Upload 8 files
Browse files- config.json +28 -0
- generation_config.json +6 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- trainer_state.json +3191 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "hidude562/maestro-4-0619",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"Qwen2ForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"bos_token_id": 151643,
|
| 8 |
+
"eos_token_id": 151643,
|
| 9 |
+
"hidden_act": "silu",
|
| 10 |
+
"hidden_size": 1024,
|
| 11 |
+
"initializer_range": 0.02,
|
| 12 |
+
"intermediate_size": 2816,
|
| 13 |
+
"max_position_embeddings": 32768,
|
| 14 |
+
"max_window_layers": 21,
|
| 15 |
+
"model_type": "qwen2",
|
| 16 |
+
"num_attention_heads": 16,
|
| 17 |
+
"num_hidden_layers": 24,
|
| 18 |
+
"num_key_value_heads": 16,
|
| 19 |
+
"rms_norm_eps": 1e-06,
|
| 20 |
+
"rope_theta": 1000000.0,
|
| 21 |
+
"sliding_window": 32768,
|
| 22 |
+
"tie_word_embeddings": true,
|
| 23 |
+
"torch_dtype": "float32",
|
| 24 |
+
"transformers_version": "4.42.4",
|
| 25 |
+
"use_cache": true,
|
| 26 |
+
"use_sliding_window": false,
|
| 27 |
+
"vocab_size": 151936
|
| 28 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 151643,
|
| 4 |
+
"eos_token_id": 151643,
|
| 5 |
+
"transformers_version": "4.42.4"
|
| 6 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bd819371c6a9c53e128a8f49d9cd1745e699552c01897302690ce8c6b3787b04
|
| 3 |
+
size 1855983640
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1d9ca622983dbaa2a1e90ba4a1924f8874b155ee001ade1ef16a147598b20b49
|
| 3 |
+
size 3712145082
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1ff264f99d31b522cc7e2a4eac9d38606d0c58a34c0adc74d71e0ca8b371dc36
|
| 3 |
+
size 14244
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbb670a2619e1344c7a72f15164db3e5988e7cbe56d3bedb9568b7f4ed01a9e1
|
| 3 |
+
size 1064
|
trainer_state.json
ADDED
|
@@ -0,0 +1,3191 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.15151515151515152,
|
| 5 |
+
"eval_steps": 50000,
|
| 6 |
+
"global_step": 90000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0003367003367003367,
|
| 13 |
+
"grad_norm": 0.3848826289176941,
|
| 14 |
+
"learning_rate": 4.998316498316499e-05,
|
| 15 |
+
"loss": 0.483,
|
| 16 |
+
"step": 200
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0006734006734006734,
|
| 20 |
+
"grad_norm": 0.28108373284339905,
|
| 21 |
+
"learning_rate": 4.9966329966329964e-05,
|
| 22 |
+
"loss": 0.4907,
|
| 23 |
+
"step": 400
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.00101010101010101,
|
| 27 |
+
"grad_norm": 1.1632909774780273,
|
| 28 |
+
"learning_rate": 4.994949494949495e-05,
|
| 29 |
+
"loss": 0.4656,
|
| 30 |
+
"step": 600
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0013468013468013469,
|
| 34 |
+
"grad_norm": 0.38611674308776855,
|
| 35 |
+
"learning_rate": 4.993265993265993e-05,
|
| 36 |
+
"loss": 0.4801,
|
| 37 |
+
"step": 800
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0016835016835016834,
|
| 41 |
+
"grad_norm": 0.44287604093551636,
|
| 42 |
+
"learning_rate": 4.991582491582492e-05,
|
| 43 |
+
"loss": 0.5212,
|
| 44 |
+
"step": 1000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.00202020202020202,
|
| 48 |
+
"grad_norm": 0.3867124617099762,
|
| 49 |
+
"learning_rate": 4.9899074074074075e-05,
|
| 50 |
+
"loss": 0.4678,
|
| 51 |
+
"step": 1200
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0023569023569023568,
|
| 55 |
+
"grad_norm": 0.5485501885414124,
|
| 56 |
+
"learning_rate": 4.988223905723906e-05,
|
| 57 |
+
"loss": 0.4865,
|
| 58 |
+
"step": 1400
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.0026936026936026937,
|
| 62 |
+
"grad_norm": 0.20481592416763306,
|
| 63 |
+
"learning_rate": 4.986548821548822e-05,
|
| 64 |
+
"loss": 0.5055,
|
| 65 |
+
"step": 1600
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.0030303030303030303,
|
| 69 |
+
"grad_norm": 0.2336668223142624,
|
| 70 |
+
"learning_rate": 4.98486531986532e-05,
|
| 71 |
+
"loss": 0.4695,
|
| 72 |
+
"step": 1800
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.003367003367003367,
|
| 76 |
+
"grad_norm": 0.5381026268005371,
|
| 77 |
+
"learning_rate": 4.9831818181818186e-05,
|
| 78 |
+
"loss": 0.4399,
|
| 79 |
+
"step": 2000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.003703703703703704,
|
| 83 |
+
"grad_norm": 0.7313366532325745,
|
| 84 |
+
"learning_rate": 4.981498316498317e-05,
|
| 85 |
+
"loss": 0.4459,
|
