,
+      "step": 3780
+    },
+    {
+      "epoch": 2.701354240912331,
+      "grad_norm": 4.90625,
+      "learning_rate": 1.6829800532645447e-05,
+      "loss": 0.7542,
+      "step": 3790
+    },
+    {
+      "epoch": 2.7084818246614395,
+      "grad_norm": 6.46875,
+      "learning_rate": 1.6569418481150595e-05,
+      "loss": 0.8126,
+      "step": 3800
+    },
+    {
+      "epoch": 2.7156094084105487,
+      "grad_norm": 4.6875,
+      "learning_rate": 1.6310665789035467e-05,
+      "loss": 0.7445,
+      "step": 3810
+    },
+    {
+      "epoch": 2.722736992159658,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.605355506754121e-05,
+      "loss": 0.8296,
+      "step": 3820
+    },
+    {
+      "epoch": 2.729864575908767,
+      "grad_norm": 4.84375,
+      "learning_rate": 1.5798098847881666e-05,
+      "loss": 0.6684,
+      "step": 3830
+    },
+    {
+      "epoch": 2.736992159657876,
+      "grad_norm": 4.5625,
+      "learning_rate": 1.554430958063259e-05,
+      "loss": 0.7063,
+      "step": 3840
+    },
+    {
+      "epoch": 2.7441197434069853,
+      "grad_norm": 4.625,
+      "learning_rate": 1.529219963512481e-05,
+      "loss": 0.6668,
+      "step": 3850
+    },
+    {
+      "epoch": 2.751247327156094,
+      "grad_norm": 5.21875,
+      "learning_rate": 1.5041781298841423e-05,
+      "loss": 0.7464,
+      "step": 3860
+    },
+    {
+      "epoch": 2.758374910905203,
+      "grad_norm": 4.78125,
+      "learning_rate": 1.4793066776818842e-05,
+      "loss": 0.7625,
+      "step": 3870
+    },
+    {
+      "epoch": 2.7655024946543123,
+      "grad_norm": 5.625,
+      "learning_rate": 1.4546068191051987e-05,
+      "loss": 0.738,
+      "step": 3880
+    },
+    {
+      "epoch": 2.772630078403421,
+      "grad_norm": 6.125,
+      "learning_rate": 1.4300797579903474e-05,
+      "loss": 0.8185,
+      "step": 3890
+    },
+    {
+      "epoch": 2.77975766215253,
+      "grad_norm": 5.4375,
+      "learning_rate": 1.4057266897516841e-05,
+      "loss": 0.7021,
+      "step": 3900
+    },
+    {
+      "epoch": 2.7868852459016393,
+      "grad_norm": 6.78125,
+      "learning_rate": 1.3815488013233984e-05,
+      "loss": 0.8398,
+      "step": 3910
+    },
+    {
+      "epoch": 2.7940128296507485,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.3575472711016635e-05,
+      "loss": 0.8821,
+      "step": 3920
+    },
+    {
+      "epoch": 2.8011404133998576,
+      "grad_norm": 4.75,
+      "learning_rate": 1.3337232688872009e-05,
+      "loss": 0.7238,
+      "step": 3930
+    },
+    {
+      "epoch": 2.8082679971489664,
+      "grad_norm": 5.1875,
+      "learning_rate": 1.3100779558282672e-05,
+      "loss": 0.6887,
+      "step": 3940
+    },
+    {
+      "epoch": 2.8153955808980755,
+      "grad_norm": 6.03125,
+      "learning_rate": 1.2866124843640615e-05,
+      "loss": 0.7804,
+      "step": 3950
+    },
+    {
+      "epoch": 2.8225231646471847,
+      "grad_norm": 5.78125,
+      "learning_rate": 1.2633279981685608e-05,
+      "loss": 0.871,
+      "step": 3960
+    },
+    {
+      "epoch": 2.8296507483962934,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.240225632094773e-05,
+      "loss": 0.764,
+      "step": 3970
+    },
+    {
+      "epoch": 2.8367783321454025,
+      "grad_norm": 4.65625,
+      "learning_rate": 1.217306512119425e-05,
+      "loss": 0.7754,
+      "step": 3980
+    },
+    {
+      "epoch": 2.8439059158945117,
+      "grad_norm": 5.71875,
+      "learning_rate": 1.194571755288092e-05,
+      "loss": 0.7185,
+      "step": 3990
+    },
+    {
