Upload YASIN V2
Browse files- checkpoint-8000/config.json +29 -0
- checkpoint-8000/generation_config.json +7 -0
- checkpoint-8000/model.safetensors +3 -0
- checkpoint-8000/optimizer.pt +3 -0
- checkpoint-8000/rng_state.pth +3 -0
- checkpoint-8000/scaler.pt +3 -0
- checkpoint-8000/scheduler.pt +3 -0
- checkpoint-8000/trainer_state.json +1154 -0
- checkpoint-8000/training_args.bin +3 -0
- checkpoint-9000/config.json +29 -0
- checkpoint-9000/generation_config.json +7 -0
- checkpoint-9000/model.safetensors +3 -0
- checkpoint-9000/optimizer.pt +3 -0
- checkpoint-9000/rng_state.pth +3 -0
- checkpoint-9000/scaler.pt +3 -0
- checkpoint-9000/scheduler.pt +3 -0
- checkpoint-9000/trainer_state.json +1294 -0
- checkpoint-9000/training_args.bin +3 -0
checkpoint-8000/config.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"YasinForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 1,
|
| 8 |
+
"dtype": "float32",
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 48,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 3072,
|
| 15 |
+
"max_position_embeddings": 1024,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "yasin",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 16,
|
| 21 |
+
"pretraining_tp": 1,
|
| 22 |
+
"rms_norm_eps": 1e-06,
|
| 23 |
+
"rope_scaling": null,
|
| 24 |
+
"rope_theta": 10000.0,
|
| 25 |
+
"tie_word_embeddings": false,
|
| 26 |
+
"transformers_version": "4.57.1",
|
| 27 |
+
"use_cache": false,
|
| 28 |
+
"vocab_size": 50257
|
| 29 |
+
}
|
checkpoint-8000/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.57.1",
|
| 6 |
+
"use_cache": false
|
| 7 |
+
}
|
checkpoint-8000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e54a94b4a0394ce9f2c1148d17a5253da67db0bb8efc9f5c5144365c37b342f0
|
| 3 |
+
size 912876512
|
checkpoint-8000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:63e12e5a118b9d112314c2e99ddda40792a9ff0cf6dac2c1949896784d92626d
|
| 3 |
+
size 1825846859
|
checkpoint-8000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61c19bab1174704a4a4441475683bf1270277af15d2e2c95e964789128e482c4
|
| 3 |
+
size 14645
|
checkpoint-8000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:022915fa8ac65fae0f13853519b8aa7ee79efd632a399c1f64c6b87bab0b057a
|
| 3 |
+
size 1383
|
checkpoint-8000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:703fbf281c65b18b081956dbdc91da7ba5d0f7374dd0632e1ca82a90f2073597
|
| 3 |
+
size 1465
|
checkpoint-8000/trainer_state.json
ADDED
|
@@ -0,0 +1,1154 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.008,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 8000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 5e-05,
|
| 14 |
+
"grad_norm": 4.426811695098877,
|
| 15 |
+
"learning_rate": 4.999755000000001e-05,
|
| 16 |
+
"loss": 6.3755,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0001,
|
| 21 |
+
"grad_norm": 1.7730491161346436,
|
| 22 |
+
"learning_rate": 4.999505e-05,
|
| 23 |
+
"loss": 3.3233,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.00015,
|
| 28 |
+
"grad_norm": 1.4473446607589722,
|
| 29 |
+
"learning_rate": 4.999255e-05,
|
| 30 |
+
"loss": 2.7523,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.0002,
|
| 35 |
+
"grad_norm": 1.4645352363586426,
|
| 36 |
+
"learning_rate": 4.9990050000000004e-05,
|
| 37 |
+
"loss": 2.4618,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.00025,
|
| 42 |
+
"grad_norm": 1.1990822553634644,
|
| 43 |
+
"learning_rate": 4.998755e-05,
|
| 44 |
+
"loss": 2.3417,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.0003,
|
| 49 |
+
"grad_norm": 3.0459673404693604,
|
| 50 |
+
"learning_rate": 4.998505e-05,
|
| 51 |
+
"loss": 2.2581,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.00035,
|
| 56 |
+
"grad_norm": 1.2380167245864868,
|
| 57 |
+
"learning_rate": 4.998255e-05,
|
| 58 |
+
"loss": 2.1563,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.0004,
|
| 63 |
+
"grad_norm": 1.2060939073562622,
|
| 64 |
+
"learning_rate": 4.998005e-05,
|
| 65 |
+
"loss": 1.8521,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.00045,
|
| 70 |
+
"grad_norm": 0.9207198619842529,
|
| 71 |
+
"learning_rate": 4.9977550000000004e-05,
|
| 72 |
+
"loss": 1.9201,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.0005,
|
| 77 |
+
"grad_norm": 1.0762739181518555,
|
| 78 |
+
"learning_rate": 4.997505e-05,
|
| 79 |
+
"loss": 1.8104,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.00055,
|
| 84 |
+
"grad_norm": 1.0367337465286255,
|
| 85 |
+
"learning_rate": 4.997255e-05,
|
| 86 |
+
"loss": 1.8756,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.0006,
|
| 91 |
+
"grad_norm": 0.9291325211524963,
|
| 92 |
+
"learning_rate": 4.997005e-05,
|
| 93 |
+
"loss": 1.7404,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.00065,
|
| 98 |
+
"grad_norm": 0.7452297210693359,
|
| 99 |
+
"learning_rate": 4.996755e-05,
|
| 100 |
+
"loss": 1.6843,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.0007,
|
| 105 |
+
"grad_norm": 0.9317042231559753,
|
| 106 |
+
"learning_rate": 4.9965050000000004e-05,
|
| 107 |
+
"loss": 1.7006,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.00075,
|
| 112 |
+
"grad_norm": 1.0354136228561401,
|
| 113 |
+
"learning_rate": 4.9962550000000005e-05,
|
| 114 |
+
"loss": 1.5652,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.0008,
|
| 119 |
+
"grad_norm": 1.050106406211853,
|
| 120 |
+
"learning_rate": 4.996005e-05,
|
| 121 |
+
"loss": 1.5966,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.00085,
|
| 126 |
+
"grad_norm": 0.9646019339561462,
|
| 127 |
+
"learning_rate": 4.995755e-05,
|
| 128 |
+
"loss": 1.6297,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0009,
|
| 133 |
+
"grad_norm": 0.9152287840843201,
|
| 134 |
+
"learning_rate": 4.995505e-05,
|
| 135 |
+
"loss": 1.6328,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.00095,
|
| 140 |
+
"grad_norm": 0.9403690099716187,
|
| 141 |
+
"learning_rate": 4.995255e-05,
|
| 142 |
+
"loss": 1.6056,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.001,
|
| 147 |
+
"grad_norm": 1.1822874546051025,
|
| 148 |
+
"learning_rate": 4.9950050000000005e-05,
|
| 149 |
+
"loss": 1.6016,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.00105,
|
| 154 |
+
"grad_norm": 1.3227542638778687,
|
| 155 |
+
"learning_rate": 4.9947550000000006e-05,
|
| 156 |
+
"loss": 1.5448,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.0011,
|
| 161 |
+
"grad_norm": 1.0503350496292114,
|
| 162 |
+
"learning_rate": 4.994505000000001e-05,
|
| 163 |
+
"loss": 1.6355,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.00115,
|
| 168 |
+
"grad_norm": 1.1647204160690308,
|
| 169 |
+
"learning_rate": 4.994255e-05,
|
| 170 |
+
"loss": 1.4673,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.0012,
|
| 175 |
+
"grad_norm": 0.7281339168548584,
|
| 176 |
+
"learning_rate": 4.994005e-05,
|
| 177 |
+
"loss": 1.4641,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.00125,
|
| 182 |
+
"grad_norm": 1.2438446283340454,
|
| 183 |
+
"learning_rate": 4.9937550000000004e-05,
|
| 184 |
+
"loss": 1.5177,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.0013,
|
| 189 |
+
"grad_norm": 0.7967873811721802,
|
| 190 |
+
"learning_rate": 4.993505e-05,
|
| 191 |
+
"loss": 1.4399,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.00135,
|
| 196 |
+
"grad_norm": 0.8938255310058594,
|
| 197 |
+
"learning_rate": 4.993255e-05,
|
| 198 |
+
"loss": 1.4453,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.0014,
|
| 203 |
+
"grad_norm": 1.4659217596054077,
|
| 204 |
+
"learning_rate": 4.993005e-05,
|
| 205 |
+
"loss": 1.6517,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.00145,
|
| 210 |
+
"grad_norm": 0.7856793403625488,
|
| 211 |
+
"learning_rate": 4.992755e-05,
|
| 212 |
+
"loss": 1.4721,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.0015,
|
| 217 |
+
"grad_norm": 0.6772142648696899,
|
| 218 |
+
"learning_rate": 4.9925050000000004e-05,
|
| 219 |
+
"loss": 1.5314,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.00155,
|
| 224 |
+
"grad_norm": 0.7613831758499146,
|
| 225 |
+
"learning_rate": 4.992255e-05,
|
| 226 |
+
"loss": 1.4603,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0016,
|
| 231 |
+
"grad_norm": 1.1216529607772827,
|
| 232 |
+
"learning_rate": 4.992005e-05,
|
| 233 |
+
"loss": 1.54,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.00165,
|
| 238 |
+
"grad_norm": 0.7359323501586914,
|
| 239 |
+
"learning_rate": 4.991755e-05,
|
| 240 |
+
"loss": 1.5121,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0017,
|
| 245 |
+
"grad_norm": 0.8343626260757446,
|
| 246 |
+
"learning_rate": 4.991505e-05,
|
| 247 |
+
"loss": 1.6107,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.00175,
|
| 252 |
+
"grad_norm": 0.9381140470504761,
|
| 253 |
+
"learning_rate": 4.9912550000000004e-05,
|
| 254 |
+
"loss": 1.5059,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.0018,
|
| 259 |
+
"grad_norm": 0.7317638993263245,
|
| 260 |
+
"learning_rate": 4.9910050000000005e-05,
|
| 261 |
+
