Training in progress, step 2500, checkpoint
Browse files- checkpoint-2500/config.json +35 -0
- checkpoint-2500/generation_config.json +9 -0
- checkpoint-2500/model.safetensors +3 -0
- checkpoint-2500/optimizer.pt +3 -0
- checkpoint-2500/rng_state.pth +3 -0
- checkpoint-2500/scheduler.pt +3 -0
- checkpoint-2500/trainer_state.json +1784 -0
- checkpoint-2500/training_args.bin +3 -0
checkpoint-2500/config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_bias": false,
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 128000,
|
| 8 |
+
"dtype": "bfloat16",
|
| 9 |
+
"eos_token_id": 128001,
|
| 10 |
+
"head_dim": 64,
|
| 11 |
+
"hidden_act": "silu",
|
| 12 |
+
"hidden_size": 2048,
|
| 13 |
+
"initializer_range": 0.02,
|
| 14 |
+
"intermediate_size": 8192,
|
| 15 |
+
"max_position_embeddings": 131072,
|
| 16 |
+
"mlp_bias": false,
|
| 17 |
+
"model_type": "llama",
|
| 18 |
+
"num_attention_heads": 32,
|
| 19 |
+
"num_hidden_layers": 16,
|
| 20 |
+
"num_key_value_heads": 8,
|
| 21 |
+
"pretraining_tp": 1,
|
| 22 |
+
"rms_norm_eps": 1e-05,
|
| 23 |
+
"rope_scaling": {
|
| 24 |
+
"factor": 32.0,
|
| 25 |
+
"high_freq_factor": 4.0,
|
| 26 |
+
"low_freq_factor": 1.0,
|
| 27 |
+
"original_max_position_embeddings": 8192,
|
| 28 |
+
"rope_type": "llama3"
|
| 29 |
+
},
|
| 30 |
+
"rope_theta": 500000.0,
|
| 31 |
+
"tie_word_embeddings": true,
|
| 32 |
+
"transformers_version": "4.57.1",
|
| 33 |
+
"use_cache": true,
|
| 34 |
+
"vocab_size": 128260
|
| 35 |
+
}
|
checkpoint-2500/generation_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 128000,
|
| 4 |
+
"do_sample": true,
|
| 5 |
+
"eos_token_id": 128001,
|
| 6 |
+
"temperature": 0.6,
|
| 7 |
+
"top_p": 0.9,
|
| 8 |
+
"transformers_version": "4.57.1"
|
| 9 |
+
}
|
checkpoint-2500/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ec83359a696f1c3266ae7ea2719b9a4c736d4224517faefc3de6e87d0b979387
|
| 3 |
+
size 2471661992
|
checkpoint-2500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5adb260bfd823bbf4e8bb0a462004c87f3a79a22ea1240ff7ef38859faa55660
|
| 3 |
+
size 3725003
|
checkpoint-2500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:22a795ff36907ad78bdca4cb57bf3d653b199e7729e1a57e1668373c18670ec0
|
| 3 |
+
size 14709
|
checkpoint-2500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7452baf421186c7ecb17aca0bed967e1dda02ffe91de0d6aa55e68a1398920e2
|
| 3 |
+
size 1465
|
checkpoint-2500/trainer_state.json
ADDED
|
@@ -0,0 +1,1784 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 1.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.004,
|
| 14 |
+
"grad_norm": 194.0,
|
| 15 |
+
"learning_rate": 7.2e-07,
|
| 16 |
+
"loss": 1.6508,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.008,
|
| 21 |
+
"grad_norm": 166.0,
|
| 22 |
+
"learning_rate": 1.52e-06,
|
| 23 |
+
"loss": 1.6883,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.012,
|
| 28 |
+
"grad_norm": 161.0,
|
| 29 |
+
"learning_rate": 2.3200000000000002e-06,
|
| 30 |
+
"loss": 1.4875,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.016,
|
| 35 |
+
"grad_norm": 219.0,
|
| 36 |
+
"learning_rate": 3.12e-06,
|
| 37 |
+
"loss": 1.6227,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.02,
|
| 42 |
+
"grad_norm": 116.0,
|
| 43 |
+
"learning_rate": 3.920000000000001e-06,
|
| 44 |
+
"loss": 1.4516,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.024,
|
| 49 |
+
"grad_norm": 23.375,
|
| 50 |
+
"learning_rate": 4.7200000000000005e-06,
|
| 51 |
+
"loss": 1.3695,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.028,
|
| 56 |
+
"grad_norm": 8.8125,
|
| 57 |
+
"learning_rate": 5.5200000000000005e-06,
|
| 58 |
+
"loss": 1.1266,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.032,
|
| 63 |
+
"grad_norm": 27.125,
|
| 64 |
+
"learning_rate": 6.3200000000000005e-06,
|
| 65 |
+
"loss": 0.993,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.036,
|
| 70 |
+
"grad_norm": 11.875,
|
| 71 |
+
"learning_rate": 7.1200000000000004e-06,
|
| 72 |
+
"loss": 0.9371,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.04,
|
| 77 |
+
"grad_norm": 234.0,
|
| 78 |
+
"learning_rate": 7.92e-06,
|
| 79 |
+
"loss": 0.9625,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.044,
|
| 84 |
+
"grad_norm": 4.25,
|
| 85 |
+
"learning_rate": 8.720000000000001e-06,
|
| 86 |
+
"loss": 0.8336,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.048,
|
| 91 |
+
"grad_norm": 4.0,
|
| 92 |
+
"learning_rate": 9.52e-06,
|
| 93 |
+
"loss": 0.7688,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.052,
|
| 98 |
+
"grad_norm": 3.890625,
|
| 99 |
+
"learning_rate": 1.0320000000000001e-05,
|
| 100 |
+
"loss": 0.8051,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.056,
|
| 105 |
+
"grad_norm": 3.015625,
|
| 106 |
+
"learning_rate": 1.1120000000000002e-05,
|
| 107 |
+
"loss": 0.7707,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.06,
|
| 112 |
+
"grad_norm": 3.890625,
|
| 113 |
+
"learning_rate": 1.1920000000000001e-05,
|
| 114 |
+
"loss": 0.7824,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.064,
|
| 119 |
+
"grad_norm": 3.4375,
|
| 120 |
+
"learning_rate": 1.2720000000000002e-05,
|
| 121 |
+
"loss": 0.7668,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.068,
|
| 126 |
+
"grad_norm": 2.953125,
|
| 127 |
+
"learning_rate": 1.3520000000000003e-05,
|
| 128 |
+
"loss": 0.6629,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.072,
|
| 133 |
+
"grad_norm": 4.1875,
|
| 134 |
+
"learning_rate": 1.432e-05,
|
| 135 |
+
"loss": 0.7113,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.076,
|
| 140 |
+
"grad_norm": 3.03125,
|
| 141 |
+
"learning_rate": 1.5120000000000001e-05,
|
| 142 |
+
"loss": 0.6758,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.08,
|
| 147 |
+
"grad_norm": 4.21875,
|
| 148 |
+
"learning_rate": 1.5920000000000003e-05,
|
| 149 |
+
"loss": 0.6562,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.084,
|
| 154 |
+