+      "epoch": 2.851033499643621,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.1720224696607474e-05,
+      "loss": 0.6823,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.arguana": 0.47570368623108616,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.climate_fever": 0.27306481927622206,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.dbpedia_entity": 0.3271142777426427,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.fever": 0.6267850303425887,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.fiqa": 0.42142137153910825,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.hotpotqa": 0.6860936998599775,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.nfcorpus": 0.3525089997861853,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.nq": 0.40835064000757015,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.quora": 0.8347230240552653,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.scidocs": 0.2466296038296945,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.scifact": 0.7298579737170866,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/nano_beir.webis_touche2020": 0.2863659661103585,
+      "step": 4000
+    },
+    {
+      "epoch": 2.851033499643621,
+      "eval/avg": 0.4723849243748155,
+      "step": 4000
+    },
+    {
+      "epoch": 2.85816108339273,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.1496597542577603e-05,
+      "loss": 0.8187,
+      "step": 4010
+    },
+    {
+      "epoch": 2.865288667141839,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.1274846990063315e-05,
+      "loss": 0.7384,
+      "step": 4020
+    },
+    {
+      "epoch": 2.872416250890948,
+      "grad_norm": 5.375,
+      "learning_rate": 1.1054983846873685e-05,
+      "loss": 0.7379,
+      "step": 4030
+    },
+    {
+      "epoch": 2.879543834640057,
+      "grad_norm": 6.5,
+      "learning_rate": 1.0837018828828133e-05,
+      "loss": 0.8139,
+      "step": 4040
+    },
+    {
+      "epoch": 2.886671418389166,
+      "grad_norm": 5.96875,
+      "learning_rate": 1.0620962559234143e-05,
+      "loss": 0.8204,
+      "step": 4050
+    },
+    {
+      "epoch": 2.893799002138275,
+      "grad_norm": 5.8125,
+      "learning_rate": 1.0406825568369477e-05,
+      "loss": 0.8589,
+      "step": 4060
+    },
+    {
+      "epoch": 2.900926585887384,
+      "grad_norm": 4.96875,
+      "learning_rate": 1.0194618292968972e-05,
+      "loss": 0.7727,
+      "step": 4070
+    },
+    {
+      "epoch": 2.908054169636493,
+      "grad_norm": 5.0625,
+      "learning_rate": 9.984351075715848e-06,
+      "loss": 0.7422,
+      "step": 4080
+    },
+    {
+      "epoch": 2.9151817533856024,
+      "grad_norm": 4.8125,
+      "learning_rate": 9.77603416473763e-06,
+      "loss": 0.8394,
+      "step": 4090
+    },
+    {
+      "epoch": 2.9223093371347115,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.569677713106674e-06,
+      "loss": 0.6983,
+      "step": 4100
+    },
+    {
+      "epoch": 2.9294369208838202,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.365291778345303e-06,
+      "loss": 0.7259,
+      "step": 4110
+    },
+    {
+      "epoch": 2.9365645046329294,
+      "grad_norm": 5.15625,
+      "learning_rate": 9.162886321935632e-06,
+      "loss": 0.7249,
+      "step": 4120
+    },
+    {
+      "epoch": 2.9436920883820386,
+      "grad_norm": 5.03125,
+      "learning_rate": 8.962471208834055e-06,
+      "loss": 0.7308,
+      "step": 4130
+    },
+    {
+      "epoch": 2.9508196721311473,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.764056206990445e-06,
+      "loss": 0.6875,
+      "step": 4140
+    },
+    {