"loss": 1.5621,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.00185,
|
| 266 |
+
"grad_norm": 0.8273525834083557,
|
| 267 |
+
"learning_rate": 4.990755e-05,
|
| 268 |
+
"loss": 1.4737,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.0019,
|
| 273 |
+
"grad_norm": 0.7206686735153198,
|
| 274 |
+
"learning_rate": 4.990505e-05,
|
| 275 |
+
"loss": 1.5242,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.00195,
|
| 280 |
+
"grad_norm": 0.9094993472099304,
|
| 281 |
+
"learning_rate": 4.990255e-05,
|
| 282 |
+
"loss": 1.5336,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.002,
|
| 287 |
+
"grad_norm": 0.7432169914245605,
|
| 288 |
+
"learning_rate": 4.9900050000000003e-05,
|
| 289 |
+
"loss": 1.4913,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.00205,
|
| 294 |
+
"grad_norm": 1.337276816368103,
|
| 295 |
+
"learning_rate": 4.9897550000000005e-05,
|
| 296 |
+
"loss": 1.5643,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.0021,
|
| 301 |
+
"grad_norm": 0.6664676666259766,
|
| 302 |
+
"learning_rate": 4.9895050000000006e-05,
|
| 303 |
+
"loss": 1.4728,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.00215,
|
| 308 |
+
"grad_norm": 0.7946372628211975,
|
| 309 |
+
"learning_rate": 4.989255000000001e-05,
|
| 310 |
+
"loss": 1.4832,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.0022,
|
| 315 |
+
"grad_norm": 0.9259161949157715,
|
| 316 |
+
"learning_rate": 4.989005e-05,
|
| 317 |
+
"loss": 1.5674,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.00225,
|
| 322 |
+
"grad_norm": 0.9703247547149658,
|
| 323 |
+
"learning_rate": 4.988755e-05,
|
| 324 |
+
"loss": 1.3973,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.0023,
|
| 329 |
+
"grad_norm": 0.7390087842941284,
|
| 330 |
+
"learning_rate": 4.988505e-05,
|
| 331 |
+
"loss": 1.3919,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.00235,
|
| 336 |
+
"grad_norm": 0.768277645111084,
|
| 337 |
+
"learning_rate": 4.988255e-05,
|
| 338 |
+
"loss": 1.4815,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.0024,
|
| 343 |
+
"grad_norm": 0.76382976770401,
|
| 344 |
+
"learning_rate": 4.988005e-05,
|
| 345 |
+
"loss": 1.4458,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.00245,
|
| 350 |
+
"grad_norm": 0.7780851721763611,
|
| 351 |
+
"learning_rate": 4.987755e-05,
|
| 352 |
+
"loss": 1.4382,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.0025,
|
| 357 |
+
"grad_norm": 0.7184464335441589,
|
| 358 |
+
"learning_rate": 4.987505e-05,
|
| 359 |
+
"loss": 1.521,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.00255,
|
| 364 |
+
"grad_norm": 0.6212390065193176,
|
| 365 |
+
"learning_rate": 4.9872550000000004e-05,
|
| 366 |
+
"loss": 1.3889,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.0026,
|
| 371 |
+
"grad_norm": 0.8539580702781677,
|
| 372 |
+
"learning_rate": 4.987005e-05,
|
| 373 |
+
"loss": 1.415,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.00265,
|
| 378 |
+
"grad_norm": 0.7129775881767273,
|
| 379 |
+
"learning_rate": 4.986755e-05,
|
| 380 |
+
"loss": 1.4102,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.0027,
|
| 385 |
+
"grad_norm": 0.5899195671081543,
|
| 386 |
+
"learning_rate": 4.986505e-05,
|
| 387 |
+
"loss": 1.3074,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.00275,
|
| 392 |
+
"grad_norm": 0.6940101981163025,
|
| 393 |
+
"learning_rate": 4.986255e-05,
|
| 394 |
+
"loss": 1.3748,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.0028,
|
| 399 |
+
"grad_norm": 0.6420891880989075,
|
| 400 |
+
"learning_rate": 4.9860050000000004e-05,
|
| 401 |
+
"loss": 1.4089,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.00285,
|
| 406 |
+
"grad_norm": 0.8561428189277649,
|
| 407 |
+
"learning_rate": 4.9857550000000005e-05,
|
| 408 |
+
"loss": 1.4052,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.0029,
|
| 413 |
+
"grad_norm": 0.6900970935821533,
|
| 414 |
+
"learning_rate": 4.9855050000000006e-05,
|
| 415 |
+
"loss": 1.4323,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.00295,
|
| 420 |
+
"grad_norm": 0.8071371912956238,
|
| 421 |
+
"learning_rate": 4.985255e-05,
|
| 422 |
+
"loss": 1.3485,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.003,
|
| 427 |
+
"grad_norm": 0.6493249535560608,
|
| 428 |
+
"learning_rate": 4.985005e-05,
|
| 429 |
+
"loss": 1.4321,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.00305,
|
| 434 |
+
"grad_norm": 0.8712514042854309,
|
| 435 |
+
"learning_rate": 4.9847550000000004e-05,
|
| 436 |
+
"loss": 1.406,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.0031,
|
| 441 |
+
"grad_norm": 0.8464570045471191,
|
| 442 |
+
"learning_rate": 4.9845050000000005e-05,
|
| 443 |
+
"loss": 1.3322,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.00315,
|
| 448 |
+
"grad_norm": 0.7779001593589783,
|
| 449 |
+
"learning_rate": 4.9842550000000006e-05,
|
| 450 |
+
"loss": 1.3905,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.0032,
|
| 455 |
+
"grad_norm": 0.8314748406410217,
|
| 456 |
+
"learning_rate": 4.984005000000001e-05,
|
| 457 |
+
"loss": 1.4134,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.00325,
|
| 462 |
+
"grad_norm": 0.7180835604667664,
|
| 463 |
+
"learning_rate": 4.983755e-05,
|
| 464 |
+
"loss": 1.3629,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.0033,
|
| 469 |
+
"grad_norm": 0.6165933012962341,
|
| 470 |
+
"learning_rate": 4.9835049999999996e-05,
|
| 471 |
+
"loss": 1.3833,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.00335,
|
| 476 |
+
"grad_norm": 1.8662049770355225,
|
| 477 |
+
"learning_rate": 4.983255e-05,
|
| 478 |
+
"loss": 1.3557,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.0034,
|
| 483 |
+
"grad_norm": 0.9091550707817078,
|
| 484 |
+
"learning_rate": 4.983005e-05,
|
| 485 |
+
"loss": 1.4092,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.00345,
|
| 490 |
+
"grad_norm": 0.9908722639083862,
|
| 491 |
+
"learning_rate": 4.982755e-05,
|
| 492 |
+
"loss": 1.3465,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.0035,
|
| 497 |
+
"grad_norm": 0.862427830696106,
|
| 498 |
+
"learning_rate": 4.982505e-05,
|
| 499 |
+
"loss": 1.3143,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.00355,
|
| 504 |
+
"grad_norm": 0.755211591720581,
|
| 505 |
+
"learning_rate": 4.982255e-05,
|
| 506 |
+
"loss": 1.358,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.0036,
|
| 511 |
+
"grad_norm": 0.7872670888900757,
|
| 512 |
+
"learning_rate": 4.9820050000000004e-05,
|
| 513 |
+
"loss": 1.3952,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.00365,
|
| 518 |
+
"grad_norm": 1.5124619007110596,
|
| 519 |
+
"learning_rate": 4.981755e-05,
|
| 520 |
+
"loss": 1.2929,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.0037,
|
| 525 |
+
"grad_norm": 0.7712079286575317,
|
| 526 |
+
"learning_rate": 4.981505e-05,
|
| 527 |
+
"loss": 1.3429,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.00375,
|
| 532 |
+
"grad_norm": 0.7001494765281677,
|
| 533 |
+
"learning_rate": 4.981255e-05,
|
| 534 |
+
"loss": 1.272,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.0038,
|
| 539 |
+
"grad_norm": 0.674104630947113,
|
| 540 |
+
"learning_rate": 4.981005e-05,
|
| 541 |
+
"loss": 1.3891,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.00385,
|
| 546 |
+
"grad_norm": 1.0494478940963745,
|
| 547 |
+
"learning_rate": 4.9807550000000004e-05,
|
| 548 |
+
"loss": 1.333,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.0039,
|
| 553 |
+
"grad_norm": 0.6674365401268005,
|
| 554 |
+
"learning_rate": 4.9805050000000005e-05,
|
| 555 |
+
"loss": 1.4418,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.00395,
|
| 560 |
+
"grad_norm": 0.7624682784080505,
|
| 561 |
+
"learning_rate": 4.9802550000000007e-05,
|
| 562 |
+
"loss": 1.2837,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.004,
|
| 567 |
+
"grad_norm": 1.5128511190414429,
|
| 568 |
+
"learning_rate": 4.980005e-05,
|
| 569 |
+
"loss": 1.2609,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.00405,
|
| 574 |
+
"grad_norm": 0.5422250628471375,
|
| 575 |
+
"learning_rate": 4.979755e-05,
|
| 576 |
+
"loss": 1.3253,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.0041,
|
| 581 |
+
"grad_norm": 0.63419508934021,
|
| 582 |
+
"learning_rate": 4.9795050000000004e-05,
|
| 583 |
+
"loss": 1.3369,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.00415,
|
| 588 |
+
"grad_norm": 0.7608025670051575,
|
| 589 |
+
"learning_rate": 4.9792550000000005e-05,
|
| 590 |
+
"loss": 1.3186,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.0042,
|
| 595 |
+
"grad_norm": 0.6282067894935608,
|
| 596 |
+
"learning_rate": 4.9790050000000006e-05,
|
| 597 |
+
"loss": 1.3755,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.00425,
|
| 602 |
+
"grad_norm": 0.7882742881774902,
|
| 603 |
+
"learning_rate": 4.978755000000001e-05,
|
| 604 |
+
"loss": 1.3326,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.0043,
|
| 609 |
+
"grad_norm": 0.6797521710395813,
|
| 610 |
+
"learning_rate": 4.978505e-05,
|
| 611 |
+