"grad_norm": 2.40625,
|
| 155 |
+
"learning_rate": 1.672e-05,
|
| 156 |
+
"loss": 0.6422,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.088,
|
| 161 |
+
"grad_norm": 2.4375,
|
| 162 |
+
"learning_rate": 1.752e-05,
|
| 163 |
+
"loss": 0.6508,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.092,
|
| 168 |
+
"grad_norm": 3.09375,
|
| 169 |
+
"learning_rate": 1.832e-05,
|
| 170 |
+
"loss": 0.6668,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.096,
|
| 175 |
+
"grad_norm": 2.28125,
|
| 176 |
+
"learning_rate": 1.912e-05,
|
| 177 |
+
"loss": 0.6273,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.1,
|
| 182 |
+
"grad_norm": 3.671875,
|
| 183 |
+
"learning_rate": 1.9920000000000002e-05,
|
| 184 |
+
"loss": 0.6539,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.104,
|
| 189 |
+
"grad_norm": 2.875,
|
| 190 |
+
"learning_rate": 1.9999210442038164e-05,
|
| 191 |
+
"loss": 0.657,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.108,
|
| 196 |
+
"grad_norm": 2.421875,
|
| 197 |
+
"learning_rate": 1.9996481265944146e-05,
|
| 198 |
+
"loss": 0.6375,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.112,
|
| 203 |
+
"grad_norm": 2.265625,
|
| 204 |
+
"learning_rate": 1.9991803256020393e-05,
|
| 205 |
+
"loss": 0.6051,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.116,
|
| 210 |
+
"grad_norm": 2.3125,
|
| 211 |
+
"learning_rate": 1.99851773242542e-05,
|
| 212 |
+
"loss": 0.591,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.12,
|
| 217 |
+
"grad_norm": 2.96875,
|
| 218 |
+
"learning_rate": 1.99766047623841e-05,
|
| 219 |
+
"loss": 0.5902,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.124,
|
| 224 |
+
"grad_norm": 2.375,
|
| 225 |
+
"learning_rate": 1.996608724164801e-05,
|
| 226 |
+
"loss": 0.6107,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.128,
|
| 231 |
+
"grad_norm": 2.4375,
|
| 232 |
+
"learning_rate": 1.995362681245744e-05,
|
| 233 |
+
"loss": 0.5801,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.132,
|
| 238 |
+
"grad_norm": 2.421875,
|
| 239 |
+
"learning_rate": 1.9939225903997748e-05,
|
| 240 |
+
"loss": 0.6066,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.136,
|
| 245 |
+
"grad_norm": 2.265625,
|
| 246 |
+
"learning_rate": 1.992288732375458e-05,
|
| 247 |
+
"loss": 0.5309,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.14,
|
| 252 |
+
"grad_norm": 2.109375,
|
| 253 |
+
"learning_rate": 1.9904614256966514e-05,
|
| 254 |
+
"loss": 0.5473,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.144,
|
| 259 |
+
"grad_norm": 2.015625,
|
| 260 |
+
"learning_rate": 1.9884410266004134e-05,
|
| 261 |
+
"loss": 0.5816,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.148,
|
| 266 |
+
"grad_norm": 3.0625,
|
| 267 |
+
"learning_rate": 1.986227928967551e-05,
|
| 268 |
+
"loss": 0.5543,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.152,
|
| 273 |
+
"grad_norm": 1.9296875,
|
| 274 |
+
"learning_rate": 1.983822564245833e-05,
|
| 275 |
+
"loss": 0.5283,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.156,
|
| 280 |
+
"grad_norm": 2.421875,
|
| 281 |
+
"learning_rate": 1.981225401365877e-05,
|
| 282 |
+
"loss": 0.5176,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.16,
|
| 287 |
+
"grad_norm": 1.7890625,
|
| 288 |
+
"learning_rate": 1.9784369466497333e-05,
|
| 289 |
+
"loss": 0.5184,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.164,
|
| 294 |
+
"grad_norm": 2.0625,
|
| 295 |
+
"learning_rate": 1.9754577437121733e-05,
|
| 296 |
+
"loss": 0.5652,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.168,
|
| 301 |
+
"grad_norm": 3.125,
|
| 302 |
+
"learning_rate": 1.9722883733547128e-05,
|
| 303 |
+
"loss": 0.5262,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.172,
|
| 308 |
+
"grad_norm": 1.953125,
|
| 309 |
+
"learning_rate": 1.968929453452383e-05,
|
| 310 |
+
"loss": 0.5352,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.176,
|
| 315 |
+
"grad_norm": 1.6015625,
|
| 316 |
+
"learning_rate": 1.965381638833274e-05,
|
| 317 |
+
"loss": 0.4859,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.18,
|
| 322 |
+
"grad_norm": 1.78125,
|
| 323 |
+
"learning_rate": 1.9616456211508756e-05,
|
| 324 |
+
"loss": 0.5305,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.184,
|
| 329 |
+
"grad_norm": 1.859375,
|
| 330 |
+
"learning_rate": 1.9577221287492368e-05,
|
| 331 |
+
"loss": 0.5123,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.188,
|
| 336 |
+
"grad_norm": 1.8828125,
|
| 337 |
+
"learning_rate": 1.9536119265209763e-05,
|
| 338 |
+
"loss": 0.5275,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.192,
|
| 343 |
+
"grad_norm": 1.9765625,
|
| 344 |
+
"learning_rate": 1.9493158157581617e-05,
|
| 345 |
+
"loss": 0.5264,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.196,
|
| 350 |
+
"grad_norm": 1.875,
|
| 351 |
+
"learning_rate": 1.9448346339960984e-05,
|
| 352 |
+
"loss": 0.5463,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.2,
|
| 357 |
+
"grad_norm": 1.8359375,
|
| 358 |
+
"learning_rate": 1.9401692548500504e-05,
|
| 359 |
+
"loss": 0.4748,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.204,
|
| 364 |
+
"grad_norm": 1.875,
|
| 365 |
+
"learning_rate": 1.935320587844926e-05,
|
| 366 |
+
"loss": 0.5252,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.208,
|
| 371 |
+
"grad_norm": 1.890625,
|
| 372 |
+
"learning_rate": 1.9302895782379648e-05,
|
| 373 |
+
"loss": 0.4865,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.212,
|
| 378 |
+
"grad_norm": 1.65625,
|
| 379 |
+
"learning_rate": 1.925077206834458e-05,
|
| 380 |
+
"loss": 0.4857,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.216,
|
| 385 |
+
"grad_norm": 1.8359375,
|
| 386 |
+
"learning_rate": 1.9196844897965393e-05,
|
| 387 |
+
"loss": 0.4979,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.22,
|
| 392 |
+
"grad_norm": 1.7734375,
|
| 393 |
+
"learning_rate": 1.914112478445079e-05,