| 86 |
+
"step": 2200
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.00404040404040404,
|
| 90 |
+
"grad_norm": 0.36844003200531006,
|
| 91 |
+
"learning_rate": 4.979814814814815e-05,
|
| 92 |
+
"loss": 0.4606,
|
| 93 |
+
"step": 2400
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.004377104377104377,
|
| 97 |
+
"grad_norm": 0.3787059187889099,
|
| 98 |
+
"learning_rate": 4.978131313131313e-05,
|
| 99 |
+
"loss": 0.5135,
|
| 100 |
+
"step": 2600
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.0047138047138047135,
|
| 104 |
+
"grad_norm": 0.1680217981338501,
|
| 105 |
+
"learning_rate": 4.9764478114478116e-05,
|
| 106 |
+
"loss": 0.4913,
|
| 107 |
+
"step": 2800
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.005050505050505051,
|
| 111 |
+
"grad_norm": 0.2730850875377655,
|
| 112 |
+
"learning_rate": 4.97476430976431e-05,
|
| 113 |
+
"loss": 0.4262,
|
| 114 |
+
"step": 3000
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.0053872053872053875,
|
| 118 |
+
"grad_norm": 0.9977230429649353,
|
| 119 |
+
"learning_rate": 4.9730808080808085e-05,
|
| 120 |
+
"loss": 0.4774,
|
| 121 |
+
"step": 3200
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.005723905723905724,
|
| 125 |
+
"grad_norm": 0.38187840580940247,
|
| 126 |
+
"learning_rate": 4.971397306397307e-05,
|
| 127 |
+
"loss": 0.4642,
|
| 128 |
+
"step": 3400
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.006060606060606061,
|
| 132 |
+
"grad_norm": 0.33152151107788086,
|
| 133 |
+
"learning_rate": 4.969713804713805e-05,
|
| 134 |
+
"loss": 0.4542,
|
| 135 |
+
"step": 3600
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.006397306397306397,
|
| 139 |
+
"grad_norm": 0.4819263815879822,
|
| 140 |
+
"learning_rate": 4.968030303030303e-05,
|
| 141 |
+
"loss": 0.456,
|
| 142 |
+
"step": 3800
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.006734006734006734,
|
| 146 |
+
"grad_norm": 0.7936732172966003,
|
| 147 |
+
"learning_rate": 4.9663468013468016e-05,
|
| 148 |
+
"loss": 0.5043,
|
| 149 |
+
"step": 4000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.007070707070707071,
|
| 153 |
+
"grad_norm": 0.15946243703365326,
|
| 154 |
+
"learning_rate": 4.9646632996632993e-05,
|
| 155 |
+
"loss": 0.4722,
|
| 156 |
+
"step": 4200
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.007407407407407408,
|
| 160 |
+
"grad_norm": 0.32496747374534607,
|
| 161 |
+
"learning_rate": 4.9629797979797985e-05,
|
| 162 |
+
"loss": 0.5135,
|
| 163 |
+
"step": 4400
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.007744107744107744,
|
| 167 |
+
"grad_norm": 0.4165472388267517,
|
| 168 |
+
"learning_rate": 4.961304713804714e-05,
|
| 169 |
+
"loss": 0.5593,
|
| 170 |
+
"step": 4600
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.00808080808080808,
|
| 174 |
+
"grad_norm": 0.25771287083625793,
|
| 175 |
+
"learning_rate": 4.959621212121212e-05,
|
| 176 |
+
"loss": 0.4484,
|
| 177 |
+
"step": 4800
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.008417508417508417,
|
| 181 |
+
"grad_norm": 0.3406078815460205,
|
| 182 |
+
"learning_rate": 4.9579377104377104e-05,
|
| 183 |
+
"loss": 0.4823,
|
| 184 |
+
"step": 5000
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.008754208754208754,
|
| 188 |
+
"grad_norm": 8.295658111572266,
|
| 189 |
+
"learning_rate": 4.956254208754209e-05,
|
| 190 |
+
"loss": 0.4635,
|
| 191 |
+
"step": 5200
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.00909090909090909,
|
| 195 |
+
"grad_norm": 0.44405078887939453,
|
| 196 |
+
"learning_rate": 4.954570707070707e-05,
|
| 197 |
+
"loss": 0.4985,
|
| 198 |
+
"step": 5400
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.009427609427609427,
|
| 202 |
+
"grad_norm": 0.4546993374824524,
|
| 203 |
+
"learning_rate": 4.952887205387206e-05,
|
| 204 |
+
"loss": 0.5113,
|
| 205 |
+
"step": 5600
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.009764309764309764,
|
| 209 |
+
"grad_norm": 0.5952478647232056,
|
| 210 |
+
"learning_rate": 4.951203703703704e-05,
|
| 211 |
+
"loss": 0.514,
|
| 212 |
+
"step": 5800
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.010101010101010102,
|
| 216 |
+
"grad_norm": 0.27716225385665894,
|
| 217 |
+
"learning_rate": 4.9495202020202026e-05,
|
| 218 |
+
"loss": 0.5375,
|
| 219 |
+
"step": 6000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.010437710437710438,
|
| 223 |
+
"grad_norm": 0.7038645148277283,
|