+      "epoch": 2.9579472558802564,
+      "grad_norm": 5.40625,
+      "learning_rate": 8.567650986872062e-06,
+      "loss": 0.7786,
+      "step": 4150
+    },
+    {
+      "epoch": 2.9650748396293656,
+      "grad_norm": 4.46875,
+      "learning_rate": 8.373265120992252e-06,
+      "loss": 0.7357,
+      "step": 4160
+    },
+    {
+      "epoch": 2.9722024233784747,
+      "grad_norm": 5.46875,
+      "learning_rate": 8.180908083443884e-06,
+      "loss": 0.7817,
+      "step": 4170
+    },
+    {
+      "epoch": 2.979330007127584,
+      "grad_norm": 4.75,
+      "learning_rate": 7.99058924943759e-06,
+      "loss": 0.7892,
+      "step": 4180
+    },
+    {
+      "epoch": 2.986457590876693,
+      "grad_norm": 6.25,
+      "learning_rate": 7.802317894844835e-06,
+      "loss": 0.7669,
+      "step": 4190
+    },
+    {
+      "epoch": 2.9935851746258018,
+      "grad_norm": 5.5,
+      "learning_rate": 7.6161031957458494e-06,
+      "loss": 0.8586,
+      "step": 4200
+    },
+    {
+      "epoch": 3.000712758374911,
+      "grad_norm": 4.40625,
+      "learning_rate": 7.43195422798233e-06,
+      "loss": 0.7776,
+      "step": 4210
+    },
+    {
+      "epoch": 3.00784034212402,
+      "grad_norm": 5.34375,
+      "learning_rate": 7.249879966715173e-06,
+      "loss": 0.7948,
+      "step": 4220
+    },
+    {
+      "epoch": 3.0149679258731292,
+      "grad_norm": 5.6875,
+      "learning_rate": 7.0698892859870245e-06,
+      "loss": 0.7505,
+      "step": 4230
+    },
+    {
+      "epoch": 3.022095509622238,
+      "grad_norm": 5.5,
+      "learning_rate": 6.891990958289723e-06,
+      "loss": 0.8047,
+      "step": 4240
+    },
+    {
+      "epoch": 3.029223093371347,
+      "grad_norm": 5.1875,
+      "learning_rate": 6.716193654136788e-06,
+      "loss": 0.7658,
+      "step": 4250
+    },
+    {
+      "epoch": 3.0363506771204563,
+      "grad_norm": 4.8125,
+      "learning_rate": 6.542505941640803e-06,
+      "loss": 0.675,
+      "step": 4260
+    },
+    {
+      "epoch": 3.0434782608695654,
+      "grad_norm": 5.09375,
+      "learning_rate": 6.370936286095841e-06,
+      "loss": 0.8296,
+      "step": 4270
+    },
+    {
+      "epoch": 3.050605844618674,
+      "grad_norm": 4.875,
+      "learning_rate": 6.201493049564882e-06,
+      "loss": 0.7138,
+      "step": 4280
+    },
+    {
+      "epoch": 3.0577334283677833,
+      "grad_norm": 3.984375,
+      "learning_rate": 6.034184490472195e-06,
+      "loss": 0.6907,
+      "step": 4290
+    },
+    {
+      "epoch": 3.0648610121168924,
+      "grad_norm": 5.625,
+      "learning_rate": 5.8690187632009285e-06,
+      "loss": 0.779,
+      "step": 4300
+    },
+    {
+      "epoch": 3.0719885958660016,
+      "grad_norm": 4.5,
+      "learning_rate": 5.706003917695618e-06,
+      "loss": 0.7086,
+      "step": 4310
+    },
+    {
+      "epoch": 3.0791161796151103,
+      "grad_norm": 5.96875,
+      "learning_rate": 5.545147899069836e-06,
+      "loss": 0.7896,
+      "step": 4320
+    },
+    {
+      "epoch": 3.0862437633642195,
+      "grad_norm": 6.5625,
+      "learning_rate": 5.3864585472190255e-06,
+      "loss": 0.7141,
+      "step": 4330
+    },
+    {
+      "epoch": 3.0933713471133286,
+      "grad_norm": 5.78125,
+      "learning_rate": 5.229943596438297e-06,
+      "loss": 0.7517,
+      "step": 4340
+    },
+    {
+      "epoch": 3.1004989308624378,
+      "grad_norm": 5.78125,
+      "learning_rate": 5.075610675045567e-06,
+      "loss": 0.7389,
+      "step": 4350
+    },
+    {
+      "epoch": 3.1076265146115465,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.92346730500966e-06,