|
| 394 |
+
"loss": 0.4949,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.224,
|
| 399 |
+
"grad_norm": 1.65625,
|
| 400 |
+
"learning_rate": 1.9083622590547313e-05,
|
| 401 |
+
"loss": 0.4988,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.228,
|
| 406 |
+
"grad_norm": 1.9296875,
|
| 407 |
+
"learning_rate": 1.9024349526421596e-05,
|
| 408 |
+
"loss": 0.5363,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.232,
|
| 413 |
+
"grad_norm": 2.078125,
|
| 414 |
+
"learning_rate": 1.896331714747493e-05,
|
| 415 |
+
"loss": 0.5225,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.236,
|
| 420 |
+
"grad_norm": 1.9140625,
|
| 421 |
+
"learning_rate": 1.8900537352090523e-05,
|
| 422 |
+
"loss": 0.5123,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.24,
|
| 427 |
+
"grad_norm": 1.59375,
|
| 428 |
+
"learning_rate": 1.8836022379313884e-05,
|
| 429 |
+
"loss": 0.4934,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.244,
|
| 434 |
+
"grad_norm": 1.765625,
|
| 435 |
+
"learning_rate": 1.8769784806466768e-05,
|
| 436 |
+
"loss": 0.4902,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.248,
|
| 441 |
+
"grad_norm": 1.6875,
|
| 442 |
+
"learning_rate": 1.870183754669526e-05,
|
| 443 |
+
"loss": 0.4932,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.252,
|
| 448 |
+
"grad_norm": 1.9453125,
|
| 449 |
+
"learning_rate": 1.863219384645227e-05,
|
| 450 |
+
"loss": 0.4648,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.256,
|
| 455 |
+
"grad_norm": 2.109375,
|
| 456 |
+
"learning_rate": 1.8560867282915164e-05,
|
| 457 |
+
"loss": 0.5135,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.26,
|
| 462 |
+
"grad_norm": 1.859375,
|
| 463 |
+
"learning_rate": 1.848787176133882e-05,
|
| 464 |
+
"loss": 0.4836,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.264,
|
| 469 |
+
"grad_norm": 2.734375,
|
| 470 |
+
"learning_rate": 1.8413221512344805e-05,
|
| 471 |
+
"loss": 0.5049,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.268,
|
| 476 |
+
"grad_norm": 2.171875,
|
| 477 |
+
"learning_rate": 1.8336931089147076e-05,
|
| 478 |
+
"loss": 0.5375,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.272,
|
| 483 |
+
"grad_norm": 1.7109375,
|
| 484 |
+
"learning_rate": 1.8259015364714786e-05,
|
| 485 |
+
"loss": 0.5139,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.276,
|
| 490 |
+
"grad_norm": 1.578125,
|
| 491 |
+
"learning_rate": 1.8179489528872808e-05,
|
| 492 |
+
"loss": 0.5223,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.28,
|
| 497 |
+
"grad_norm": 1.6484375,
|
| 498 |
+
"learning_rate": 1.80983690853404e-05,
|
| 499 |
+
"loss": 0.5131,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.284,
|
| 504 |
+
"grad_norm": 1.9921875,
|
| 505 |
+
"learning_rate": 1.8015669848708768e-05,
|
| 506 |
+
"loss": 0.468,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.288,
|
| 511 |
+
"grad_norm": 1.8828125,
|
| 512 |
+
"learning_rate": 1.793140794135795e-05,
|
| 513 |
+
"loss": 0.4797,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.292,
|
| 518 |
+
"grad_norm": 2.015625,
|
| 519 |
+
"learning_rate": 1.7845599790313735e-05,
|
| 520 |
+
"loss": 0.4699,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.296,
|
| 525 |
+
"grad_norm": 1.6171875,
|
| 526 |
+
"learning_rate": 1.7758262124045195e-05,
|
| 527 |
+
"loss": 0.5324,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.3,
|
| 532 |
+
"grad_norm": 2.03125,
|
| 533 |
+
"learning_rate": 1.7669411969203417e-05,
|
| 534 |
+
"loss": 0.4785,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.304,
|
| 539 |
+
"grad_norm": 1.71875,
|
| 540 |
+
"learning_rate": 1.7579066647302134e-05,
|
| 541 |
+
"loss": 0.4814,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.308,
|
| 546 |
+
"grad_norm": 1.6171875,
|
| 547 |
+
"learning_rate": 1.7487243771340862e-05,
|
| 548 |
+
"loss": 0.4537,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.312,
|
| 553 |
+
"grad_norm": 4.3125,
|
| 554 |
+
"learning_rate": 1.7393961242371203e-05,
|
| 555 |
+
"loss": 0.4742,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.316,
|
| 560 |
+
"grad_norm": 2.125,
|
| 561 |
+
"learning_rate": 1.7299237246007018e-05,
|
| 562 |
+
"loss": 0.4926,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.32,
|
| 567 |
+
"grad_norm": 1.75,
|
| 568 |
+
"learning_rate": 1.720309024887907e-05,
|
| 569 |
+
"loss": 0.4881,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.324,
|
| 574 |
+
"grad_norm": 1.8984375,
|
| 575 |
+
"learning_rate": 1.710553899503496e-05,
|
| 576 |
+
"loss": 0.4857,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.328,
|
| 581 |
+
"grad_norm": 1.7578125,
|
| 582 |
+
"learning_rate": 1.700660250228492e-05,
|
| 583 |
+
"loss": 0.5352,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.332,
|
| 588 |
+
"grad_norm": 1.5390625,
|
| 589 |
+
"learning_rate": 1.690630005849423e-05,
|
| 590 |
+
"loss": 0.4445,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.336,
|
| 595 |
+
"grad_norm": 1.890625,
|
| 596 |
+
"learning_rate": 1.6804651217823055e-05,
|
| 597 |
+
"loss": 0.5189,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.34,
|
| 602 |
+
"grad_norm": 1.546875,
|
| 603 |
+
"learning_rate": 1.6701675796914284e-05,
|
| 604 |
+
"loss": 0.4311,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.344,
|
| 609 |
+
"grad_norm": 1.7109375,
|
| 610 |
+
"learning_rate": 1.6597393871030264e-05,
|
| 611 |
+
"loss": 0.4873,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.348,
|
| 616 |
+
"grad_norm": 1.6875,
|
| 617 |
+
"learning_rate": 1.649182577013906e-05,
|
| 618 |
+
"loss": 0.4633,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.352,
|
| 623 |
+
"grad_norm": 1.5625,
|
| 624 |
+
"learning_rate": 1.6384992074951124e-05,
|
| 625 |
+
"loss": 0.4436,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.356,
|
| 630 |
+
"grad_norm": 1.65625,
|
| 631 |
+
"learning_rate": 1.6276913612907005e-05,