"loss": 1.2774,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.00435,
|
| 616 |
+
"grad_norm": 0.6129476428031921,
|
| 617 |
+
"learning_rate": 4.978255e-05,
|
| 618 |
+
"loss": 1.324,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.0044,
|
| 623 |
+
"grad_norm": 0.7544896006584167,
|
| 624 |
+
"learning_rate": 4.978005e-05,
|
| 625 |
+
"loss": 1.299,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.00445,
|
| 630 |
+
"grad_norm": 0.6218116283416748,
|
| 631 |
+
"learning_rate": 4.977755e-05,
|
| 632 |
+
"loss": 1.3084,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.0045,
|
| 637 |
+
"grad_norm": 0.5591554045677185,
|
| 638 |
+
"learning_rate": 4.977505e-05,
|
| 639 |
+
"loss": 1.3001,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.00455,
|
| 644 |
+
"grad_norm": 0.854743242263794,
|
| 645 |
+
"learning_rate": 4.977255e-05,
|
| 646 |
+
"loss": 1.3357,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.0046,
|
| 651 |
+
"grad_norm": 0.7265657186508179,
|
| 652 |
+
"learning_rate": 4.977005e-05,
|
| 653 |
+
"loss": 1.3362,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.00465,
|
| 658 |
+
"grad_norm": 0.8089432120323181,
|
| 659 |
+
"learning_rate": 4.9767550000000004e-05,
|
| 660 |
+
"loss": 1.3087,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.0047,
|
| 665 |
+
"grad_norm": 0.8999931216239929,
|
| 666 |
+
"learning_rate": 4.976505e-05,
|
| 667 |
+
"loss": 1.3894,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.00475,
|
| 672 |
+
"grad_norm": 0.7705693244934082,
|
| 673 |
+
"learning_rate": 4.976255e-05,
|
| 674 |
+
"loss": 1.2585,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.0048,
|
| 679 |
+
"grad_norm": 0.7201912999153137,
|
| 680 |
+
"learning_rate": 4.976005e-05,
|
| 681 |
+
"loss": 1.221,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.00485,
|
| 686 |
+
"grad_norm": 0.6199264526367188,
|
| 687 |
+
"learning_rate": 4.975755e-05,
|
| 688 |
+
"loss": 1.2878,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.0049,
|
| 693 |
+
"grad_norm": 0.6883070468902588,
|
| 694 |
+
"learning_rate": 4.9755050000000004e-05,
|
| 695 |
+
"loss": 1.3032,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.00495,
|
| 700 |
+
"grad_norm": 0.524889349937439,
|
| 701 |
+
"learning_rate": 4.9752550000000005e-05,
|
| 702 |
+
"loss": 1.2706,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.005,
|
| 707 |
+
"grad_norm": 0.7067034244537354,
|
| 708 |
+
"learning_rate": 4.9750050000000007e-05,
|
| 709 |
+
"loss": 1.2424,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.00505,
|
| 714 |
+
"grad_norm": 0.7097471356391907,
|
| 715 |
+
"learning_rate": 4.974755e-05,
|
| 716 |
+
"loss": 1.3175,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.0051,
|
| 721 |
+
"grad_norm": 0.6991677284240723,
|
| 722 |
+
"learning_rate": 4.974505e-05,
|
| 723 |
+
"loss": 1.2983,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.00515,
|
| 728 |
+
"grad_norm": 1.4025177955627441,
|
| 729 |
+
"learning_rate": 4.9742550000000004e-05,
|
| 730 |
+
"loss": 1.2695,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.0052,
|
| 735 |
+
"grad_norm": 0.6075018048286438,
|
| 736 |
+
"learning_rate": 4.9740050000000005e-05,
|
| 737 |
+
"loss": 1.2973,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.00525,
|
| 742 |
+
"grad_norm": 0.5344257950782776,
|
| 743 |
+
"learning_rate": 4.9737550000000006e-05,
|
| 744 |
+
"loss": 1.2353,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.0053,
|
| 749 |
+
"grad_norm": 0.5963959693908691,
|
| 750 |
+
"learning_rate": 4.973505e-05,
|
| 751 |
+
"loss": 1.2359,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.00535,
|
| 756 |
+
"grad_norm": 0.9493517279624939,
|
| 757 |
+
"learning_rate": 4.973255e-05,
|
| 758 |
+
"loss": 1.252,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.0054,
|
| 763 |
+
"grad_norm": 0.6466087102890015,
|
| 764 |
+
"learning_rate": 4.9730050000000003e-05,
|
| 765 |
+
"loss": 1.2012,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.00545,
|
| 770 |
+
"grad_norm": 0.6407445669174194,
|
| 771 |
+
"learning_rate": 4.972755e-05,
|
| 772 |
+
"loss": 1.2615,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.0055,
|
| 777 |
+
"grad_norm": 0.617421567440033,
|
| 778 |
+
"learning_rate": 4.972505e-05,
|
| 779 |
+
"loss": 1.2718,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.00555,
|
| 784 |
+
"grad_norm": 0.5649735331535339,
|
| 785 |
+
"learning_rate": 4.972255e-05,
|
| 786 |
+
"loss": 1.2439,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.0056,
|
| 791 |
+
"grad_norm": 0.675645649433136,
|
| 792 |
+
"learning_rate": 4.972005e-05,
|
| 793 |
+
"loss": 1.2847,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.00565,
|
| 798 |
+
"grad_norm": 0.6887643337249756,
|
| 799 |
+
"learning_rate": 4.971755e-05,
|
| 800 |
+
"loss": 1.2464,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.0057,
|
| 805 |
+
"grad_norm": 0.6572535037994385,
|
| 806 |
+
"learning_rate": 4.9715050000000004e-05,
|
| 807 |
+
"loss": 1.2517,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.00575,
|
| 812 |
+
"grad_norm": 0.6076003313064575,
|
| 813 |
+
"learning_rate": 4.971255e-05,
|
| 814 |
+
"loss": 1.2183,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.0058,
|
| 819 |
+
"grad_norm": 0.6134760975837708,
|
| 820 |
+
"learning_rate": 4.971005e-05,
|
| 821 |
+
"loss": 1.2303,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.00585,
|
| 826 |
+
"grad_norm": 0.6791936159133911,
|
| 827 |
+
"learning_rate": 4.970755e-05,
|
| 828 |
+
"loss": 1.2351,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.0059,
|
| 833 |
+
"grad_norm": 0.6307588219642639,
|
| 834 |
+
"learning_rate": 4.970505e-05,
|
| 835 |
+
"loss": 1.254,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.00595,
|
| 840 |
+
"grad_norm": 1.1272459030151367,
|
| 841 |
+
"learning_rate": 4.9702550000000004e-05,
|
| 842 |
+
"loss": 1.265,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.006,
|
| 847 |
+
"grad_norm": 0.5415930151939392,
|
| 848 |
+
"learning_rate": 4.9700050000000005e-05,
|
| 849 |
+
"loss": 1.2068,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.00605,
|
| 854 |
+
"grad_norm": 0.5450507402420044,
|
| 855 |
+
"learning_rate": 4.969755000000001e-05,
|
| 856 |
+
"loss": 1.2408,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.0061,
|
| 861 |
+
"grad_norm": 0.6528737545013428,
|
| 862 |
+
"learning_rate": 4.969505e-05,
|
| 863 |
+
"loss": 1.2035,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.00615,
|
| 868 |
+
"grad_norm": 0.6139540672302246,
|
| 869 |
+
"learning_rate": 4.969255e-05,
|
| 870 |
+
"loss": 1.2077,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.0062,
|
| 875 |
+
"grad_norm": 0.6503865718841553,
|
| 876 |
+
"learning_rate": 4.9690050000000004e-05,
|
| 877 |
+
"loss": 1.2055,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.00625,
|
| 882 |
+
"grad_norm": 0.6888736486434937,
|
| 883 |
+
"learning_rate": 4.9687550000000005e-05,
|
| 884 |
+
"loss": 1.2161,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.0063,
|
| 889 |
+
"grad_norm": 1.1457703113555908,
|
| 890 |
+
"learning_rate": 4.968505e-05,
|
| 891 |
+
"loss": 1.1939,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.00635,
|
| 896 |
+
"grad_norm": 0.5861665606498718,
|
| 897 |
+
"learning_rate": 4.968255e-05,
|
| 898 |
+
"loss": 1.2177,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.0064,
|
| 903 |
+
"grad_norm": 0.8666422963142395,
|
| 904 |
+
"learning_rate": 4.968005e-05,
|
| 905 |
+
"loss": 1.1945,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.00645,
|
| 910 |
+
"grad_norm": 0.6094856858253479,
|
| 911 |
+
"learning_rate": 4.9677550000000003e-05,
|
| 912 |
+
"loss": 1.2394,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.0065,
|
| 917 |
+
"grad_norm": 0.6510404348373413,
|
| 918 |
+
"learning_rate": 4.967505e-05,
|
| 919 |
+
"loss": 1.1764,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.00655,
|
| 924 |
+
"grad_norm": 0.7044185996055603,
|
| 925 |
+
"learning_rate": 4.967255e-05,
|
| 926 |
+
"loss": 1.1951,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.0066,
|
| 931 |
+
"grad_norm": 0.6577385663986206,
|
| 932 |
+
"learning_rate": 4.967005e-05,
|
| 933 |
+
"loss": 1.2186,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.00665,
|
| 938 |
+
"grad_norm": 0.6468052268028259,
|
| 939 |
+
"learning_rate": 4.966755e-05,
|
| 940 |
+
"loss": 1.2491,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.0067,
|
| 945 |
+
"grad_norm": 0.6153103709220886,
|
| 946 |
+
"learning_rate": 4.966505e-05,
|
| 947 |
+
"loss": 1.25,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.00675,
|
| 952 |
+
"grad_norm": 0.6066887974739075,
|
| 953 |
+
"learning_rate": 4.9662550000000004e-05,
|
| 954 |
+
"loss": 1.1852,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.0068,
|
| 959 |
+
"grad_norm": 0.5735403895378113,
|
| 960 |
+
"learning_rate": 4.9660050000000006e-05,
|
| 961 |
+
"loss": 1.2139,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.00685,