| 224 |
+
"learning_rate": 4.9478367003367004e-05,
|
| 225 |
+
"loss": 0.4905,
|
| 226 |
+
"step": 6200
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.010774410774410775,
|
| 230 |
+
"grad_norm": 0.48266687989234924,
|
| 231 |
+
"learning_rate": 4.946153198653199e-05,
|
| 232 |
+
"loss": 0.4924,
|
| 233 |
+
"step": 6400
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.011111111111111112,
|
| 237 |
+
"grad_norm": 0.5230734944343567,
|
| 238 |
+
"learning_rate": 4.944469696969697e-05,
|
| 239 |
+
"loss": 0.533,
|
| 240 |
+
"step": 6600
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.011447811447811448,
|
| 244 |
+
"grad_norm": 1.7186241149902344,
|
| 245 |
+
"learning_rate": 4.942794612794613e-05,
|
| 246 |
+
"loss": 0.4852,
|
| 247 |
+
"step": 6800
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.011784511784511785,
|
| 251 |
+
"grad_norm": 0.5641151666641235,
|
| 252 |
+
"learning_rate": 4.9411111111111114e-05,
|
| 253 |
+
"loss": 0.492,
|
| 254 |
+
"step": 7000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.012121212121212121,
|
| 258 |
+
"grad_norm": 0.18955977261066437,
|
| 259 |
+
"learning_rate": 4.93942760942761e-05,
|
| 260 |
+
"loss": 0.4994,
|
| 261 |
+
"step": 7200
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.012457912457912458,
|
| 265 |
+
"grad_norm": 0.4500541687011719,
|
| 266 |
+
"learning_rate": 4.9377441077441076e-05,
|
| 267 |
+
"loss": 0.4761,
|
| 268 |
+
"step": 7400
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.012794612794612794,
|
| 272 |
+
"grad_norm": 0.5736501812934875,
|
| 273 |
+
"learning_rate": 4.936060606060606e-05,
|
| 274 |
+
"loss": 0.4835,
|
| 275 |
+
"step": 7600
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.013131313131313131,
|
| 279 |
+
"grad_norm": 0.12210117280483246,
|
| 280 |
+
"learning_rate": 4.9343771043771045e-05,
|
| 281 |
+
"loss": 0.4753,
|
| 282 |
+
"step": 7800
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.013468013468013467,
|
| 286 |
+
"grad_norm": 0.6027535796165466,
|
| 287 |
+
"learning_rate": 4.932693602693603e-05,
|
| 288 |
+
"loss": 0.442,
|
| 289 |
+
"step": 8000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.013804713804713804,
|
| 293 |
+
"grad_norm": 0.6002740859985352,
|
| 294 |
+
"learning_rate": 4.9310101010101014e-05,
|
| 295 |
+
"loss": 0.4668,
|
| 296 |
+
"step": 8200
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.014141414141414142,
|
| 300 |
+
"grad_norm": 0.45654693245887756,
|
| 301 |
+
"learning_rate": 4.9293265993266e-05,
|
| 302 |
+
"loss": 0.4792,
|
| 303 |
+
"step": 8400
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.014478114478114479,
|
| 307 |
+
"grad_norm": 0.5119714736938477,
|
| 308 |
+
"learning_rate": 4.9276430976430976e-05,
|
| 309 |
+
"loss": 0.4767,
|
| 310 |
+
"step": 8600
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.014814814814814815,
|
| 314 |
+
"grad_norm": 0.22278934717178345,
|
| 315 |
+
"learning_rate": 4.925959595959596e-05,
|
| 316 |
+
"loss": 0.4826,
|
| 317 |
+
"step": 8800
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.015151515151515152,
|
| 321 |
+
"grad_norm": 0.265720933675766,
|
| 322 |
+
"learning_rate": 4.9242845117845124e-05,
|
| 323 |
+
"loss": 0.5267,
|
| 324 |
+
"step": 9000
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.015488215488215488,
|
| 328 |
+
"grad_norm": 0.4445594847202301,
|
| 329 |
+
"learning_rate": 4.922601010101011e-05,
|
| 330 |
+
"loss": 0.507,
|
| 331 |
+
"step": 9200
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.015824915824915825,
|
| 335 |
+
"grad_norm": 0.6113983988761902,
|
| 336 |
+
"learning_rate": 4.9209175084175086e-05,
|
| 337 |
+
"loss": 0.5192,
|
| 338 |
+
"step": 9400
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.01616161616161616,
|
| 342 |
+
"grad_norm": 0.2967589497566223,
|
| 343 |
+
"learning_rate": 4.919234006734007e-05,
|
| 344 |
+
"loss": 0.4728,
|
| 345 |
+
"step": 9600
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.016498316498316498,
|
| 349 |
+
"grad_norm": 1.5599192380905151,
|
| 350 |
+
"learning_rate": 4.9175505050505055e-05,
|
| 351 |
+
"loss": 0.5145,
|
| 352 |
+
"step": 9800
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.016835016835016835,
|
| 356 |
+
"grad_norm": 0.6804638504981995,
|
| 357 |
+
"learning_rate": 4.915867003367003e-05,
|
| 358 |
+
"loss": 0.4293,
|
| 359 |
+
"step": 10000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.01717171717171717,