|
| 632 |
+
"loss": 0.4871,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.36,
|
| 637 |
+
"grad_norm": 1.7421875,
|
| 638 |
+
"learning_rate": 1.6167611454117027e-05,
|
| 639 |
+
"loss": 0.4613,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.364,
|
| 644 |
+
"grad_norm": 1.65625,
|
| 645 |
+
"learning_rate": 1.6057106907253617e-05,
|
| 646 |
+
"loss": 0.4684,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.368,
|
| 651 |
+
"grad_norm": 1.734375,
|
| 652 |
+
"learning_rate": 1.5945421515397135e-05,
|
| 653 |
+
"loss": 0.5123,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.372,
|
| 658 |
+
"grad_norm": 1.7890625,
|
| 659 |
+
"learning_rate": 1.5832577051836016e-05,
|
| 660 |
+
"loss": 0.5119,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.376,
|
| 665 |
+
"grad_norm": 1.4375,
|
| 666 |
+
"learning_rate": 1.5718595515822027e-05,
|
| 667 |
+
"loss": 0.4494,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.38,
|
| 672 |
+
"grad_norm": 1.609375,
|
| 673 |
+
"learning_rate": 1.5603499128281447e-05,
|
| 674 |
+
"loss": 0.4803,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.384,
|
| 679 |
+
"grad_norm": 1.7265625,
|
| 680 |
+
"learning_rate": 1.5487310327483087e-05,
|
| 681 |
+
"loss": 0.5055,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.388,
|
| 686 |
+
"grad_norm": 1.546875,
|
| 687 |
+
"learning_rate": 1.5370051764663872e-05,
|
| 688 |
+
"loss": 0.4959,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.392,
|
| 693 |
+
"grad_norm": 1.8515625,
|
| 694 |
+
"learning_rate": 1.5251746299612959e-05,
|
| 695 |
+
"loss": 0.4793,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.396,
|
| 700 |
+
"grad_norm": 1.8203125,
|
| 701 |
+
"learning_rate": 1.5132416996215171e-05,
|
| 702 |
+
"loss": 0.516,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.4,
|
| 707 |
+
"grad_norm": 1.578125,
|
| 708 |
+
"learning_rate": 1.5012087117954643e-05,
|
| 709 |
+
"loss": 0.4549,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.404,
|
| 714 |
+
"grad_norm": 1.46875,
|
| 715 |
+
"learning_rate": 1.4890780123379565e-05,
|
| 716 |
+
"loss": 0.467,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.408,
|
| 721 |
+
"grad_norm": 1.6953125,
|
| 722 |
+
"learning_rate": 1.4768519661528879e-05,
|
| 723 |
+
"loss": 0.492,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.412,
|
| 728 |
+
"grad_norm": 1.6640625,
|
| 729 |
+
"learning_rate": 1.464532956732188e-05,
|
| 730 |
+
"loss": 0.4711,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.416,
|
| 735 |
+
"grad_norm": 4.03125,
|
| 736 |
+
"learning_rate": 1.4521233856911507e-05,
|
| 737 |
+
"loss": 0.4842,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.42,
|
| 742 |
+
"grad_norm": 1.59375,
|
| 743 |
+
"learning_rate": 1.43962567230024e-05,
|
| 744 |
+
"loss": 0.4535,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.424,
|
| 749 |
+
"grad_norm": 1.6171875,
|
| 750 |
+
"learning_rate": 1.4270422530134433e-05,
|
| 751 |
+
"loss": 0.4813,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.428,
|
| 756 |
+
"grad_norm": 1.2578125,
|
| 757 |
+
"learning_rate": 1.4143755809932843e-05,
|
| 758 |
+
"loss": 0.4453,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.432,
|
| 763 |
+
"grad_norm": 1.5625,
|
| 764 |
+
"learning_rate": 1.4016281256325702e-05,
|
| 765 |
+
"loss": 0.4578,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.436,
|
| 770 |
+
"grad_norm": 1.953125,
|
| 771 |
+
"learning_rate": 1.388802372072981e-05,
|
| 772 |
+
"loss": 0.4559,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.44,
|
| 777 |
+
"grad_norm": 1.9296875,
|
| 778 |
+
"learning_rate": 1.3759008207205869e-05,
|
| 779 |
+
"loss": 0.5078,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.444,
|
| 784 |
+
"grad_norm": 1.75,
|
| 785 |
+
"learning_rate": 1.3629259867583864e-05,
|
| 786 |
+
"loss": 0.4645,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.448,
|
| 791 |
+
"grad_norm": 1.8046875,
|
| 792 |
+
"learning_rate": 1.349880399655969e-05,
|
| 793 |
+
"loss": 0.4395,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.452,
|
| 798 |
+
"grad_norm": 1.4921875,
|
| 799 |
+
"learning_rate": 1.3367666026763884e-05,
|
| 800 |
+
"loss": 0.4641,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.456,
|
| 805 |
+
"grad_norm": 1.625,
|
| 806 |
+
"learning_rate": 1.3235871523803496e-05,
|
| 807 |
+
"loss": 0.4324,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.46,
|
| 812 |
+
"grad_norm": 1.6796875,
|
| 813 |
+
"learning_rate": 1.3103446181278015e-05,
|
| 814 |
+
"loss": 0.4684,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.464,
|
| 819 |
+
"grad_norm": 1.78125,
|
| 820 |
+
"learning_rate": 1.297041581577035e-05,
|
| 821 |
+
"loss": 0.4635,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.468,
|
| 826 |
+
"grad_norm": 1.765625,
|
| 827 |
+
"learning_rate": 1.2836806361813846e-05,
|
| 828 |
+
"loss": 0.4705,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.472,
|
| 833 |
+
"grad_norm": 1.53125,
|
| 834 |
+
"learning_rate": 1.270264386683628e-05,
|
| 835 |
+
"loss": 0.4299,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.476,
|
| 840 |
+
"grad_norm": 1.6328125,
|
| 841 |
+
"learning_rate": 1.256795448608188e-05,
|
| 842 |
+
"loss": 0.4813,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.48,
|
| 847 |
+
"grad_norm": 1.6953125,
|
| 848 |
+
"learning_rate": 1.2432764477512294e-05,
|
| 849 |
+
"loss": 0.4609,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.484,
|
| 854 |
+
"grad_norm": 1.8046875,
|
| 855 |
+
"learning_rate": 1.2297100196687557e-05,
|
| 856 |
+
"loss": 0.4916,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.488,
|
| 861 |
+
"grad_norm": 1.5703125,
|
| 862 |
+
"learning_rate": 1.2160988091628023e-05,
|
| 863 |
+
"loss": 0.459,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.492,
|
| 868 |
+
"grad_norm": 2.078125,
|
| 869 |
+