|
| 966 |
+
"grad_norm": 0.6247462630271912,
|
| 967 |
+
"learning_rate": 4.965755e-05,
|
| 968 |
+
"loss": 1.1597,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.0069,
|
| 973 |
+
"grad_norm": 0.6690845489501953,
|
| 974 |
+
"learning_rate": 4.965505e-05,
|
| 975 |
+
"loss": 1.2095,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.00695,
|
| 980 |
+
"grad_norm": 0.5683197975158691,
|
| 981 |
+
"learning_rate": 4.965255e-05,
|
| 982 |
+
"loss": 1.3077,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.007,
|
| 987 |
+
"grad_norm": 0.6314153671264648,
|
| 988 |
+
"learning_rate": 4.9650050000000004e-05,
|
| 989 |
+
"loss": 1.2556,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.00705,
|
| 994 |
+
"grad_norm": 0.641153872013092,
|
| 995 |
+
"learning_rate": 4.9647550000000005e-05,
|
| 996 |
+
"loss": 1.1643,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.0071,
|
| 1001 |
+
"grad_norm": 0.6113337278366089,
|
| 1002 |
+
"learning_rate": 4.964505000000001e-05,
|
| 1003 |
+
"loss": 1.1688,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.00715,
|
| 1008 |
+
"grad_norm": 0.5102031230926514,
|
| 1009 |
+
"learning_rate": 4.964255e-05,
|
| 1010 |
+
"loss": 1.1835,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.0072,
|
| 1015 |
+
"grad_norm": 0.6926486492156982,
|
| 1016 |
+
"learning_rate": 4.964005e-05,
|
| 1017 |
+
"loss": 1.1704,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.00725,
|
| 1022 |
+
"grad_norm": 0.594326376914978,
|
| 1023 |
+
"learning_rate": 4.9637550000000004e-05,
|
| 1024 |
+
"loss": 1.1434,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.0073,
|
| 1029 |
+
"grad_norm": 0.5609344840049744,
|
| 1030 |
+
"learning_rate": 4.963505e-05,
|
| 1031 |
+
"loss": 1.2389,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.00735,
|
| 1036 |
+
"grad_norm": 0.6464629173278809,
|
| 1037 |
+
"learning_rate": 4.963255e-05,
|
| 1038 |
+
"loss": 1.2097,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.0074,
|
| 1043 |
+
"grad_norm": 0.5782963633537292,
|
| 1044 |
+
"learning_rate": 4.963005e-05,
|
| 1045 |
+
"loss": 1.2607,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.00745,
|
| 1050 |
+
"grad_norm": 0.7301307320594788,
|
| 1051 |
+
"learning_rate": 4.962755e-05,
|
| 1052 |
+
"loss": 1.1759,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.0075,
|
| 1057 |
+
"grad_norm": 0.5845964550971985,
|
| 1058 |
+
"learning_rate": 4.9625050000000004e-05,
|
| 1059 |
+
"loss": 1.1806,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.00755,
|
| 1064 |
+
"grad_norm": 0.618295431137085,
|
| 1065 |
+
"learning_rate": 4.962255e-05,
|
| 1066 |
+
"loss": 1.1339,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.0076,
|
| 1071 |
+
"grad_norm": 0.5799853801727295,
|
| 1072 |
+
"learning_rate": 4.962005e-05,
|
| 1073 |
+
"loss": 1.1641,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.00765,
|
| 1078 |
+
"grad_norm": 0.5227323174476624,
|
| 1079 |
+
"learning_rate": 4.961755e-05,
|
| 1080 |
+
"loss": 1.2375,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.0077,
|
| 1085 |
+
"grad_norm": 0.699111819267273,
|
| 1086 |
+
"learning_rate": 4.961505e-05,
|
| 1087 |
+
"loss": 1.0943,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.00775,
|
| 1092 |
+
"grad_norm": 0.5230436325073242,
|
| 1093 |
+
"learning_rate": 4.961255e-05,
|
| 1094 |
+
"loss": 1.1762,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.0078,
|
| 1099 |
+
"grad_norm": 0.5776082873344421,
|
| 1100 |
+
"learning_rate": 4.9610050000000005e-05,
|
| 1101 |
+
"loss": 1.18,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.00785,
|
| 1106 |
+
"grad_norm": 0.584697425365448,
|
| 1107 |
+
"learning_rate": 4.9607550000000006e-05,
|
| 1108 |
+
"loss": 1.2878,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.0079,
|
| 1113 |
+
"grad_norm": 0.66282057762146,
|
| 1114 |
+
"learning_rate": 4.960505e-05,
|
| 1115 |
+
"loss": 1.219,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.00795,
|
| 1120 |
+
"grad_norm": 1.2405346632003784,
|
| 1121 |
+
"learning_rate": 4.960255e-05,
|
| 1122 |
+
"loss": 1.0984,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.008,
|
| 1127 |
+
"grad_norm": 0.6882185339927673,
|
| 1128 |
+
"learning_rate": 4.960005e-05,
|
| 1129 |
+
"loss": 1.2291,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
}
|
| 1132 |
+
],
|
| 1133 |
+
"logging_steps": 50,
|
| 1134 |
+
"max_steps": 1000000,
|
| 1135 |
+
"num_input_tokens_seen": 0,
|
| 1136 |
+
"num_train_epochs": 9223372036854775807,
|
| 1137 |
+
"save_steps": 1000,
|
| 1138 |
+
"stateful_callbacks": {
|
| 1139 |
+
"TrainerControl": {
|
| 1140 |
+
"args": {
|
| 1141 |
+
"should_epoch_stop": false,
|
| 1142 |
+
"should_evaluate": false,
|
| 1143 |
+
"should_log": false,
|
| 1144 |
+
"should_save": true,
|
| 1145 |
+
"should_training_stop": false
|
| 1146 |
+
},
|
| 1147 |
+
"attributes": {}
|
| 1148 |
+
}
|
| 1149 |
+
},
|
| 1150 |
+
"total_flos": 1.49121398734848e+17,
|
| 1151 |
+
"train_batch_size": 4,
|
| 1152 |
+
"trial_name": null,
|
| 1153 |
+
"trial_params": null
|
| 1154 |
+
}
|
checkpoint-8000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f1f06799524e8433f4ea5813803d7bd7b4539e7b44d7846bd777c6941f1e37c
|
| 3 |
+
size 5841
|
checkpoint-9000/config.json
ADDED
|
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"YasinForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 1,
|
| 8 |
+
"dtype": "float32",
|
| 9 |
+
"eos_token_id": 2,
|
| 10 |
+
"head_dim": 48,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 3072,
|
| 15 |
+
"max_position_embeddings": 1024,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "yasin",
|
| 18 |
+
"num_attention_heads": 16,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 16,
|
| 21 |
+
"pretraining_tp": 1,
|
| 22 |
+
"rms_norm_eps": 1e-06,
|
| 23 |
+
"rope_scaling": null,
|
| 24 |
+
"rope_theta": 10000.0,
|
| 25 |
+
"tie_word_embeddings": false,
|
| 26 |
+
"transformers_version": "4.57.1",
|
| 27 |
+
"use_cache": false,
|
| 28 |
+
"vocab_size": 50257
|
| 29 |
+
}
|
checkpoint-9000/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.57.1",
|
| 6 |
+
"use_cache": false
|
| 7 |
+
}
|
checkpoint-9000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f25d4fd6edbfc4cf8153ac40dafe3105e19751bce39ed2b4dd71ee5ca8f07409
|
| 3 |
+
size 912876512
|
checkpoint-9000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ccf591c09ac9a236d05591e4c5adedeed47e23cba5dec29440a520451a48bb5
|
| 3 |
+
size 1825846859
|
checkpoint-9000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61c19bab1174704a4a4441475683bf1270277af15d2e2c95e964789128e482c4
|
| 3 |
+
size 14645
|
checkpoint-9000/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1b0ffe1c5339266a7640f64d74cbe7720984c034c061903c7cafd92501da2b55
|
| 3 |
+
size 1383
|
checkpoint-9000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eba484843bfc8f44ba71a7c82b9990f51769097f58eba94a55a38f79aa56cc8a
|
| 3 |
+
size 1465
|
checkpoint-9000/trainer_state.json
ADDED
|
@@ -0,0 +1,1294 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.009,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 9000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 5e-05,
|
| 14 |
+
"grad_norm": 4.426811695098877,
|
| 15 |
+
"learning_rate": 4.999755000000001e-05,
|
| 16 |
+
"loss": 6.3755,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0001,
|
| 21 |
+
"grad_norm": 1.7730491161346436,
|
| 22 |
+
"learning_rate": 4.999505e-05,
|
| 23 |
+
"loss": 3.3233,
|
| 24 |
+
"step": 100
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.00015,
|
| 28 |
+
"grad_norm": 1.4473446607589722,
|
| 29 |
+
"learning_rate": 4.999255e-05,
|
| 30 |
+
"loss": 2.7523,
|
| 31 |
+
"step": 150
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.0002,
|
| 35 |
+
"grad_norm": 1.4645352363586426,
|
| 36 |
+
"learning_rate": 4.9990050000000004e-05,
|
| 37 |
+
"loss": 2.4618,
|
| 38 |
+
"step": 200
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.00025,
|
| 42 |
+
"grad_norm": 1.1990822553634644,
|
| 43 |
+
"learning_rate": 4.998755e-05,
|
| 44 |
+
"loss": 2.3417,
|
| 45 |
+
"step": 250
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.0003,
|
| 49 |
+
"grad_norm": 3.0459673404693604,
|
| 50 |
+
"learning_rate": 4.998505e-05,
|
| 51 |
+
"loss": 2.2581,
|
| 52 |
+
"step": 300
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.00035,
|
| 56 |
+
"grad_norm": 1.2380167245864868,
|
| 57 |
+
"learning_rate": 4.998255e-05,
|
| 58 |
+
"loss": 2.1563,
|
| 59 |
+
"step": 350
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.0004,
|
| 63 |
+
"grad_norm": 1.2060939073562622,
|
| 64 |
+
"learning_rate": 4.998005e-05,
|
| 65 |
+
"loss": 1.8521,
|
| 66 |
+
"step": 400
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.00045,
|
| 70 |
+
"grad_norm": 0.9207198619842529,
|
| 71 |
+
"learning_rate": 4.9977550000000004e-05,
|
| 72 |
+
"loss": 1.9201,
|
| 73 |
+
"step": 450
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.0005,
|
| 77 |
+
"grad_norm": 1.0762739181518555,