+      "loss": 0.6878,
+      "step": 4360
+    },
+    {
+      "epoch": 3.1147540983606556,
+      "grad_norm": 5.96875,
+      "learning_rate": 4.7735209015838e-06,
+      "loss": 0.7404,
+      "step": 4370
+    },
+    {
+      "epoch": 3.121881682109765,
+      "grad_norm": 4.9375,
+      "learning_rate": 4.625778772944156e-06,
+      "loss": 0.702,
+      "step": 4380
+    },
+    {
+      "epoch": 3.129009265858874,
+      "grad_norm": 5.0625,
+      "learning_rate": 4.480248119833641e-06,
+      "loss": 0.7602,
+      "step": 4390
+    },
+    {
+      "epoch": 3.1361368496079827,
+      "grad_norm": 6.5,
+      "learning_rate": 4.33693603521097e-06,
+      "loss": 0.7341,
+      "step": 4400
+    },
+    {
+      "epoch": 3.143264433357092,
+      "grad_norm": 5.75,
+      "learning_rate": 4.195849503904975e-06,
+      "loss": 0.7399,
+      "step": 4410
+    },
+    {
+      "epoch": 3.150392017106201,
+      "grad_norm": 4.1875,
+      "learning_rate": 4.056995402274122e-06,
+      "loss": 0.7524,
+      "step": 4420
+    },
+    {
+      "epoch": 3.15751960085531,
+      "grad_norm": 5.0625,
+      "learning_rate": 3.920380497871473e-06,
+      "loss": 0.78,
+      "step": 4430
+    },
+    {
+      "epoch": 3.1646471846044193,
+      "grad_norm": 5.71875,
+      "learning_rate": 3.7860114491147016e-06,
+      "loss": 0.7816,
+      "step": 4440
+    },
+    {
+      "epoch": 3.171774768353528,
+      "grad_norm": 3.953125,
+      "learning_rate": 3.6538948049616882e-06,
+      "loss": 0.7733,
+      "step": 4450
+    },
+    {
+      "epoch": 3.178902352102637,
+      "grad_norm": 5.4375,
+      "learning_rate": 3.524037004591274e-06,
+      "loss": 0.8646,
+      "step": 4460
+    },
+    {
+      "epoch": 3.1860299358517463,
+      "grad_norm": 4.5,
+      "learning_rate": 3.3964443770894528e-06,
+      "loss": 0.806,
+      "step": 4470
+    },
+    {
+      "epoch": 3.1931575196008555,
+      "grad_norm": 5.25,
+      "learning_rate": 3.271123141140886e-06,
+      "loss": 0.7918,
+      "step": 4480
+    },
+    {
+      "epoch": 3.200285103349964,
+      "grad_norm": 6.0,
+      "learning_rate": 3.148079404725801e-06,
+      "loss": 0.8287,
+      "step": 4490
+    },
+    {
+      "epoch": 3.2074126870990733,
+      "grad_norm": 4.0,
+      "learning_rate": 3.0273191648223287e-06,
+      "loss": 0.6243,
+      "step": 4500
+    },
+    {
+      "epoch": 3.2145402708481825,
+      "grad_norm": 4.9375,
+      "learning_rate": 2.9088483071141983e-06,
+      "loss": 0.7122,
+      "step": 4510
+    },
+    {
+      "epoch": 3.2216678545972917,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.7926726057038666e-06,
+      "loss": 0.7312,
+      "step": 4520
+    },
+    {
+      "epoch": 3.2287954383464004,
+      "grad_norm": 4.5625,
+      "learning_rate": 2.6787977228311334e-06,
+      "loss": 0.7337,
+      "step": 4530
+    },
+    {
+      "epoch": 3.2359230220955095,
+      "grad_norm": 5.28125,
+      "learning_rate": 2.567229208597127e-06,
+      "loss": 0.7347,
+      "step": 4540
+    },
+    {
+      "epoch": 3.2430506058446187,
+      "grad_norm": 4.28125,
+      "learning_rate": 2.4579725006938337e-06,
+      "loss": 0.7344,
+      "step": 4550
+    },
+    {
+      "epoch": 3.250178189593728,
+      "grad_norm": 5.46875,
+      "learning_rate": 2.351032924139063e-06,
+      "loss": 0.8731,
+      "step": 4560
+    },
+    {
+      "epoch": 3.2573057733428366,
+      "grad_norm": 5.15625,
+      "learning_rate": 2.2464156910168954e-06,
+      "loss": 0.7059,
+      "step": 4570
+    },
+    {
+      "epoch": 3.2644333570919457,