"learning_rate": 1.202445469765826e-05,
|
| 870 |
+
"loss": 0.4824,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.496,
|
| 875 |
+
"grad_norm": 1.78125,
|
| 876 |
+
"learning_rate": 1.1887526632233954e-05,
|
| 877 |
+
"loss": 0.4771,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.5,
|
| 882 |
+
"grad_norm": 1.8359375,
|
| 883 |
+
"learning_rate": 1.1750230589752763e-05,
|
| 884 |
+
"loss": 0.4734,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.504,
|
| 889 |
+
"grad_norm": 1.734375,
|
| 890 |
+
"learning_rate": 1.1612593336350209e-05,
|
| 891 |
+
"loss": 0.5332,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.508,
|
| 896 |
+
"grad_norm": 1.40625,
|
| 897 |
+
"learning_rate": 1.1474641704681551e-05,
|
| 898 |
+
"loss": 0.4105,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.512,
|
| 903 |
+
"grad_norm": 1.40625,
|
| 904 |
+
"learning_rate": 1.1336402588690727e-05,
|
| 905 |
+
"loss": 0.4398,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.516,
|
| 910 |
+
"grad_norm": 1.59375,
|
| 911 |
+
"learning_rate": 1.1197902938367297e-05,
|
| 912 |
+
"loss": 0.4486,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.52,
|
| 917 |
+
"grad_norm": 1.375,
|
| 918 |
+
"learning_rate": 1.105916975449252e-05,
|
| 919 |
+
"loss": 0.4705,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.524,
|
| 924 |
+
"grad_norm": 1.5859375,
|
| 925 |
+
"learning_rate": 1.0920230083375474e-05,
|
| 926 |
+
"loss": 0.4553,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.528,
|
| 931 |
+
"grad_norm": 1.2734375,
|
| 932 |
+
"learning_rate": 1.0781111011580336e-05,
|
| 933 |
+
"loss": 0.4098,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.532,
|
| 938 |
+
"grad_norm": 1.609375,
|
| 939 |
+
"learning_rate": 1.0641839660645806e-05,
|
| 940 |
+
"loss": 0.4551,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.536,
|
| 945 |
+
"grad_norm": 1.6015625,
|
| 946 |
+
"learning_rate": 1.0502443181797696e-05,
|
| 947 |
+
"loss": 0.4047,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.54,
|
| 952 |
+
"grad_norm": 1.609375,
|
| 953 |
+
"learning_rate": 1.036294875065576e-05,
|
| 954 |
+
"loss": 0.4584,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.544,
|
| 959 |
+
"grad_norm": 1.4921875,
|
| 960 |
+
"learning_rate": 1.0223383561935738e-05,
|
| 961 |
+
"loss": 0.4338,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.548,
|
| 966 |
+
"grad_norm": 1.8125,
|
| 967 |
+
"learning_rate": 1.0083774824147707e-05,
|
| 968 |
+
"loss": 0.4584,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.552,
|
| 973 |
+
"grad_norm": 1.5,
|
| 974 |
+
"learning_rate": 9.944149754291719e-06,
|
| 975 |
+
"loss": 0.442,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.556,
|
| 980 |
+
"grad_norm": 1.296875,
|
| 981 |
+
"learning_rate": 9.80453557255179e-06,
|
| 982 |
+
"loss": 0.4059,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.56,
|
| 987 |
+
"grad_norm": 1.4375,
|
| 988 |
+
"learning_rate": 9.664959496989286e-06,
|
| 989 |
+
"loss": 0.4334,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.564,
|
| 994 |
+
"grad_norm": 1.515625,
|
| 995 |
+
"learning_rate": 9.525448738236691e-06,
|
| 996 |
+
"loss": 0.4059,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.568,
|
| 1001 |
+
"grad_norm": 1.5078125,
|
| 1002 |
+
"learning_rate": 9.386030494192847e-06,
|
| 1003 |
+
"loss": 0.4322,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.572,
|
| 1008 |
+
"grad_norm": 1.546875,
|
| 1009 |
+
"learning_rate": 9.246731944720675e-06,
|
| 1010 |
+
"loss": 0.435,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.576,
|
| 1015 |
+
"grad_norm": 1.5,
|
| 1016 |
+
"learning_rate": 9.107580246348395e-06,
|
| 1017 |
+
"loss": 0.4232,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.58,
|
| 1022 |
+
"grad_norm": 1.6328125,
|
| 1023 |
+
"learning_rate": 8.968602526975329e-06,
|
| 1024 |
+
"loss": 0.4484,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.584,
|
| 1029 |
+
"grad_norm": 1.4609375,
|
| 1030 |
+
"learning_rate": 8.829825880583228e-06,
|
| 1031 |
+
"loss": 0.4379,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.588,
|
| 1036 |
+
"grad_norm": 1.7734375,
|
| 1037 |
+
"learning_rate": 8.69127736195428e-06,
|
| 1038 |
+
"loss": 0.4096,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.592,
|
| 1043 |
+
"grad_norm": 1.3984375,
|
| 1044 |
+
"learning_rate": 8.552983981396709e-06,
|
| 1045 |
+
"loss": 0.4352,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.596,
|
| 1050 |
+
"grad_norm": 1.546875,
|
| 1051 |
+
"learning_rate": 8.414972699479076e-06,
|
| 1052 |
+
"loss": 0.4191,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.6,
|
| 1057 |
+
"grad_norm": 1.46875,
|
| 1058 |
+
"learning_rate": 8.277270421774234e-06,
|
| 1059 |
+
"loss": 0.4219,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.604,
|
| 1064 |
+
"grad_norm": 1.734375,
|
| 1065 |
+
"learning_rate": 8.139903993614069e-06,
|
| 1066 |
+
"loss": 0.4318,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.608,
|
| 1071 |
+
"grad_norm": 1.3359375,
|
| 1072 |
+
"learning_rate": 8.00290019485593e-06,
|
| 1073 |
+
"loss": 0.3777,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.612,
|
| 1078 |
+
"grad_norm": 1.6015625,
|
| 1079 |
+
"learning_rate": 7.866285734661842e-06,
|
| 1080 |
+
"loss": 0.4092,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.616,
|
| 1085 |
+
"grad_norm": 1.7109375,
|
| 1086 |
+
"learning_rate": 7.730087246291503e-06,
|
| 1087 |
+
"loss": 0.468,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.62,
|
| 1092 |
+
"grad_norm": 1.71875,
|
| 1093 |
+
"learning_rate": 7.594331281910082e-06,
|
| 1094 |
+
"loss": 0.4207,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.624,
|
| 1099 |
+
"grad_norm": 1.6015625,
|
| 1100 |
+
"learning_rate": 7.4590443074118325e-06,
|
| 1101 |
+
"loss": 0.3861,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.628,
|
| 1106 |
+
"grad_norm": 1.8203125,