|
| 363 |
+
"grad_norm": 0.5212819576263428,
|
| 364 |
+
"learning_rate": 4.914183501683502e-05,
|
| 365 |
+
"loss": 0.5484,
|
| 366 |
+
"step": 10200
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.017508417508417508,
|
| 370 |
+
"grad_norm": 0.5872311592102051,
|
| 371 |
+
"learning_rate": 4.9125e-05,
|
| 372 |
+
"loss": 0.502,
|
| 373 |
+
"step": 10400
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.017845117845117844,
|
| 377 |
+
"grad_norm": 0.2819989323616028,
|
| 378 |
+
"learning_rate": 4.9108164983164986e-05,
|
| 379 |
+
"loss": 0.474,
|
| 380 |
+
"step": 10600
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.01818181818181818,
|
| 384 |
+
"grad_norm": 0.2001451998949051,
|
| 385 |
+
"learning_rate": 4.909132996632997e-05,
|
| 386 |
+
"loss": 0.4348,
|
| 387 |
+
"step": 10800
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.018518518518518517,
|
| 391 |
+
"grad_norm": 2.0528833866119385,
|
| 392 |
+
"learning_rate": 4.9074494949494955e-05,
|
| 393 |
+
"loss": 0.4569,
|
| 394 |
+
"step": 11000
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.018855218855218854,
|
| 398 |
+
"grad_norm": 0.7602177858352661,
|
| 399 |
+
"learning_rate": 4.905765993265993e-05,
|
| 400 |
+
"loss": 0.4319,
|
| 401 |
+
"step": 11200
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.01919191919191919,
|
| 405 |
+
"grad_norm": 0.2091585099697113,
|
| 406 |
+
"learning_rate": 4.904090909090909e-05,
|
| 407 |
+
"loss": 0.4193,
|
| 408 |
+
"step": 11400
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.019528619528619527,
|
| 412 |
+
"grad_norm": 0.20664581656455994,
|
| 413 |
+
"learning_rate": 4.902407407407408e-05,
|
| 414 |
+
"loss": 0.493,
|
| 415 |
+
"step": 11600
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.019865319865319864,
|
| 419 |
+
"grad_norm": 0.40701425075531006,
|
| 420 |
+
"learning_rate": 4.900723905723906e-05,
|
| 421 |
+
"loss": 0.5205,
|
| 422 |
+
"step": 11800
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.020202020202020204,
|
| 426 |
+
"grad_norm": 1.0059823989868164,
|
| 427 |
+
"learning_rate": 4.899040404040404e-05,
|
| 428 |
+
"loss": 0.4822,
|
| 429 |
+
"step": 12000
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.02053872053872054,
|
| 433 |
+
"grad_norm": 0.2437160164117813,
|
| 434 |
+
"learning_rate": 4.897356902356903e-05,
|
| 435 |
+
"loss": 0.4796,
|
| 436 |
+
"step": 12200
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.020875420875420877,
|
| 440 |
+
"grad_norm": 0.3996870219707489,
|
| 441 |
+
"learning_rate": 4.8956734006734005e-05,
|
| 442 |
+
"loss": 0.4243,
|
| 443 |
+
"step": 12400
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.021212121212121213,
|
| 447 |
+
"grad_norm": 1.2844949960708618,
|
| 448 |
+
"learning_rate": 4.893989898989899e-05,
|
| 449 |
+
"loss": 0.4628,
|
| 450 |
+
"step": 12600
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.02154882154882155,
|
| 454 |
+
"grad_norm": 0.5382771492004395,
|
| 455 |
+
"learning_rate": 4.8923063973063974e-05,
|
| 456 |
+
"loss": 0.5122,
|
| 457 |
+
"step": 12800
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.021885521885521887,
|
| 461 |
+
"grad_norm": 1.4352107048034668,
|
| 462 |
+
"learning_rate": 4.890622895622896e-05,
|
| 463 |
+
"loss": 0.4959,
|
| 464 |
+
"step": 13000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.022222222222222223,
|
| 468 |
+
"grad_norm": 0.3689097464084625,
|
| 469 |
+
"learning_rate": 4.888939393939394e-05,
|
| 470 |
+
"loss": 0.4711,
|
| 471 |
+
"step": 13200
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.02255892255892256,
|
| 475 |
+
"grad_norm": 0.13188982009887695,
|
| 476 |
+
"learning_rate": 4.887255892255893e-05,
|
| 477 |
+
"loss": 0.4761,
|
| 478 |
+
"step": 13400
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.022895622895622896,
|
| 482 |
+
"grad_norm": 0.18019753694534302,
|
| 483 |
+
"learning_rate": 4.885572390572391e-05,
|
| 484 |
+
"loss": 0.4604,
|
| 485 |
+
"step": 13600
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.023232323232323233,
|
| 489 |
+
"grad_norm": 0.4621998071670532,
|
| 490 |
+
"learning_rate": 4.883888888888889e-05,
|
| 491 |
+
"loss": 0.5366,
|
| 492 |
+
"step": 13800
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.02356902356902357,
|
| 496 |
+
"grad_norm": 0.21382929384708405,
|
| 497 |
+
"learning_rate": 4.882205387205387e-05,
|