|
| 78 |
+
"learning_rate": 4.997505e-05,
|
| 79 |
+
"loss": 1.8104,
|
| 80 |
+
"step": 500
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.00055,
|
| 84 |
+
"grad_norm": 1.0367337465286255,
|
| 85 |
+
"learning_rate": 4.997255e-05,
|
| 86 |
+
"loss": 1.8756,
|
| 87 |
+
"step": 550
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.0006,
|
| 91 |
+
"grad_norm": 0.9291325211524963,
|
| 92 |
+
"learning_rate": 4.997005e-05,
|
| 93 |
+
"loss": 1.7404,
|
| 94 |
+
"step": 600
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.00065,
|
| 98 |
+
"grad_norm": 0.7452297210693359,
|
| 99 |
+
"learning_rate": 4.996755e-05,
|
| 100 |
+
"loss": 1.6843,
|
| 101 |
+
"step": 650
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.0007,
|
| 105 |
+
"grad_norm": 0.9317042231559753,
|
| 106 |
+
"learning_rate": 4.9965050000000004e-05,
|
| 107 |
+
"loss": 1.7006,
|
| 108 |
+
"step": 700
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.00075,
|
| 112 |
+
"grad_norm": 1.0354136228561401,
|
| 113 |
+
"learning_rate": 4.9962550000000005e-05,
|
| 114 |
+
"loss": 1.5652,
|
| 115 |
+
"step": 750
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.0008,
|
| 119 |
+
"grad_norm": 1.050106406211853,
|
| 120 |
+
"learning_rate": 4.996005e-05,
|
| 121 |
+
"loss": 1.5966,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.00085,
|
| 126 |
+
"grad_norm": 0.9646019339561462,
|
| 127 |
+
"learning_rate": 4.995755e-05,
|
| 128 |
+
"loss": 1.6297,
|
| 129 |
+
"step": 850
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0009,
|
| 133 |
+
"grad_norm": 0.9152287840843201,
|
| 134 |
+
"learning_rate": 4.995505e-05,
|
| 135 |
+
"loss": 1.6328,
|
| 136 |
+
"step": 900
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.00095,
|
| 140 |
+
"grad_norm": 0.9403690099716187,
|
| 141 |
+
"learning_rate": 4.995255e-05,
|
| 142 |
+
"loss": 1.6056,
|
| 143 |
+
"step": 950
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.001,
|
| 147 |
+
"grad_norm": 1.1822874546051025,
|
| 148 |
+
"learning_rate": 4.9950050000000005e-05,
|
| 149 |
+
"loss": 1.6016,
|
| 150 |
+
"step": 1000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.00105,
|
| 154 |
+
"grad_norm": 1.3227542638778687,
|
| 155 |
+
"learning_rate": 4.9947550000000006e-05,
|
| 156 |
+
"loss": 1.5448,
|
| 157 |
+
"step": 1050
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.0011,
|
| 161 |
+
"grad_norm": 1.0503350496292114,
|
| 162 |
+
"learning_rate": 4.994505000000001e-05,
|
| 163 |
+
"loss": 1.6355,
|
| 164 |
+
"step": 1100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.00115,
|
| 168 |
+
"grad_norm": 1.1647204160690308,
|
| 169 |
+
"learning_rate": 4.994255e-05,
|
| 170 |
+
"loss": 1.4673,
|
| 171 |
+
"step": 1150
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.0012,
|
| 175 |
+
"grad_norm": 0.7281339168548584,
|
| 176 |
+
"learning_rate": 4.994005e-05,
|
| 177 |
+
"loss": 1.4641,
|
| 178 |
+
"step": 1200
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.00125,
|
| 182 |
+
"grad_norm": 1.2438446283340454,
|
| 183 |
+
"learning_rate": 4.9937550000000004e-05,
|
| 184 |
+
"loss": 1.5177,
|
| 185 |
+
"step": 1250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.0013,
|
| 189 |
+
"grad_norm": 0.7967873811721802,
|
| 190 |
+
"learning_rate": 4.993505e-05,
|
| 191 |
+
"loss": 1.4399,
|
| 192 |
+
"step": 1300
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.00135,
|
| 196 |
+
"grad_norm": 0.8938255310058594,
|
| 197 |
+
"learning_rate": 4.993255e-05,
|
| 198 |
+
"loss": 1.4453,
|
| 199 |
+
"step": 1350
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.0014,
|
| 203 |
+
"grad_norm": 1.4659217596054077,
|
| 204 |
+
"learning_rate": 4.993005e-05,
|
| 205 |
+
"loss": 1.6517,
|
| 206 |
+
"step": 1400
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.00145,
|
| 210 |
+
"grad_norm": 0.7856793403625488,
|
| 211 |
+
"learning_rate": 4.992755e-05,
|
| 212 |
+
"loss": 1.4721,
|
| 213 |
+
"step": 1450
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.0015,
|
| 217 |
+
"grad_norm": 0.6772142648696899,
|
| 218 |
+
"learning_rate": 4.9925050000000004e-05,
|
| 219 |
+
"loss": 1.5314,
|
| 220 |
+
"step": 1500
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.00155,
|
| 224 |
+
"grad_norm": 0.7613831758499146,
|
| 225 |
+
"learning_rate": 4.992255e-05,
|
| 226 |
+
"loss": 1.4603,
|
| 227 |
+
"step": 1550
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0016,
|
| 231 |
+
"grad_norm": 1.1216529607772827,
|
| 232 |
+
"learning_rate": 4.992005e-05,
|
| 233 |
+
"loss": 1.54,
|
| 234 |
+
"step": 1600
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.00165,
|
| 238 |
+
"grad_norm": 0.7359323501586914,
|
| 239 |
+
"learning_rate": 4.991755e-05,
|
| 240 |
+
"loss": 1.5121,
|
| 241 |
+
"step": 1650
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0017,
|
| 245 |
+
"grad_norm": 0.8343626260757446,
|
| 246 |
+
"learning_rate": 4.991505e-05,
|
| 247 |
+
"loss": 1.6107,
|
| 248 |
+
"step": 1700
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.00175,
|
| 252 |
+
"grad_norm": 0.9381140470504761,
|
| 253 |
+
"learning_rate": 4.9912550000000004e-05,
|
| 254 |
+
"loss": 1.5059,
|
| 255 |
+
"step": 1750
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.0018,
|
| 259 |
+
"grad_norm": 0.7317638993263245,
|
| 260 |
+
"learning_rate": 4.9910050000000005e-05,
|
| 261 |
+
"loss": 1.5621,
|
| 262 |
+
"step": 1800
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.00185,
|
| 266 |
+
"grad_norm": 0.8273525834083557,
|
| 267 |
+
"learning_rate": 4.990755e-05,
|
| 268 |
+
"loss": 1.4737,
|
| 269 |
+
"step": 1850
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.0019,
|
| 273 |
+
"grad_norm": 0.7206686735153198,
|
| 274 |
+
"learning_rate": 4.990505e-05,
|
| 275 |
+
"loss": 1.5242,
|
| 276 |
+
"step": 1900
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.00195,
|
| 280 |
+
"grad_norm": 0.9094993472099304,
|
| 281 |
+
"learning_rate": 4.990255e-05,
|
| 282 |
+
"loss": 1.5336,
|
| 283 |
+
"step": 1950
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.002,
|
| 287 |
+
"grad_norm": 0.7432169914245605,
|
| 288 |
+
"learning_rate": 4.9900050000000003e-05,
|
| 289 |
+
"loss": 1.4913,
|
| 290 |
+
"step": 2000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.00205,
|
| 294 |
+
"grad_norm": 1.337276816368103,
|
| 295 |
+
"learning_rate": 4.9897550000000005e-05,
|
| 296 |
+
"loss": 1.5643,
|
| 297 |
+
"step": 2050
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.0021,
|
| 301 |
+
"grad_norm": 0.6664676666259766,
|
| 302 |
+
"learning_rate": 4.9895050000000006e-05,
|
| 303 |
+
"loss": 1.4728,
|
| 304 |
+
"step": 2100
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.00215,
|
| 308 |
+
"grad_norm": 0.7946372628211975,
|
| 309 |
+
"learning_rate": 4.989255000000001e-05,
|
| 310 |
+
"loss": 1.4832,
|
| 311 |
+
"step": 2150
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.0022,
|
| 315 |
+
"grad_norm": 0.9259161949157715,
|
| 316 |
+
"learning_rate": 4.989005e-05,
|
| 317 |
+
"loss": 1.5674,
|
| 318 |
+
"step": 2200
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.00225,
|
| 322 |
+
"grad_norm": 0.9703247547149658,
|
| 323 |
+
"learning_rate": 4.988755e-05,
|
| 324 |
+
"loss": 1.3973,
|
| 325 |
+
"step": 2250
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.0023,
|
| 329 |
+
"grad_norm": 0.7390087842941284,
|
| 330 |
+
"learning_rate": 4.988505e-05,
|
| 331 |
+
"loss": 1.3919,
|
| 332 |
+
"step": 2300
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.00235,
|
| 336 |
+
"grad_norm": 0.768277645111084,
|
| 337 |
+
"learning_rate": 4.988255e-05,
|
| 338 |
+
"loss": 1.4815,
|
| 339 |
+
"step": 2350
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.0024,
|
| 343 |
+
"grad_norm": 0.76382976770401,
|
| 344 |
+
"learning_rate": 4.988005e-05,
|
| 345 |
+
"loss": 1.4458,
|
| 346 |
+
"step": 2400
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.00245,
|
| 350 |
+
"grad_norm": 0.7780851721763611,
|
| 351 |
+
"learning_rate": 4.987755e-05,
|
| 352 |
+
"loss": 1.4382,
|
| 353 |
+
"step": 2450
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.0025,
|
| 357 |
+
"grad_norm": 0.7184464335441589,
|
| 358 |
+
"learning_rate": 4.987505e-05,
|
| 359 |
+
"loss": 1.521,
|
| 360 |
+
"step": 2500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.00255,
|
| 364 |
+
"grad_norm": 0.6212390065193176,
|
| 365 |
+
"learning_rate": 4.9872550000000004e-05,
|
| 366 |
+
"loss": 1.3889,
|
| 367 |
+
"step": 2550
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.0026,
|
| 371 |
+
"grad_norm": 0.8539580702781677,
|
| 372 |
+
"learning_rate": 4.987005e-05,
|
| 373 |
+
"loss": 1.415,
|
| 374 |
+
"step": 2600
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.00265,
|
| 378 |
+
"grad_norm": 0.7129775881767273,
|
| 379 |
+
"learning_rate": 4.986755e-05,
|
| 380 |
+
"loss": 1.4102,
|
| 381 |
+
"step": 2650
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.0027,
|
| 385 |
+
"grad_norm": 0.5899195671081543,
|
| 386 |
+
"learning_rate": 4.986505e-05,