|
| 1107 |
+
"learning_rate": 7.324252697260475e-06,
|
| 1108 |
+
"loss": 0.3957,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.632,
|
| 1113 |
+
"grad_norm": 1.6875,
|
| 1114 |
+
"learning_rate": 7.189982729347491e-06,
|
| 1115 |
+
"loss": 0.4182,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.636,
|
| 1120 |
+
"grad_norm": 1.6328125,
|
| 1121 |
+
"learning_rate": 7.056260579869165e-06,
|
| 1122 |
+
"loss": 0.4166,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.64,
|
| 1127 |
+
"grad_norm": 1.6484375,
|
| 1128 |
+
"learning_rate": 6.923112318223497e-06,
|
| 1129 |
+
"loss": 0.4355,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.644,
|
| 1134 |
+
"grad_norm": 1.5703125,
|
| 1135 |
+
"learning_rate": 6.790563901927907e-06,
|
| 1136 |
+
"loss": 0.385,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.648,
|
| 1141 |
+
"grad_norm": 1.640625,
|
| 1142 |
+
"learning_rate": 6.658641171558785e-06,
|
| 1143 |
+
"loss": 0.409,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.652,
|
| 1148 |
+
"grad_norm": 1.375,
|
| 1149 |
+
"learning_rate": 6.52736984571381e-06,
|
| 1150 |
+
"loss": 0.3865,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.656,
|
| 1155 |
+
"grad_norm": 1.6015625,
|
| 1156 |
+
"learning_rate": 6.396775515998055e-06,
|
| 1157 |
+
"loss": 0.3941,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.66,
|
| 1162 |
+
"grad_norm": 1.4296875,
|
| 1163 |
+
"learning_rate": 6.2668836420348535e-06,
|
| 1164 |
+
"loss": 0.3771,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.664,
|
| 1169 |
+
"grad_norm": 1.7421875,
|
| 1170 |
+
"learning_rate": 6.137719546502401e-06,
|
| 1171 |
+
"loss": 0.4125,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.668,
|
| 1176 |
+
"grad_norm": 1.6796875,
|
| 1177 |
+
"learning_rate": 6.009308410197048e-06,
|
| 1178 |
+
"loss": 0.399,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.672,
|
| 1183 |
+
"grad_norm": 1.7421875,
|
| 1184 |
+
"learning_rate": 5.881675267124254e-06,
|
| 1185 |
+
"loss": 0.4168,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.676,
|
| 1190 |
+
"grad_norm": 1.5078125,
|
| 1191 |
+
"learning_rate": 5.754844999618144e-06,
|
| 1192 |
+
"loss": 0.3693,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.68,
|
| 1197 |
+
"grad_norm": 1.6484375,
|
| 1198 |
+
"learning_rate": 5.628842333490674e-06,
|
| 1199 |
+
"loss": 0.4236,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.684,
|
| 1204 |
+
"grad_norm": 1.5234375,
|
| 1205 |
+
"learning_rate": 5.50369183321126e-06,
|
| 1206 |
+
"loss": 0.4057,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.688,
|
| 1211 |
+
"grad_norm": 1.703125,
|
| 1212 |
+
"learning_rate": 5.379417897117917e-06,
|
| 1213 |
+
"loss": 0.3937,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.692,
|
| 1218 |
+
"grad_norm": 1.78125,
|
| 1219 |
+
"learning_rate": 5.256044752660709e-06,
|
| 1220 |
+
"loss": 0.3998,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.696,
|
| 1225 |
+
"grad_norm": 1.8984375,
|
| 1226 |
+
"learning_rate": 5.133596451678603e-06,
|
| 1227 |
+
"loss": 0.4027,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.7,
|
| 1232 |
+
"grad_norm": 1.625,
|
| 1233 |
+
"learning_rate": 5.012096865710494e-06,
|
| 1234 |
+
"loss": 0.4459,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.704,
|
| 1239 |
+
"grad_norm": 1.5078125,
|
| 1240 |
+
"learning_rate": 4.891569681341403e-06,
|
| 1241 |
+
"loss": 0.3854,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.708,
|
| 1246 |
+
"grad_norm": 1.7734375,
|
| 1247 |
+
"learning_rate": 4.772038395584735e-06,
|
| 1248 |
+
"loss": 0.4125,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.712,
|
| 1253 |
+
"grad_norm": 1.3359375,
|
| 1254 |
+
"learning_rate": 4.6535263113014885e-06,
|
| 1255 |
+
"loss": 0.3865,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.716,
|
| 1260 |
+
"grad_norm": 1.578125,
|
| 1261 |
+
"learning_rate": 4.53605653265731e-06,
|
| 1262 |
+
"loss": 0.3973,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.72,
|
| 1267 |
+
"grad_norm": 1.4609375,
|
| 1268 |
+
"learning_rate": 4.419651960618302e-06,
|
| 1269 |
+
"loss": 0.4137,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.724,
|
| 1274 |
+
"grad_norm": 1.6015625,
|
| 1275 |
+
"learning_rate": 4.304335288486426e-06,
|
| 1276 |
+
"loss": 0.4145,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.728,
|
| 1281 |
+
"grad_norm": 1.4296875,
|
| 1282 |
+
"learning_rate": 4.190128997475402e-06,
|
| 1283 |
+
"loss": 0.3926,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.732,
|
| 1288 |
+
"grad_norm": 1.71875,
|
| 1289 |
+
"learning_rate": 4.0770553523279535e-06,
|
| 1290 |
+
"loss": 0.3898,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.736,
|
| 1295 |
+
"grad_norm": 1.5625,
|
| 1296 |
+
"learning_rate": 3.965136396975235e-06,
|
| 1297 |
+
"loss": 0.4184,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.74,
|
| 1302 |
+
"grad_norm": 1.953125,
|
| 1303 |
+
"learning_rate": 3.854393950239356e-06,
|
| 1304 |
+
"loss": 0.4225,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.744,
|
| 1309 |
+
"grad_norm": 1.75,
|
| 1310 |
+
"learning_rate": 3.7448496015797296e-06,
|
| 1311 |
+
"loss": 0.3936,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.748,
|
| 1316 |
+
"grad_norm": 1.5703125,
|
| 1317 |
+
"learning_rate": 3.636524706884181e-06,
|
| 1318 |
+
"loss": 0.4172,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.752,
|
| 1323 |
+
"grad_norm": 1.4765625,
|
| 1324 |
+
"learning_rate": 3.5294403843055604e-06,
|
| 1325 |
+
"loss": 0.3791,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.756,
|
| 1330 |
+
"grad_norm": 1.5625,
|
| 1331 |
+
"learning_rate": 3.4236175101447265e-06,
|
| 1332 |
+
"loss": 0.3812,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.76,
|
| 1337 |
+
"grad_norm": 1.53125,
|
| 1338 |
+
"learning_rate": 3.3190767147806825e-06,
|
| 1339 |
+