| 498 |
+
"loss": 0.4864,
|
| 499 |
+
"step": 14000
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.023905723905723906,
|
| 503 |
+
"grad_norm": 0.2051325887441635,
|
| 504 |
+
"learning_rate": 4.880521885521886e-05,
|
| 505 |
+
"loss": 0.458,
|
| 506 |
+
"step": 14200
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.024242424242424242,
|
| 510 |
+
"grad_norm": 0.3620564341545105,
|
| 511 |
+
"learning_rate": 4.8788468013468015e-05,
|
| 512 |
+
"loss": 0.499,
|
| 513 |
+
"step": 14400
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.02457912457912458,
|
| 517 |
+
"grad_norm": 0.9814438223838806,
|
| 518 |
+
"learning_rate": 4.8771632996633e-05,
|
| 519 |
+
"loss": 0.442,
|
| 520 |
+
"step": 14600
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.024915824915824916,
|
| 524 |
+
"grad_norm": 0.46618032455444336,
|
| 525 |
+
"learning_rate": 4.8754797979797984e-05,
|
| 526 |
+
"loss": 0.4635,
|
| 527 |
+
"step": 14800
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.025252525252525252,
|
| 531 |
+
"grad_norm": 0.2610645592212677,
|
| 532 |
+
"learning_rate": 4.873796296296296e-05,
|
| 533 |
+
"loss": 0.458,
|
| 534 |
+
"step": 15000
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.02558922558922559,
|
| 538 |
+
"grad_norm": 0.3188152611255646,
|
| 539 |
+
"learning_rate": 4.8721127946127946e-05,
|
| 540 |
+
"loss": 0.4726,
|
| 541 |
+
"step": 15200
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.025925925925925925,
|
| 545 |
+
"grad_norm": 0.3566981852054596,
|
| 546 |
+
"learning_rate": 4.870429292929293e-05,
|
| 547 |
+
"loss": 0.4847,
|
| 548 |
+
"step": 15400
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.026262626262626262,
|
| 552 |
+
"grad_norm": 1.0731638669967651,
|
| 553 |
+
"learning_rate": 4.8687457912457914e-05,
|
| 554 |
+
"loss": 0.4912,
|
| 555 |
+
"step": 15600
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.0265993265993266,
|
| 559 |
+
"grad_norm": 0.4343542456626892,
|
| 560 |
+
"learning_rate": 4.86706228956229e-05,
|
| 561 |
+
"loss": 0.471,
|
| 562 |
+
"step": 15800
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.026936026936026935,
|
| 566 |
+
"grad_norm": 0.37956860661506653,
|
| 567 |
+
"learning_rate": 4.865378787878788e-05,
|
| 568 |
+
"loss": 0.4797,
|
| 569 |
+
"step": 16000
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.02727272727272727,
|
| 573 |
+
"grad_norm": 0.49000558257102966,
|
| 574 |
+
"learning_rate": 4.863695286195287e-05,
|
| 575 |
+
"loss": 0.5084,
|
| 576 |
+
"step": 16200
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.027609427609427608,
|
| 580 |
+
"grad_norm": 0.28972625732421875,
|
| 581 |
+
"learning_rate": 4.8620117845117845e-05,
|
| 582 |
+
"loss": 0.493,
|
| 583 |
+
"step": 16400
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.027946127946127945,
|
| 587 |
+
"grad_norm": 0.5928806662559509,
|
| 588 |
+
"learning_rate": 4.860328282828283e-05,
|
| 589 |
+
"loss": 0.5001,
|
| 590 |
+
"step": 16600
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.028282828282828285,
|
| 594 |
+
"grad_norm": 0.4121922552585602,
|
| 595 |
+
"learning_rate": 4.8586447811447814e-05,
|
| 596 |
+
"loss": 0.448,
|
| 597 |
+
"step": 16800
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.02861952861952862,
|
| 601 |
+
"grad_norm": 0.3214101195335388,
|
| 602 |
+
"learning_rate": 4.85696127946128e-05,
|
| 603 |
+
"loss": 0.5098,
|
| 604 |
+
"step": 17000
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.028956228956228958,
|
| 608 |
+
"grad_norm": 2.317594289779663,
|
| 609 |
+
"learning_rate": 4.855277777777778e-05,
|
| 610 |
+
"loss": 0.4628,
|
| 611 |
+
"step": 17200
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.029292929292929294,
|
| 615 |
+
"grad_norm": 0.46101972460746765,
|
| 616 |
+
"learning_rate": 4.853594276094277e-05,
|
| 617 |
+
"loss": 0.4556,
|
| 618 |
+
"step": 17400
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.02962962962962963,
|
| 622 |
+
"grad_norm": 0.24499452114105225,
|
| 623 |
+
"learning_rate": 4.8519107744107745e-05,
|
| 624 |
+
"loss": 0.4103,
|
| 625 |
+
"step": 17600
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.029966329966329967,
|
| 629 |
+
"grad_norm": 0.21861068904399872,
|
| 630 |
+
"learning_rate": 4.850227272727273e-05,
|
| 631 |
+
"loss": 0.4887,
|
| 632 |
+
"step": 17800
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.030303030303030304,
|
| 636 |