|
| 387 |
+
"loss": 1.3074,
|
| 388 |
+
"step": 2700
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.00275,
|
| 392 |
+
"grad_norm": 0.6940101981163025,
|
| 393 |
+
"learning_rate": 4.986255e-05,
|
| 394 |
+
"loss": 1.3748,
|
| 395 |
+
"step": 2750
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.0028,
|
| 399 |
+
"grad_norm": 0.6420891880989075,
|
| 400 |
+
"learning_rate": 4.9860050000000004e-05,
|
| 401 |
+
"loss": 1.4089,
|
| 402 |
+
"step": 2800
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.00285,
|
| 406 |
+
"grad_norm": 0.8561428189277649,
|
| 407 |
+
"learning_rate": 4.9857550000000005e-05,
|
| 408 |
+
"loss": 1.4052,
|
| 409 |
+
"step": 2850
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.0029,
|
| 413 |
+
"grad_norm": 0.6900970935821533,
|
| 414 |
+
"learning_rate": 4.9855050000000006e-05,
|
| 415 |
+
"loss": 1.4323,
|
| 416 |
+
"step": 2900
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.00295,
|
| 420 |
+
"grad_norm": 0.8071371912956238,
|
| 421 |
+
"learning_rate": 4.985255e-05,
|
| 422 |
+
"loss": 1.3485,
|
| 423 |
+
"step": 2950
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.003,
|
| 427 |
+
"grad_norm": 0.6493249535560608,
|
| 428 |
+
"learning_rate": 4.985005e-05,
|
| 429 |
+
"loss": 1.4321,
|
| 430 |
+
"step": 3000
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.00305,
|
| 434 |
+
"grad_norm": 0.8712514042854309,
|
| 435 |
+
"learning_rate": 4.9847550000000004e-05,
|
| 436 |
+
"loss": 1.406,
|
| 437 |
+
"step": 3050
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.0031,
|
| 441 |
+
"grad_norm": 0.8464570045471191,
|
| 442 |
+
"learning_rate": 4.9845050000000005e-05,
|
| 443 |
+
"loss": 1.3322,
|
| 444 |
+
"step": 3100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.00315,
|
| 448 |
+
"grad_norm": 0.7779001593589783,
|
| 449 |
+
"learning_rate": 4.9842550000000006e-05,
|
| 450 |
+
"loss": 1.3905,
|
| 451 |
+
"step": 3150
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.0032,
|
| 455 |
+
"grad_norm": 0.8314748406410217,
|
| 456 |
+
"learning_rate": 4.984005000000001e-05,
|
| 457 |
+
"loss": 1.4134,
|
| 458 |
+
"step": 3200
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.00325,
|
| 462 |
+
"grad_norm": 0.7180835604667664,
|
| 463 |
+
"learning_rate": 4.983755e-05,
|
| 464 |
+
"loss": 1.3629,
|
| 465 |
+
"step": 3250
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.0033,
|
| 469 |
+
"grad_norm": 0.6165933012962341,
|
| 470 |
+
"learning_rate": 4.9835049999999996e-05,
|
| 471 |
+
"loss": 1.3833,
|
| 472 |
+
"step": 3300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.00335,
|
| 476 |
+
"grad_norm": 1.8662049770355225,
|
| 477 |
+
"learning_rate": 4.983255e-05,
|
| 478 |
+
"loss": 1.3557,
|
| 479 |
+
"step": 3350
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.0034,
|
| 483 |
+
"grad_norm": 0.9091550707817078,
|
| 484 |
+
"learning_rate": 4.983005e-05,
|
| 485 |
+
"loss": 1.4092,
|
| 486 |
+
"step": 3400
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.00345,
|
| 490 |
+
"grad_norm": 0.9908722639083862,
|
| 491 |
+
"learning_rate": 4.982755e-05,
|
| 492 |
+
"loss": 1.3465,
|
| 493 |
+
"step": 3450
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.0035,
|
| 497 |
+
"grad_norm": 0.862427830696106,
|
| 498 |
+
"learning_rate": 4.982505e-05,
|
| 499 |
+
"loss": 1.3143,
|
| 500 |
+
"step": 3500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.00355,
|
| 504 |
+
"grad_norm": 0.755211591720581,
|
| 505 |
+
"learning_rate": 4.982255e-05,
|
| 506 |
+
"loss": 1.358,
|
| 507 |
+
"step": 3550
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.0036,
|
| 511 |
+
"grad_norm": 0.7872670888900757,
|
| 512 |
+
"learning_rate": 4.9820050000000004e-05,
|
| 513 |
+
"loss": 1.3952,
|
| 514 |
+
"step": 3600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.00365,
|
| 518 |
+
"grad_norm": 1.5124619007110596,
|
| 519 |
+
"learning_rate": 4.981755e-05,
|
| 520 |
+
"loss": 1.2929,
|
| 521 |
+
"step": 3650
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.0037,
|
| 525 |
+
"grad_norm": 0.7712079286575317,
|
| 526 |
+
"learning_rate": 4.981505e-05,
|
| 527 |
+
"loss": 1.3429,
|
| 528 |
+
"step": 3700
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.00375,
|
| 532 |
+
"grad_norm": 0.7001494765281677,
|
| 533 |
+
"learning_rate": 4.981255e-05,
|
| 534 |
+
"loss": 1.272,
|
| 535 |
+
"step": 3750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.0038,
|
| 539 |
+
"grad_norm": 0.674104630947113,
|
| 540 |
+
"learning_rate": 4.981005e-05,
|
| 541 |
+
"loss": 1.3891,
|
| 542 |
+
"step": 3800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.00385,
|
| 546 |
+
"grad_norm": 1.0494478940963745,
|
| 547 |
+
"learning_rate": 4.9807550000000004e-05,
|
| 548 |
+
"loss": 1.333,
|
| 549 |
+
"step": 3850
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.0039,
|
| 553 |
+
"grad_norm": 0.6674365401268005,
|
| 554 |
+
"learning_rate": 4.9805050000000005e-05,
|
| 555 |
+
"loss": 1.4418,
|
| 556 |
+
"step": 3900
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.00395,
|
| 560 |
+
"grad_norm": 0.7624682784080505,
|
| 561 |
+
"learning_rate": 4.9802550000000007e-05,
|
| 562 |
+
"loss": 1.2837,
|
| 563 |
+
"step": 3950
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.004,
|
| 567 |
+
"grad_norm": 1.5128511190414429,
|
| 568 |
+
"learning_rate": 4.980005e-05,
|
| 569 |
+
"loss": 1.2609,
|
| 570 |
+
"step": 4000
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.00405,
|
| 574 |
+
"grad_norm": 0.5422250628471375,
|
| 575 |
+
"learning_rate": 4.979755e-05,
|
| 576 |
+
"loss": 1.3253,
|
| 577 |
+
"step": 4050
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.0041,
|
| 581 |
+
"grad_norm": 0.63419508934021,
|
| 582 |
+
"learning_rate": 4.9795050000000004e-05,
|
| 583 |
+
"loss": 1.3369,
|
| 584 |
+
"step": 4100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.00415,
|
| 588 |
+
"grad_norm": 0.7608025670051575,
|
| 589 |
+
"learning_rate": 4.9792550000000005e-05,
|
| 590 |
+
"loss": 1.3186,
|
| 591 |
+
"step": 4150
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.0042,
|
| 595 |
+
"grad_norm": 0.6282067894935608,
|
| 596 |
+
"learning_rate": 4.9790050000000006e-05,
|
| 597 |
+
"loss": 1.3755,
|
| 598 |
+
"step": 4200
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.00425,
|
| 602 |
+
"grad_norm": 0.7882742881774902,
|
| 603 |
+
"learning_rate": 4.978755000000001e-05,
|
| 604 |
+
"loss": 1.3326,
|
| 605 |
+
"step": 4250
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.0043,
|
| 609 |
+
"grad_norm": 0.6797521710395813,
|
| 610 |
+
"learning_rate": 4.978505e-05,
|
| 611 |
+
"loss": 1.2774,
|
| 612 |
+
"step": 4300
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.00435,
|
| 616 |
+
"grad_norm": 0.6129476428031921,
|
| 617 |
+
"learning_rate": 4.978255e-05,
|
| 618 |
+
"loss": 1.324,
|
| 619 |
+
"step": 4350
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.0044,
|
| 623 |
+
"grad_norm": 0.7544896006584167,
|
| 624 |
+
"learning_rate": 4.978005e-05,
|
| 625 |
+
"loss": 1.299,
|
| 626 |
+
"step": 4400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.00445,
|
| 630 |
+
"grad_norm": 0.6218116283416748,
|
| 631 |
+
"learning_rate": 4.977755e-05,
|
| 632 |
+
"loss": 1.3084,
|
| 633 |
+
"step": 4450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.0045,
|
| 637 |
+
"grad_norm": 0.5591554045677185,
|
| 638 |
+
"learning_rate": 4.977505e-05,
|
| 639 |
+
"loss": 1.3001,
|
| 640 |
+
"step": 4500
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.00455,
|
| 644 |
+
"grad_norm": 0.854743242263794,
|
| 645 |
+
"learning_rate": 4.977255e-05,
|
| 646 |
+
"loss": 1.3357,
|
| 647 |
+
"step": 4550
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.0046,
|
| 651 |
+
"grad_norm": 0.7265657186508179,
|
| 652 |
+
"learning_rate": 4.977005e-05,
|
| 653 |
+
"loss": 1.3362,
|
| 654 |
+
"step": 4600
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.00465,
|
| 658 |
+
"grad_norm": 0.8089432120323181,
|
| 659 |
+
"learning_rate": 4.9767550000000004e-05,
|
| 660 |
+
"loss": 1.3087,
|
| 661 |
+
"step": 4650
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.0047,
|
| 665 |
+
"grad_norm": 0.8999931216239929,
|
| 666 |
+
"learning_rate": 4.976505e-05,
|
| 667 |
+
"loss": 1.3894,
|
| 668 |
+
"step": 4700
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.00475,
|
| 672 |
+
"grad_norm": 0.7705693244934082,
|
| 673 |
+
"learning_rate": 4.976255e-05,
|
| 674 |
+
"loss": 1.2585,
|
| 675 |
+
"step": 4750
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.0048,
|
| 679 |
+
"grad_norm": 0.7201912999153137,
|
| 680 |
+
"learning_rate": 4.976005e-05,
|
| 681 |
+
"loss": 1.221,
|
| 682 |
+
"step": 4800
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.00485,
|
| 686 |
+
"grad_norm": 0.6199264526367188,
|
| 687 |
+
"learning_rate": 4.975755e-05,
|
| 688 |
+
"loss": 1.2878,
|
| 689 |
+
"step": 4850
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.0049,
|
| 693 |
+
"grad_norm": 0.6883070468902588,