"loss": 0.4004,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.764,
|
| 1344 |
+
"grad_norm": 1.5703125,
|
| 1345 |
+
"learning_rate": 3.2158383786486204e-06,
|
| 1346 |
+
"loss": 0.3898,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.768,
|
| 1351 |
+
"grad_norm": 1.46875,
|
| 1352 |
+
"learning_rate": 3.113922628266718e-06,
|
| 1353 |
+
"loss": 0.4053,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.772,
|
| 1358 |
+
"grad_norm": 1.65625,
|
| 1359 |
+
"learning_rate": 3.013349332312451e-06,
|
| 1360 |
+
"loss": 0.4156,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.776,
|
| 1365 |
+
"grad_norm": 1.5859375,
|
| 1366 |
+
"learning_rate": 2.9141380977491373e-06,
|
| 1367 |
+
"loss": 0.3939,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.78,
|
| 1372 |
+
"grad_norm": 1.359375,
|
| 1373 |
+
"learning_rate": 2.816308266003541e-06,
|
| 1374 |
+
"loss": 0.3773,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.784,
|
| 1379 |
+
"grad_norm": 1.578125,
|
| 1380 |
+
"learning_rate": 2.7198789091951903e-06,
|
| 1381 |
+
"loss": 0.374,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.788,
|
| 1386 |
+
"grad_norm": 1.453125,
|
| 1387 |
+
"learning_rate": 2.624868826418262e-06,
|
| 1388 |
+
"loss": 0.3885,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.792,
|
| 1393 |
+
"grad_norm": 1.4921875,
|
| 1394 |
+
"learning_rate": 2.5312965400766475e-06,
|
| 1395 |
+
"loss": 0.3891,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.796,
|
| 1400 |
+
"grad_norm": 1.6328125,
|
| 1401 |
+
"learning_rate": 2.4391802922729703e-06,
|
| 1402 |
+
"loss": 0.3861,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.8,
|
| 1407 |
+
"grad_norm": 1.3125,
|
| 1408 |
+
"learning_rate": 2.3485380412522586e-06,
|
| 1409 |
+
"loss": 0.3674,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.804,
|
| 1414 |
+
"grad_norm": 1.6015625,
|
| 1415 |
+
"learning_rate": 2.259387457900948e-06,
|
| 1416 |
+
"loss": 0.391,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.808,
|
| 1421 |
+
"grad_norm": 1.6875,
|
| 1422 |
+
"learning_rate": 2.171745922301903e-06,
|
| 1423 |
+
"loss": 0.4186,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.812,
|
| 1428 |
+
"grad_norm": 1.53125,
|
| 1429 |
+
"learning_rate": 2.0856305203461436e-06,
|
| 1430 |
+
"loss": 0.3592,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.816,
|
| 1435 |
+
"grad_norm": 1.4609375,
|
| 1436 |
+
"learning_rate": 2.0010580404019066e-06,
|
| 1437 |
+
"loss": 0.3977,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.82,
|
| 1442 |
+
"grad_norm": 1.4453125,
|
| 1443 |
+
"learning_rate": 1.918044970041729e-06,
|
| 1444 |
+
"loss": 0.398,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.824,
|
| 1449 |
+
"grad_norm": 1.6171875,
|
| 1450 |
+
"learning_rate": 1.8366074928281608e-06,
|
| 1451 |
+
"loss": 0.3736,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.828,
|
| 1456 |
+
"grad_norm": 1.703125,
|
| 1457 |
+
"learning_rate": 1.7567614851587444e-06,
|
| 1458 |
+
"loss": 0.3521,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.832,
|
| 1463 |
+
"grad_norm": 1.6484375,
|
| 1464 |
+
"learning_rate": 1.6785225131708749e-06,
|
| 1465 |
+
"loss": 0.3947,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.836,
|
| 1470 |
+
"grad_norm": 1.5234375,
|
| 1471 |
+
"learning_rate": 1.601905829707171e-06,
|
| 1472 |
+
"loss": 0.377,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.84,
|
| 1477 |
+
"grad_norm": 1.375,
|
| 1478 |
+
"learning_rate": 1.526926371341878e-06,
|
| 1479 |
+
"loss": 0.3893,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.844,
|
| 1484 |
+
"grad_norm": 1.3984375,
|
| 1485 |
+
"learning_rate": 1.4535987554689712e-06,
|
| 1486 |
+
"loss": 0.4205,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.848,
|
| 1491 |
+
"grad_norm": 1.546875,
|
| 1492 |
+
"learning_rate": 1.381937277452451e-06,
|
| 1493 |
+
"loss": 0.4303,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.852,
|
| 1498 |
+
"grad_norm": 1.6953125,
|
| 1499 |
+
"learning_rate": 1.3119559078394462e-06,
|
| 1500 |
+
"loss": 0.4213,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.856,
|
| 1505 |
+
"grad_norm": 1.8828125,
|
| 1506 |
+
"learning_rate": 1.2436682896366282e-06,
|
| 1507 |
+
"loss": 0.391,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.86,
|
| 1512 |
+
"grad_norm": 1.6015625,
|
| 1513 |
+
"learning_rate": 1.1770877356504684e-06,
|
| 1514 |
+
"loss": 0.3861,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.864,
|
| 1519 |
+
"grad_norm": 1.640625,
|
| 1520 |
+
"learning_rate": 1.1122272258918864e-06,
|
| 1521 |
+
"loss": 0.4026,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.868,
|
| 1526 |
+
"grad_norm": 1.4765625,
|
| 1527 |
+
"learning_rate": 1.0490994050457748e-06,
|
| 1528 |
+
"loss": 0.3893,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.872,
|
| 1533 |
+
"grad_norm": 3.59375,
|
| 1534 |
+
"learning_rate": 9.877165800058874e-07,
|
| 1535 |
+
"loss": 0.4078,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.876,
|
| 1540 |
+
"grad_norm": 1.75,
|
| 1541 |
+
"learning_rate": 9.280907174755916e-07,
|
| 1542 |
+
"loss": 0.373,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.88,
|
| 1547 |
+
"grad_norm": 1.6796875,
|
| 1548 |
+
"learning_rate": 8.702334416349279e-07,
|
| 1549 |
+
"loss": 0.4193,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.884,
|
| 1554 |
+
"grad_norm": 1.5234375,
|
| 1555 |
+
"learning_rate": 8.141560318744601e-07,
|
| 1556 |
+
"loss": 0.3754,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.888,
|
| 1561 |
+
"grad_norm": 1.6484375,
|
| 1562 |
+
"learning_rate": 7.598694205963331e-07,
|
| 1563 |
+
"loss": 0.3652,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.892,
|
| 1568 |
+
"grad_norm": 1.53125,
|
| 1569 |
+
"learning_rate": 7.073841910829771e-07,
|
| 1570 |
+
"loss": 0.3934,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.896,