+
"grad_norm": 0.6664220094680786,
|
| 637 |
+
"learning_rate": 4.848543771043771e-05,
|
| 638 |
+
"loss": 0.4822,
|
| 639 |
+
"step": 18000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.03063973063973064,
|
| 643 |
+
"grad_norm": 0.5134005546569824,
|
| 644 |
+
"learning_rate": 4.846860269360269e-05,
|
| 645 |
+
"loss": 0.4432,
|
| 646 |
+
"step": 18200
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.030976430976430977,
|
| 650 |
+
"grad_norm": 0.31726887822151184,
|
| 651 |
+
"learning_rate": 4.8451851851851855e-05,
|
| 652 |
+
"loss": 0.4757,
|
| 653 |
+
"step": 18400
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.031313131313131314,
|
| 657 |
+
"grad_norm": 0.5236911177635193,
|
| 658 |
+
"learning_rate": 4.843501683501684e-05,
|
| 659 |
+
"loss": 0.4522,
|
| 660 |
+
"step": 18600
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.03164983164983165,
|
| 664 |
+
"grad_norm": 0.359935998916626,
|
| 665 |
+
"learning_rate": 4.841818181818182e-05,
|
| 666 |
+
"loss": 0.4901,
|
| 667 |
+
"step": 18800
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.03198653198653199,
|
| 671 |
+
"grad_norm": 0.5292563438415527,
|
| 672 |
+
"learning_rate": 4.84013468013468e-05,
|
| 673 |
+
"loss": 0.4962,
|
| 674 |
+
"step": 19000
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.03232323232323232,
|
| 678 |
+
"grad_norm": 0.5163784623146057,
|
| 679 |
+
"learning_rate": 4.8384511784511786e-05,
|
| 680 |
+
"loss": 0.4427,
|
| 681 |
+
"step": 19200
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.03265993265993266,
|
| 685 |
+
"grad_norm": 0.19916895031929016,
|
| 686 |
+
"learning_rate": 4.836767676767677e-05,
|
| 687 |
+
"loss": 0.4778,
|
| 688 |
+
"step": 19400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.032996632996632996,
|
| 692 |
+
"grad_norm": 0.16879796981811523,
|
| 693 |
+
"learning_rate": 4.8350841750841755e-05,
|
| 694 |
+
"loss": 0.4561,
|
| 695 |
+
"step": 19600
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.03333333333333333,
|
| 699 |
+
"grad_norm": 0.40591439604759216,
|
| 700 |
+
"learning_rate": 4.833400673400674e-05,
|
| 701 |
+
"loss": 0.532,
|
| 702 |
+
"step": 19800
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.03367003367003367,
|
| 706 |
+
"grad_norm": 0.27528542280197144,
|
| 707 |
+
"learning_rate": 4.8317171717171723e-05,
|
| 708 |
+
"loss": 0.5181,
|
| 709 |
+
"step": 20000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.034006734006734006,
|
| 713 |
+
"grad_norm": 0.46540895104408264,
|
| 714 |
+
"learning_rate": 4.83003367003367e-05,
|
| 715 |
+
"loss": 0.5005,
|
| 716 |
+
"step": 20200
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.03434343434343434,
|
| 720 |
+
"grad_norm": 0.4676566421985626,
|
| 721 |
+
"learning_rate": 4.8283501683501685e-05,
|
| 722 |
+
"loss": 0.4752,
|
| 723 |
+
"step": 20400
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.03468013468013468,
|
| 727 |
+
"grad_norm": 0.5396921038627625,
|
| 728 |
+
"learning_rate": 4.826666666666667e-05,
|
| 729 |
+
"loss": 0.4566,
|
| 730 |
+
"step": 20600
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.035016835016835016,
|
| 734 |
+
"grad_norm": 0.1875556856393814,
|
| 735 |
+
"learning_rate": 4.824983164983165e-05,
|
| 736 |
+
"loss": 0.4705,
|
| 737 |
+
"step": 20800
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.03535353535353535,
|
| 741 |
+
"grad_norm": 0.5470389723777771,
|
| 742 |
+
"learning_rate": 4.823299663299664e-05,
|
| 743 |
+
"loss": 0.5035,
|
| 744 |
+
"step": 21000
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.03569023569023569,
|
| 748 |
+
"grad_norm": 0.2772787809371948,
|
| 749 |
+
"learning_rate": 4.821616161616162e-05,
|
| 750 |
+
"loss": 0.4857,
|
| 751 |
+
"step": 21200
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.036026936026936025,
|
| 755 |
+
"grad_norm": 0.43938860297203064,
|
| 756 |
+
"learning_rate": 4.81993265993266e-05,
|
| 757 |
+
"loss": 0.5107,
|
| 758 |
+
"step": 21400
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.03636363636363636,
|
| 762 |
+
"grad_norm": 0.2839397192001343,
|
| 763 |
+
"learning_rate": 4.818257575757576e-05,
|
| 764 |
+
"loss": 0.451,
|
| 765 |
+
"step": 21600
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.0367003367003367,
|
| 769 |
+
"grad_norm": 0.46151599287986755,
|
| 770 |
+
"learning_rate": 4.816574074074074e-05,
|
| 771 |
+
"loss": 0.4984,
|
| 772 |
+