|
| 694 |
+
"learning_rate": 4.9755050000000004e-05,
|
| 695 |
+
"loss": 1.3032,
|
| 696 |
+
"step": 4900
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.00495,
|
| 700 |
+
"grad_norm": 0.524889349937439,
|
| 701 |
+
"learning_rate": 4.9752550000000005e-05,
|
| 702 |
+
"loss": 1.2706,
|
| 703 |
+
"step": 4950
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.005,
|
| 707 |
+
"grad_norm": 0.7067034244537354,
|
| 708 |
+
"learning_rate": 4.9750050000000007e-05,
|
| 709 |
+
"loss": 1.2424,
|
| 710 |
+
"step": 5000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.00505,
|
| 714 |
+
"grad_norm": 0.7097471356391907,
|
| 715 |
+
"learning_rate": 4.974755e-05,
|
| 716 |
+
"loss": 1.3175,
|
| 717 |
+
"step": 5050
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.0051,
|
| 721 |
+
"grad_norm": 0.6991677284240723,
|
| 722 |
+
"learning_rate": 4.974505e-05,
|
| 723 |
+
"loss": 1.2983,
|
| 724 |
+
"step": 5100
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.00515,
|
| 728 |
+
"grad_norm": 1.4025177955627441,
|
| 729 |
+
"learning_rate": 4.9742550000000004e-05,
|
| 730 |
+
"loss": 1.2695,
|
| 731 |
+
"step": 5150
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.0052,
|
| 735 |
+
"grad_norm": 0.6075018048286438,
|
| 736 |
+
"learning_rate": 4.9740050000000005e-05,
|
| 737 |
+
"loss": 1.2973,
|
| 738 |
+
"step": 5200
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.00525,
|
| 742 |
+
"grad_norm": 0.5344257950782776,
|
| 743 |
+
"learning_rate": 4.9737550000000006e-05,
|
| 744 |
+
"loss": 1.2353,
|
| 745 |
+
"step": 5250
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.0053,
|
| 749 |
+
"grad_norm": 0.5963959693908691,
|
| 750 |
+
"learning_rate": 4.973505e-05,
|
| 751 |
+
"loss": 1.2359,
|
| 752 |
+
"step": 5300
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.00535,
|
| 756 |
+
"grad_norm": 0.9493517279624939,
|
| 757 |
+
"learning_rate": 4.973255e-05,
|
| 758 |
+
"loss": 1.252,
|
| 759 |
+
"step": 5350
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.0054,
|
| 763 |
+
"grad_norm": 0.6466087102890015,
|
| 764 |
+
"learning_rate": 4.9730050000000003e-05,
|
| 765 |
+
"loss": 1.2012,
|
| 766 |
+
"step": 5400
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.00545,
|
| 770 |
+
"grad_norm": 0.6407445669174194,
|
| 771 |
+
"learning_rate": 4.972755e-05,
|
| 772 |
+
"loss": 1.2615,
|
| 773 |
+
"step": 5450
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.0055,
|
| 777 |
+
"grad_norm": 0.617421567440033,
|
| 778 |
+
"learning_rate": 4.972505e-05,
|
| 779 |
+
"loss": 1.2718,
|
| 780 |
+
"step": 5500
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.00555,
|
| 784 |
+
"grad_norm": 0.5649735331535339,
|
| 785 |
+
"learning_rate": 4.972255e-05,
|
| 786 |
+
"loss": 1.2439,
|
| 787 |
+
"step": 5550
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.0056,
|
| 791 |
+
"grad_norm": 0.675645649433136,
|
| 792 |
+
"learning_rate": 4.972005e-05,
|
| 793 |
+
"loss": 1.2847,
|
| 794 |
+
"step": 5600
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.00565,
|
| 798 |
+
"grad_norm": 0.6887643337249756,
|
| 799 |
+
"learning_rate": 4.971755e-05,
|
| 800 |
+
"loss": 1.2464,
|
| 801 |
+
"step": 5650
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.0057,
|
| 805 |
+
"grad_norm": 0.6572535037994385,
|
| 806 |
+
"learning_rate": 4.9715050000000004e-05,
|
| 807 |
+
"loss": 1.2517,
|
| 808 |
+
"step": 5700
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.00575,
|
| 812 |
+
"grad_norm": 0.6076003313064575,
|
| 813 |
+
"learning_rate": 4.971255e-05,
|
| 814 |
+
"loss": 1.2183,
|
| 815 |
+
"step": 5750
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.0058,
|
| 819 |
+
"grad_norm": 0.6134760975837708,
|
| 820 |
+
"learning_rate": 4.971005e-05,
|
| 821 |
+
"loss": 1.2303,
|
| 822 |
+
"step": 5800
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.00585,
|
| 826 |
+
"grad_norm": 0.6791936159133911,
|
| 827 |
+
"learning_rate": 4.970755e-05,
|
| 828 |
+
"loss": 1.2351,
|
| 829 |
+
"step": 5850
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.0059,
|
| 833 |
+
"grad_norm": 0.6307588219642639,
|
| 834 |
+
"learning_rate": 4.970505e-05,
|
| 835 |
+
"loss": 1.254,
|
| 836 |
+
"step": 5900
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.00595,
|
| 840 |
+
"grad_norm": 1.1272459030151367,
|
| 841 |
+
"learning_rate": 4.9702550000000004e-05,
|
| 842 |
+
"loss": 1.265,
|
| 843 |
+
"step": 5950
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.006,
|
| 847 |
+
"grad_norm": 0.5415930151939392,
|
| 848 |
+
"learning_rate": 4.9700050000000005e-05,
|
| 849 |
+
"loss": 1.2068,
|
| 850 |
+
"step": 6000
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.00605,
|
| 854 |
+
"grad_norm": 0.5450507402420044,
|
| 855 |
+
"learning_rate": 4.969755000000001e-05,
|
| 856 |
+
"loss": 1.2408,
|
| 857 |
+
"step": 6050
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.0061,
|
| 861 |
+
"grad_norm": 0.6528737545013428,
|
| 862 |
+
"learning_rate": 4.969505e-05,
|
| 863 |
+
"loss": 1.2035,
|
| 864 |
+
"step": 6100
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.00615,
|
| 868 |
+
"grad_norm": 0.6139540672302246,
|
| 869 |
+
"learning_rate": 4.969255e-05,
|
| 870 |
+
"loss": 1.2077,
|
| 871 |
+
"step": 6150
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.0062,
|
| 875 |
+
"grad_norm": 0.6503865718841553,
|
| 876 |
+
"learning_rate": 4.9690050000000004e-05,
|
| 877 |
+
"loss": 1.2055,
|
| 878 |
+
"step": 6200
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.00625,
|
| 882 |
+
"grad_norm": 0.6888736486434937,
|
| 883 |
+
"learning_rate": 4.9687550000000005e-05,
|
| 884 |
+
"loss": 1.2161,
|
| 885 |
+
"step": 6250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.0063,
|
| 889 |
+
"grad_norm": 1.1457703113555908,
|
| 890 |
+
"learning_rate": 4.968505e-05,
|
| 891 |
+
"loss": 1.1939,
|
| 892 |
+
"step": 6300
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.00635,
|
| 896 |
+
"grad_norm": 0.5861665606498718,
|
| 897 |
+
"learning_rate": 4.968255e-05,
|
| 898 |
+
"loss": 1.2177,
|
| 899 |
+
"step": 6350
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.0064,
|
| 903 |
+
"grad_norm": 0.8666422963142395,
|
| 904 |
+
"learning_rate": 4.968005e-05,
|
| 905 |
+
"loss": 1.1945,
|
| 906 |
+
"step": 6400
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.00645,
|
| 910 |
+
"grad_norm": 0.6094856858253479,
|
| 911 |
+
"learning_rate": 4.9677550000000003e-05,
|
| 912 |
+
"loss": 1.2394,
|
| 913 |
+
"step": 6450
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.0065,
|
| 917 |
+
"grad_norm": 0.6510404348373413,
|
| 918 |
+
"learning_rate": 4.967505e-05,
|
| 919 |
+
"loss": 1.1764,
|
| 920 |
+
"step": 6500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.00655,
|
| 924 |
+
"grad_norm": 0.7044185996055603,
|
| 925 |
+
"learning_rate": 4.967255e-05,
|
| 926 |
+
"loss": 1.1951,
|
| 927 |
+
"step": 6550
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.0066,
|
| 931 |
+
"grad_norm": 0.6577385663986206,
|
| 932 |
+
"learning_rate": 4.967005e-05,
|
| 933 |
+
"loss": 1.2186,
|
| 934 |
+
"step": 6600
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.00665,
|
| 938 |
+
"grad_norm": 0.6468052268028259,
|
| 939 |
+
"learning_rate": 4.966755e-05,
|
| 940 |
+
"loss": 1.2491,
|
| 941 |
+
"step": 6650
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.0067,
|
| 945 |
+
"grad_norm": 0.6153103709220886,
|
| 946 |
+
"learning_rate": 4.966505e-05,
|
| 947 |
+
"loss": 1.25,
|
| 948 |
+
"step": 6700
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.00675,
|
| 952 |
+
"grad_norm": 0.6066887974739075,
|
| 953 |
+
"learning_rate": 4.9662550000000004e-05,
|
| 954 |
+
"loss": 1.1852,
|
| 955 |
+
"step": 6750
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.0068,
|
| 959 |
+
"grad_norm": 0.5735403895378113,
|
| 960 |
+
"learning_rate": 4.9660050000000006e-05,
|
| 961 |
+
"loss": 1.2139,
|
| 962 |
+
"step": 6800
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.00685,
|
| 966 |
+
"grad_norm": 0.6247462630271912,
|
| 967 |
+
"learning_rate": 4.965755e-05,
|
| 968 |
+
"loss": 1.1597,
|
| 969 |
+
"step": 6850
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.0069,
|
| 973 |
+
"grad_norm": 0.6690845489501953,
|
| 974 |
+
"learning_rate": 4.965505e-05,
|
| 975 |
+
"loss": 1.2095,
|
| 976 |
+
"step": 6900
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.00695,
|
| 980 |
+
"grad_norm": 0.5683197975158691,
|
| 981 |
+
"learning_rate": 4.965255e-05,
|
| 982 |
+
"loss": 1.3077,
|
| 983 |
+
"step": 6950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.007,
|
| 987 |
+
"grad_norm": 0.6314153671264648,
|
| 988 |
+
"learning_rate": 4.9650050000000004e-05,
|
| 989 |
+
"loss": 1.2556,
|
| 990 |
+
"step": 7000
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.00705,
|
| 994 |
+
"grad_norm": 0.641153872013092,
|
| 995 |
+
"learning_rate": 4.9647550000000005e-05,
|
| 996 |
+
"loss": 1.1643,
|
| 997 |
+
"step": 7050
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.0071,