|
| 1575 |
+
"grad_norm": 1.5234375,
|
| 1576 |
+
"learning_rate": 6.567105754338798e-07,
|
| 1577 |
+
"loss": 0.4051,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.9,
|
| 1582 |
+
"grad_norm": 1.859375,
|
| 1583 |
+
"learning_rate": 6.078584525708175e-07,
|
| 1584 |
+
"loss": 0.3695,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.904,
|
| 1589 |
+
"grad_norm": 1.484375,
|
| 1590 |
+
"learning_rate": 5.608373463119354e-07,
|
| 1591 |
+
"loss": 0.4188,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.908,
|
| 1596 |
+
"grad_norm": 1.703125,
|
| 1597 |
+
"learning_rate": 5.156564235150686e-07,
|
| 1598 |
+
"loss": 0.3867,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.912,
|
| 1603 |
+
"grad_norm": 1.5625,
|
| 1604 |
+
"learning_rate": 4.723244922906356e-07,
|
| 1605 |
+
"loss": 0.4287,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.916,
|
| 1610 |
+
"grad_norm": 1.4765625,
|
| 1611 |
+
"learning_rate": 4.308500002844862e-07,
|
| 1612 |
+
"loss": 0.368,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.92,
|
| 1617 |
+
"grad_norm": 1.4140625,
|
| 1618 |
+
"learning_rate": 3.912410330310157e-07,
|
| 1619 |
+
"loss": 0.3799,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.924,
|
| 1624 |
+
"grad_norm": 1.4765625,
|
| 1625 |
+
"learning_rate": 3.5350531237686723e-07,
|
| 1626 |
+
"loss": 0.3721,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.928,
|
| 1631 |
+
"grad_norm": 1.59375,
|
| 1632 |
+
"learning_rate": 3.1765019497555617e-07,
|
| 1633 |
+
"loss": 0.3943,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.932,
|
| 1638 |
+
"grad_norm": 1.7109375,
|
| 1639 |
+
"learning_rate": 2.836826708532603e-07,
|
| 1640 |
+
"loss": 0.376,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.936,
|
| 1645 |
+
"grad_norm": 1.8125,
|
| 1646 |
+
"learning_rate": 2.516093620461124e-07,
|
| 1647 |
+
"loss": 0.398,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.94,
|
| 1652 |
+
"grad_norm": 1.8046875,
|
| 1653 |
+
"learning_rate": 2.214365213092118e-07,
|
| 1654 |
+
"loss": 0.3855,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.944,
|
| 1659 |
+
"grad_norm": 1.890625,
|
| 1660 |
+
"learning_rate": 1.9317003089764365e-07,
|
| 1661 |
+
"loss": 0.4172,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.948,
|
| 1666 |
+
"grad_norm": 1.4375,
|
| 1667 |
+
"learning_rate": 1.668154014197243e-07,
|
| 1668 |
+
"loss": 0.425,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.952,
|
| 1673 |
+
"grad_norm": 1.453125,
|
| 1674 |
+
"learning_rate": 1.4237777076268723e-07,
|
| 1675 |
+
"loss": 0.4049,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.956,
|
| 1680 |
+
"grad_norm": 1.59375,
|
| 1681 |
+
"learning_rate": 1.1986190309104861e-07,
|
| 1682 |
+
"loss": 0.3668,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.96,
|
| 1687 |
+
"grad_norm": 1.8984375,
|
| 1688 |
+
"learning_rate": 9.9272187917826e-08,
|
| 1689 |
+
"loss": 0.4242,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.964,
|
| 1694 |
+
"grad_norm": 1.4296875,
|
| 1695 |
+
"learning_rate": 8.061263924878604e-08,
|
| 1696 |
+
"loss": 0.4172,
|
| 1697 |
+
"step": 2410
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.968,
|
| 1701 |
+
"grad_norm": 1.7578125,
|
| 1702 |
+
"learning_rate": 6.388689479991606e-08,
|
| 1703 |
+
"loss": 0.3926,
|
| 1704 |
+
"step": 2420
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.972,
|
| 1708 |
+
"grad_norm": 1.6328125,
|
| 1709 |
+
"learning_rate": 4.9098215288235776e-08,
|
| 1710 |
+
"loss": 0.451,
|
| 1711 |
+
"step": 2430
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.976,
|
| 1715 |
+
"grad_norm": 1.734375,
|
| 1716 |
+
"learning_rate": 3.6249483796116924e-08,
|
| 1717 |
+
"loss": 0.4084,
|
| 1718 |
+
"step": 2440
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.98,
|
| 1722 |
+
"grad_norm": 1.671875,
|
| 1723 |
+
"learning_rate": 2.5343205209225062e-08,
|
| 1724 |
+
"loss": 0.4176,
|
| 1725 |
+
"step": 2450
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.984,
|
| 1729 |
+
"grad_norm": 1.4921875,
|
| 1730 |
+
"learning_rate": 1.6381505728176872e-08,
|
| 1731 |
+
"loss": 0.3951,
|
| 1732 |
+
"step": 2460
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.988,
|
| 1736 |
+
"grad_norm": 1.328125,
|
| 1737 |
+
"learning_rate": 9.366132454046162e-09,
|
| 1738 |
+
"loss": 0.4207,
|
| 1739 |
+
"step": 2470
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.992,
|
| 1743 |
+
"grad_norm": 1.8046875,
|
| 1744 |
+
"learning_rate": 4.298453047749674e-09,
|
| 1745 |
+
"loss": 0.409,
|
| 1746 |
+
"step": 2480
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.996,
|
| 1750 |
+
"grad_norm": 1.6640625,
|
| 1751 |
+
"learning_rate": 1.1794554634314558e-09,
|
| 1752 |
+
"loss": 0.4086,
|
| 1753 |
+
"step": 2490
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 1.0,
|
| 1757 |
+
"grad_norm": 1.8203125,
|
| 1758 |
+
"learning_rate": 9.74775584916543e-12,
|
| 1759 |
+
"loss": 0.401,
|
| 1760 |
+
"step": 2500
|
| 1761 |
+
}
|
| 1762 |
+
],
|
| 1763 |
+
"logging_steps": 10,
|
| 1764 |
+
"max_steps": 2500,
|
| 1765 |
+
"num_input_tokens_seen": 0,
|
| 1766 |
+
"num_train_epochs": 9223372036854775807,
|
| 1767 |
+
"save_steps": 2500,
|
| 1768 |
+
"stateful_callbacks": {
|
| 1769 |
+
"TrainerControl": {
|
| 1770 |
+
"args": {
|
| 1771 |
+
"should_epoch_stop": false,
|
| 1772 |
+
"should_evaluate": false,
|
| 1773 |
+
"should_log": false,
|
| 1774 |
+
"should_save": true,
|
| 1775 |
+
"should_training_stop": true
|
| 1776 |
+
},
|
| 1777 |
+
"attributes": {}
|
| 1778 |
+
}
|
| 1779 |
+
},
|
| 1780 |
+
"total_flos": 2.3916038848512e+17,
|
| 1781 |
+
"train_batch_size": 32,
|
| 1782 |
+
"trial_name": null,
|
| 1783 |
+
"trial_params": null
|
| 1784 |
+
}
|
checkpoint-2500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1285711003bb475ba52d893e14e752e3b9c164d1ba846f3e81021ee9946facf8
|
| 3 |
+
size 5841
|