"step": 21800
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.037037037037037035,
|
| 776 |
+
"grad_norm": 0.4271756410598755,
|
| 777 |
+
"learning_rate": 4.814890572390573e-05,
|
| 778 |
+
"loss": 0.4462,
|
| 779 |
+
"step": 22000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.03737373737373737,
|
| 783 |
+
"grad_norm": 0.20119212567806244,
|
| 784 |
+
"learning_rate": 4.813207070707071e-05,
|
| 785 |
+
"loss": 0.4914,
|
| 786 |
+
"step": 22200
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.03771043771043771,
|
| 790 |
+
"grad_norm": 0.7174796462059021,
|
| 791 |
+
"learning_rate": 4.8115235690235696e-05,
|
| 792 |
+
"loss": 0.4468,
|
| 793 |
+
"step": 22400
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.038047138047138045,
|
| 797 |
+
"grad_norm": 0.25557178258895874,
|
| 798 |
+
"learning_rate": 4.809840067340067e-05,
|
| 799 |
+
"loss": 0.4969,
|
| 800 |
+
"step": 22600
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.03838383838383838,
|
| 804 |
+
"grad_norm": 0.2094777226448059,
|
| 805 |
+
"learning_rate": 4.808156565656566e-05,
|
| 806 |
+
"loss": 0.4552,
|
| 807 |
+
"step": 22800
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.03872053872053872,
|
| 811 |
+
"grad_norm": 0.2642809748649597,
|
| 812 |
+
"learning_rate": 4.806473063973064e-05,
|
| 813 |
+
"loss": 0.4741,
|
| 814 |
+
"step": 23000
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.039057239057239054,
|
| 818 |
+
"grad_norm": 0.9128819704055786,
|
| 819 |
+
"learning_rate": 4.8047895622895626e-05,
|
| 820 |
+
"loss": 0.4964,
|
| 821 |
+
"step": 23200
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.03939393939393939,
|
| 825 |
+
"grad_norm": 0.6326189637184143,
|
| 826 |
+
"learning_rate": 4.8031060606060604e-05,
|
| 827 |
+
"loss": 0.4749,
|
| 828 |
+
"step": 23400
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.03973063973063973,
|
| 832 |
+
"grad_norm": 0.6259990334510803,
|
| 833 |
+
"learning_rate": 4.8014225589225595e-05,
|
| 834 |
+
"loss": 0.4915,
|
| 835 |
+
"step": 23600
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.040067340067340064,
|
| 839 |
+
"grad_norm": 0.22890082001686096,
|
| 840 |
+
"learning_rate": 4.799739057239058e-05,
|
| 841 |
+
"loss": 0.4941,
|
| 842 |
+
"step": 23800
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.04040404040404041,
|
| 846 |
+
"grad_norm": 0.28990963101387024,
|
| 847 |
+
"learning_rate": 4.798063973063973e-05,
|
| 848 |
+
"loss": 0.4874,
|
| 849 |
+
"step": 24000
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.040740740740740744,
|
| 853 |
+
"grad_norm": 0.21455387771129608,
|
| 854 |
+
"learning_rate": 4.7963804713804715e-05,
|
| 855 |
+
"loss": 0.5352,
|
| 856 |
+
"step": 24200
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.04107744107744108,
|
| 860 |
+
"grad_norm": 0.17958062887191772,
|
| 861 |
+
"learning_rate": 4.79469696969697e-05,
|
| 862 |
+
"loss": 0.421,
|
| 863 |
+
"step": 24400
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.04141414141414142,
|
| 867 |
+
"grad_norm": 0.36474618315696716,
|
| 868 |
+
"learning_rate": 4.793013468013468e-05,
|
| 869 |
+
"loss": 0.4513,
|
| 870 |
+
"step": 24600
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.041750841750841754,
|
| 874 |
+
"grad_norm": 0.4638340175151825,
|
| 875 |
+
"learning_rate": 4.791329966329967e-05,
|
| 876 |
+
"loss": 0.5075,
|
| 877 |
+
"step": 24800
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.04208754208754209,
|
| 881 |
+
"grad_norm": 0.2026415318250656,
|
| 882 |
+
"learning_rate": 4.789646464646465e-05,
|
| 883 |
+
"loss": 0.4807,
|
| 884 |
+
"step": 25000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.04242424242424243,
|
| 888 |
+
"grad_norm": 0.3414445221424103,
|
| 889 |
+
"learning_rate": 4.787962962962963e-05,
|
| 890 |
+
"loss": 0.4731,
|
| 891 |
+
"step": 25200
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.04276094276094276,
|
| 895 |
+
"grad_norm": 0.20735152065753937,
|
| 896 |
+
"learning_rate": 4.7862794612794614e-05,
|
| 897 |
+
"loss": 0.5116,
|
| 898 |
+
"step": 25400
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.0430976430976431,
|
| 902 |
+
"grad_norm": 0.3263112008571625,
|
| 903 |
+
"learning_rate": 4.78459595959596e-05,
|
| 904 |
+
"loss": 0.4675,
|
| 905 |
+
"step": 25600
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.043434343434343436,
|
| 909 |
+
"grad_norm": 1.1240352392196655,