+      "grad_norm": 4.875,
+      "learning_rate": 2.1441259002236924e-06,
+      "loss": 0.7404,
+      "step": 4580
+    },
+    {
+      "epoch": 3.271560940841055,
+      "grad_norm": 3.9375,
+      "learning_rate": 2.0441685372195484e-06,
+      "loss": 0.709,
+      "step": 4590
+    },
+    {
+      "epoch": 3.278688524590164,
+      "grad_norm": 4.625,
+      "learning_rate": 1.946548473785309e-06,
+      "loss": 0.732,
+      "step": 4600
+    },
+    {
+      "epoch": 3.285816108339273,
+      "grad_norm": 6.3125,
+      "learning_rate": 1.8512704677851488e-06,
+      "loss": 0.8489,
+      "step": 4610
+    },
+    {
+      "epoch": 3.292943692088382,
+      "grad_norm": 4.5,
+      "learning_rate": 1.758339162934658e-06,
+      "loss": 0.7027,
+      "step": 4620
+    },
+    {
+      "epoch": 3.300071275837491,
+      "grad_norm": 4.59375,
+      "learning_rate": 1.6677590885745386e-06,
+      "loss": 0.7346,
+      "step": 4630
+    },
+    {
+      "epoch": 3.3071988595866,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.5795346594498161e-06,
+      "loss": 0.8141,
+      "step": 4640
+    },
+    {
+      "epoch": 3.3143264433357094,
+      "grad_norm": 4.40625,
+      "learning_rate": 1.4936701754947101e-06,
+      "loss": 0.695,
+      "step": 4650
+    },
+    {
+      "epoch": 3.321454027084818,
+      "grad_norm": 4.6875,
+      "learning_rate": 1.4101698216230253e-06,
+      "loss": 0.7755,
+      "step": 4660
+    },
+    {
+      "epoch": 3.3285816108339272,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.3290376675242023e-06,
+      "loss": 0.7076,
+      "step": 4670
+    },
+    {
+      "epoch": 3.3357091945830364,
+      "grad_norm": 5.59375,
+      "learning_rate": 1.2502776674649774e-06,
+      "loss": 0.8417,
+      "step": 4680
+    },
+    {
+      "epoch": 3.3428367783321455,
+      "grad_norm": 5.625,
+      "learning_rate": 1.1738936600966367e-06,
+      "loss": 0.8167,
+      "step": 4690
+    },
+    {
+      "epoch": 3.3499643620812547,
+      "grad_norm": 5.65625,
+      "learning_rate": 1.0998893682679479e-06,
+      "loss": 0.8371,
+      "step": 4700
+    },
+    {
+      "epoch": 3.3570919458303634,
+      "grad_norm": 5.3125,
+      "learning_rate": 1.0282683988436792e-06,
+      "loss": 0.8034,
+      "step": 4710
+    },
+    {
+      "epoch": 3.3642195295794726,
+      "grad_norm": 6.40625,
+      "learning_rate": 9.590342425288446e-07,
+      "loss": 0.8418,
+      "step": 4720
+    },
+    {
+      "epoch": 3.3713471133285817,
+      "grad_norm": 5.3125,
+      "learning_rate": 8.921902736985399e-07,
+      "loss": 0.7213,
+      "step": 4730
+    },
+    {
+      "epoch": 3.3784746970776904,
+      "grad_norm": 5.8125,
+      "learning_rate": 8.277397502335194e-07,
+      "loss": 0.707,
+      "step": 4740
+    },
+    {
+      "epoch": 3.3856022808267996,
+      "grad_norm": 5.40625,
+      "learning_rate": 7.656858133613498e-07,
+      "loss": 0.8238,
+      "step": 4750
+    },
+    {
+      "epoch": 3.3927298645759087,
+      "grad_norm": 5.1875,
+      "learning_rate": 7.060314875033835e-07,
+      "loss": 0.6872,
+      "step": 4760
+    },
+    {
+      "epoch": 3.399857448325018,
+      "grad_norm": 5.75,
+      "learning_rate": 6.487796801272983e-07,
+      "loss": 0.8421,
+      "step": 4770
+    },
+    {
+      "epoch": 3.406985032074127,
+      "grad_norm": 4.96875,
+      "learning_rate": 5.939331816054161e-07,
+      "loss": 0.784,
+      "step": 4780
+    },
+    {
+      "epoch": 3.4141126158232358,
+      "grad_norm": 4.75,
+      "learning_rate": 5.414946650786957e-07,
+      "loss": 0.728,