|
| 1001 |
+
"grad_norm": 0.6113337278366089,
|
| 1002 |
+
"learning_rate": 4.964505000000001e-05,
|
| 1003 |
+
"loss": 1.1688,
|
| 1004 |
+
"step": 7100
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.00715,
|
| 1008 |
+
"grad_norm": 0.5102031230926514,
|
| 1009 |
+
"learning_rate": 4.964255e-05,
|
| 1010 |
+
"loss": 1.1835,
|
| 1011 |
+
"step": 7150
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.0072,
|
| 1015 |
+
"grad_norm": 0.6926486492156982,
|
| 1016 |
+
"learning_rate": 4.964005e-05,
|
| 1017 |
+
"loss": 1.1704,
|
| 1018 |
+
"step": 7200
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.00725,
|
| 1022 |
+
"grad_norm": 0.594326376914978,
|
| 1023 |
+
"learning_rate": 4.9637550000000004e-05,
|
| 1024 |
+
"loss": 1.1434,
|
| 1025 |
+
"step": 7250
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.0073,
|
| 1029 |
+
"grad_norm": 0.5609344840049744,
|
| 1030 |
+
"learning_rate": 4.963505e-05,
|
| 1031 |
+
"loss": 1.2389,
|
| 1032 |
+
"step": 7300
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.00735,
|
| 1036 |
+
"grad_norm": 0.6464629173278809,
|
| 1037 |
+
"learning_rate": 4.963255e-05,
|
| 1038 |
+
"loss": 1.2097,
|
| 1039 |
+
"step": 7350
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.0074,
|
| 1043 |
+
"grad_norm": 0.5782963633537292,
|
| 1044 |
+
"learning_rate": 4.963005e-05,
|
| 1045 |
+
"loss": 1.2607,
|
| 1046 |
+
"step": 7400
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.00745,
|
| 1050 |
+
"grad_norm": 0.7301307320594788,
|
| 1051 |
+
"learning_rate": 4.962755e-05,
|
| 1052 |
+
"loss": 1.1759,
|
| 1053 |
+
"step": 7450
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.0075,
|
| 1057 |
+
"grad_norm": 0.5845964550971985,
|
| 1058 |
+
"learning_rate": 4.9625050000000004e-05,
|
| 1059 |
+
"loss": 1.1806,
|
| 1060 |
+
"step": 7500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.00755,
|
| 1064 |
+
"grad_norm": 0.618295431137085,
|
| 1065 |
+
"learning_rate": 4.962255e-05,
|
| 1066 |
+
"loss": 1.1339,
|
| 1067 |
+
"step": 7550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.0076,
|
| 1071 |
+
"grad_norm": 0.5799853801727295,
|
| 1072 |
+
"learning_rate": 4.962005e-05,
|
| 1073 |
+
"loss": 1.1641,
|
| 1074 |
+
"step": 7600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.00765,
|
| 1078 |
+
"grad_norm": 0.5227323174476624,
|
| 1079 |
+
"learning_rate": 4.961755e-05,
|
| 1080 |
+
"loss": 1.2375,
|
| 1081 |
+
"step": 7650
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.0077,
|
| 1085 |
+
"grad_norm": 0.699111819267273,
|
| 1086 |
+
"learning_rate": 4.961505e-05,
|
| 1087 |
+
"loss": 1.0943,
|
| 1088 |
+
"step": 7700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.00775,
|
| 1092 |
+
"grad_norm": 0.5230436325073242,
|
| 1093 |
+
"learning_rate": 4.961255e-05,
|
| 1094 |
+
"loss": 1.1762,
|
| 1095 |
+
"step": 7750
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.0078,
|
| 1099 |
+
"grad_norm": 0.5776082873344421,
|
| 1100 |
+
"learning_rate": 4.9610050000000005e-05,
|
| 1101 |
+
"loss": 1.18,
|
| 1102 |
+
"step": 7800
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.00785,
|
| 1106 |
+
"grad_norm": 0.584697425365448,
|
| 1107 |
+
"learning_rate": 4.9607550000000006e-05,
|
| 1108 |
+
"loss": 1.2878,
|
| 1109 |
+
"step": 7850
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.0079,
|
| 1113 |
+
"grad_norm": 0.66282057762146,
|
| 1114 |
+
"learning_rate": 4.960505e-05,
|
| 1115 |
+
"loss": 1.219,
|
| 1116 |
+
"step": 7900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.00795,
|
| 1120 |
+
"grad_norm": 1.2405346632003784,
|
| 1121 |
+
"learning_rate": 4.960255e-05,
|
| 1122 |
+
"loss": 1.0984,
|
| 1123 |
+
"step": 7950
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.008,
|
| 1127 |
+
"grad_norm": 0.6882185339927673,
|
| 1128 |
+
"learning_rate": 4.960005e-05,
|
| 1129 |
+
"loss": 1.2291,
|
| 1130 |
+
"step": 8000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.00805,
|
| 1134 |
+
"grad_norm": 0.5376139879226685,
|
| 1135 |
+
"learning_rate": 4.9597550000000004e-05,
|
| 1136 |
+
"loss": 1.1768,
|
| 1137 |
+
"step": 8050
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.0081,
|
| 1141 |
+
"grad_norm": 0.5613966584205627,
|
| 1142 |
+
"learning_rate": 4.9595050000000006e-05,
|
| 1143 |
+
"loss": 1.1882,
|
| 1144 |
+
"step": 8100
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.00815,
|
| 1148 |
+
"grad_norm": 0.5976732969284058,
|
| 1149 |
+
"learning_rate": 4.959255000000001e-05,
|
| 1150 |
+
"loss": 1.1938,
|
| 1151 |
+
"step": 8150
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.0082,
|
| 1155 |
+
"grad_norm": 0.5053662657737732,
|
| 1156 |
+
"learning_rate": 4.959005e-05,
|
| 1157 |
+
"loss": 1.1335,
|
| 1158 |
+
"step": 8200
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.00825,
|
| 1162 |
+
"grad_norm": 0.5319002866744995,
|
| 1163 |
+
"learning_rate": 4.958755e-05,
|
| 1164 |
+
"loss": 1.23,
|
| 1165 |
+
"step": 8250
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.0083,
|
| 1169 |
+
"grad_norm": 0.6441113948822021,
|
| 1170 |
+
"learning_rate": 4.9585050000000004e-05,
|
| 1171 |
+
"loss": 1.095,
|
| 1172 |
+
"step": 8300
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.00835,
|
| 1176 |
+
"grad_norm": 0.7779256701469421,
|
| 1177 |
+
"learning_rate": 4.958255e-05,
|
| 1178 |
+
"loss": 1.1637,
|
| 1179 |
+
"step": 8350
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.0084,
|
| 1183 |
+
"grad_norm": 0.6262242794036865,
|
| 1184 |
+
"learning_rate": 4.958005e-05,
|
| 1185 |
+
"loss": 1.1756,
|
| 1186 |
+
"step": 8400
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.00845,
|
| 1190 |
+
"grad_norm": 0.6683831214904785,
|
| 1191 |
+
"learning_rate": 4.957755e-05,
|
| 1192 |
+
"loss": 1.2142,
|
| 1193 |
+
"step": 8450
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.0085,
|
| 1197 |
+
"grad_norm": 0.8390682339668274,
|
| 1198 |
+
"learning_rate": 4.957505e-05,
|
| 1199 |
+
"loss": 1.147,
|
| 1200 |
+
"step": 8500
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.00855,
|
| 1204 |
+
"grad_norm": 0.6620815396308899,
|
| 1205 |
+
"learning_rate": 4.9572550000000004e-05,
|
| 1206 |
+
"loss": 1.1619,
|
| 1207 |
+
"step": 8550
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.0086,
|
| 1211 |
+
"grad_norm": 0.5890603065490723,
|
| 1212 |
+
"learning_rate": 4.957005e-05,
|
| 1213 |
+
"loss": 1.1076,
|
| 1214 |
+
"step": 8600
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.00865,
|
| 1218 |
+
"grad_norm": 0.578610897064209,
|
| 1219 |
+
"learning_rate": 4.956755e-05,
|
| 1220 |
+
"loss": 1.0664,
|
| 1221 |
+
"step": 8650
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.0087,
|
| 1225 |
+
"grad_norm": 0.5939948558807373,
|
| 1226 |
+
"learning_rate": 4.956505e-05,
|
| 1227 |
+
"loss": 1.1613,
|
| 1228 |
+
"step": 8700
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.00875,
|
| 1232 |
+
"grad_norm": 0.6854498386383057,
|
| 1233 |
+
"learning_rate": 4.956255e-05,
|
| 1234 |
+
"loss": 1.218,
|
| 1235 |
+
"step": 8750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.0088,
|
| 1239 |
+
"grad_norm": 0.8202462196350098,
|
| 1240 |
+
"learning_rate": 4.956005e-05,
|
| 1241 |
+
"loss": 1.2239,
|
| 1242 |
+
"step": 8800
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.00885,
|
| 1246 |
+
"grad_norm": 0.5740517973899841,
|
| 1247 |
+
"learning_rate": 4.9557550000000005e-05,
|
| 1248 |
+
"loss": 1.1104,
|
| 1249 |
+
"step": 8850
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.0089,
|
| 1253 |
+
"grad_norm": 0.7225694060325623,
|
| 1254 |
+
"learning_rate": 4.9555050000000006e-05,
|
| 1255 |
+
"loss": 1.1617,
|
| 1256 |
+
"step": 8900
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.00895,
|
| 1260 |
+
"grad_norm": 0.5652306079864502,
|
| 1261 |
+
"learning_rate": 4.955255e-05,
|
| 1262 |
+
"loss": 1.0913,
|
| 1263 |
+
"step": 8950
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.009,
|
| 1267 |
+
"grad_norm": 0.9501689672470093,
|
| 1268 |
+
"learning_rate": 4.955005e-05,
|
| 1269 |
+
"loss": 1.1862,
|
| 1270 |
+
"step": 9000
|
| 1271 |
+
}
|
| 1272 |
+
],
|
| 1273 |
+
"logging_steps": 50,
|
| 1274 |
+
"max_steps": 1000000,
|
| 1275 |
+
"num_input_tokens_seen": 0,
|
| 1276 |
+
"num_train_epochs": 9223372036854775807,
|
| 1277 |
+
"save_steps": 1000,
|
| 1278 |
+
"stateful_callbacks": {
|
| 1279 |
+
"TrainerControl": {
|
| 1280 |
+
"args": {
|
| 1281 |
+
"should_epoch_stop": false,
|
| 1282 |
+
"should_evaluate": false,
|
| 1283 |
+
"should_log": false,
|
| 1284 |
+
"should_save": true,
|
| 1285 |
+
"should_training_stop": false
|
| 1286 |
+
},
|
| 1287 |
+
"attributes": {}
|
| 1288 |
+
}
|
| 1289 |
+
},
|
| 1290 |
+
"total_flos": 1.67761573576704e+17,
|
| 1291 |
+
"train_batch_size": 4,
|
| 1292 |
+
"trial_name": null,
|
| 1293 |
+
"trial_params": null
|
| 1294 |
+
}
|
checkpoint-9000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2f1f06799524e8433f4ea5813803d7bd7b4539e7b44d7846bd777c6941f1e37c
|
| 3 |
+
size 5841
|