+      "step": 4790
+    },
+    {
+      "epoch": 3.421240199572345,
+      "grad_norm": 5.34375,
+      "learning_rate": 4.914666863264528e-07,
+      "loss": 0.8101,
+      "step": 4800
+    },
+    {
+      "epoch": 3.428367783321454,
+      "grad_norm": 4.875,
+      "learning_rate": 4.438516836417994e-07,
+      "loss": 0.8649,
+      "step": 4810
+    },
+    {
+      "epoch": 3.4354953670705632,
+      "grad_norm": 4.5625,
+      "learning_rate": 3.986519777127884e-07,
+      "loss": 0.7394,
+      "step": 4820
+    },
+    {
+      "epoch": 3.442622950819672,
+      "grad_norm": 6.3125,
+      "learning_rate": 3.558697715093207e-07,
+      "loss": 0.6994,
+      "step": 4830
+    },
+    {
+      "epoch": 3.449750534568781,
+      "grad_norm": 6.6875,
+      "learning_rate": 3.1550715017575895e-07,
+      "loss": 0.7571,
+      "step": 4840
+    },
+    {
+      "epoch": 3.4568781183178903,
+      "grad_norm": 5.4375,
+      "learning_rate": 2.775660809293368e-07,
+      "loss": 0.6913,
+      "step": 4850
+    },
+    {
+      "epoch": 3.4640057020669994,
+      "grad_norm": 5.90625,
+      "learning_rate": 2.4204841296424085e-07,
+      "loss": 0.7983,
+      "step": 4860
+    },
+    {
+      "epoch": 3.4711332858161086,
+      "grad_norm": 5.46875,
+      "learning_rate": 2.0895587736149414e-07,
+      "loss": 0.7831,
+      "step": 4870
+    },
+    {
+      "epoch": 3.4782608695652173,
+      "grad_norm": 6.21875,
+      "learning_rate": 1.7829008700460115e-07,
+      "loss": 0.7475,
+      "step": 4880
+    },
+    {
+      "epoch": 3.4853884533143265,
+      "grad_norm": 4.9375,
+      "learning_rate": 1.500525365009109e-07,
+      "loss": 0.7402,
+      "step": 4890
+    },
+    {
+      "epoch": 3.4925160370634356,
+      "grad_norm": 4.59375,
+      "learning_rate": 1.2424460210881395e-07,
+      "loss": 0.7151,
+      "step": 4900
+    },
+    {
+      "epoch": 3.4996436208125443,
+      "grad_norm": 5.0,
+      "learning_rate": 1.008675416706073e-07,
+      "loss": 0.7341,
+      "step": 4910
+    },
+    {
+      "epoch": 3.5067712045616535,
+      "grad_norm": 5.9375,
+      "learning_rate": 7.99224945512489e-08,
+      "loss": 0.717,
+      "step": 4920
+    },
+    {
+      "epoch": 3.5138987883107626,
+      "grad_norm": 5.65625,
+      "learning_rate": 6.141048158277429e-08,
+      "loss": 0.7112,
+      "step": 4930
+    },
+    {
+      "epoch": 3.521026372059872,
+      "grad_norm": 5.625,
+      "learning_rate": 4.5332405014592014e-08,
+      "loss": 0.747,
+      "step": 4940
+    },
+    {
+      "epoch": 3.528153955808981,
+      "grad_norm": 5.65625,
+      "learning_rate": 3.168904846945764e-08,
+      "loss": 0.8131,
+      "step": 4950
+    },
+    {
+      "epoch": 3.5352815395580897,
+      "grad_norm": 5.34375,
+      "learning_rate": 2.0481076905332075e-08,
+      "loss": 0.7257,
+      "step": 4960
+    },
+    {
+      "epoch": 3.542409123307199,
+      "grad_norm": 6.34375,
+      "learning_rate": 1.1709036582935318e-08,
+      "loss": 0.7629,
+      "step": 4970
+    },
+    {
+      "epoch": 3.549536707056308,
+      "grad_norm": 4.21875,
+      "learning_rate": 5.373355039128835e-09,
+      "loss": 0.681,
+      "step": 4980
+    },
+    {
+      "epoch": 3.556664290805417,
+      "grad_norm": 6.71875,
+      "learning_rate": 1.4743410661044455e-09,
+      "loss": 0.8106,
+      "step": 4990
+    },
+    {
+      "epoch": 3.563791874554526,
+      "grad_norm": 5.8125,
+      "learning_rate": 1.2184696296380082e-11,
+      "loss": 0.812,
+      "step": 5000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}