|
| 910 |
+
"learning_rate": 4.7829124579124576e-05,
|
| 911 |
+
"loss": 0.4651,
|
| 912 |
+
"step": 25800
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.04377104377104377,
|
| 916 |
+
"grad_norm": 0.47903600335121155,
|
| 917 |
+
"learning_rate": 4.781228956228957e-05,
|
| 918 |
+
"loss": 0.4696,
|
| 919 |
+
"step": 26000
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.04410774410774411,
|
| 923 |
+
"grad_norm": 2.3555760383605957,
|
| 924 |
+
"learning_rate": 4.779545454545455e-05,
|
| 925 |
+
"loss": 0.4826,
|
| 926 |
+
"step": 26200
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.044444444444444446,
|
| 930 |
+
"grad_norm": 0.22116072475910187,
|
| 931 |
+
"learning_rate": 4.777861952861953e-05,
|
| 932 |
+
"loss": 0.483,
|
| 933 |
+
"step": 26400
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.04478114478114478,
|
| 937 |
+
"grad_norm": 0.7030754685401917,
|
| 938 |
+
"learning_rate": 4.7761784511784514e-05,
|
| 939 |
+
"loss": 0.4317,
|
| 940 |
+
"step": 26600
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.04511784511784512,
|
| 944 |
+
"grad_norm": 0.3827281594276428,
|
| 945 |
+
"learning_rate": 4.774503367003367e-05,
|
| 946 |
+
"loss": 0.4292,
|
| 947 |
+
"step": 26800
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.045454545454545456,
|
| 951 |
+
"grad_norm": 0.8839394450187683,
|
| 952 |
+
"learning_rate": 4.7728198653198655e-05,
|
| 953 |
+
"loss": 0.4646,
|
| 954 |
+
"step": 27000
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.04579124579124579,
|
| 958 |
+
"grad_norm": 0.5600021481513977,
|
| 959 |
+
"learning_rate": 4.771136363636364e-05,
|
| 960 |
+
"loss": 0.5081,
|
| 961 |
+
"step": 27200
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.04612794612794613,
|
| 965 |
+
"grad_norm": 0.23018187284469604,
|
| 966 |
+
"learning_rate": 4.7694528619528624e-05,
|
| 967 |
+
"loss": 0.4908,
|
| 968 |
+
"step": 27400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.046464646464646465,
|
| 972 |
+
"grad_norm": 0.456559419631958,
|
| 973 |
+
"learning_rate": 4.767769360269361e-05,
|
| 974 |
+
"loss": 0.4872,
|
| 975 |
+
"step": 27600
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.0468013468013468,
|
| 979 |
+
"grad_norm": 0.12121502310037613,
|
| 980 |
+
"learning_rate": 4.7660858585858586e-05,
|
| 981 |
+
"loss": 0.4979,
|
| 982 |
+
"step": 27800
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.04713804713804714,
|
| 986 |
+
"grad_norm": 0.38946759700775146,
|
| 987 |
+
"learning_rate": 4.764402356902357e-05,
|
| 988 |
+
"loss": 0.4651,
|
| 989 |
+
"step": 28000
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.047474747474747475,
|
| 993 |
+
"grad_norm": 2.142538547515869,
|
| 994 |
+
"learning_rate": 4.7627188552188555e-05,
|
| 995 |
+
"loss": 0.4288,
|
| 996 |
+
"step": 28200
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.04781144781144781,
|
| 1000 |
+
"grad_norm": 0.15298590064048767,
|
| 1001 |
+
"learning_rate": 4.761035353535353e-05,
|
| 1002 |
+
"loss": 0.4365,
|
| 1003 |
+
"step": 28400
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.04814814814814815,
|
| 1007 |
+
"grad_norm": 0.6363445520401001,
|
| 1008 |
+
"learning_rate": 4.7593518518518524e-05,
|
| 1009 |
+
"loss": 0.4428,
|
| 1010 |
+
"step": 28600
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.048484848484848485,
|
| 1014 |
+
"grad_norm": 0.5169795751571655,
|
| 1015 |
+
"learning_rate": 4.757668350168351e-05,
|
| 1016 |
+
"loss": 0.4666,
|
| 1017 |
+
"step": 28800
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.04882154882154882,
|
| 1021 |
+
"grad_norm": 0.2661610245704651,
|
| 1022 |
+
"learning_rate": 4.7559848484848486e-05,
|
| 1023 |
+
"loss": 0.4492,
|
| 1024 |
+
"step": 29000
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.04915824915824916,
|
| 1028 |
+
"grad_norm": 0.4399496018886566,
|
| 1029 |
+
"learning_rate": 4.754301346801347e-05,
|
| 1030 |
+
"loss": 0.4504,
|
| 1031 |
+
"step": 29200
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.049494949494949494,
|
| 1035 |
+
"grad_norm": 0.31160300970077515,
|
| 1036 |
+
"learning_rate": 4.7526178451178454e-05,
|
| 1037 |
+
"loss": 0.4527,
|
| 1038 |
+
"step": 29400
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.04983164983164983,
|
| 1042 |
+
"grad_norm": 0.6169541478157043,
|
| 1043 |
+
"learning_rate": 4.750934343434343e-05,
|
| 1044 |
+
"loss": 0.4995,
|
| 1045 |
+
"step": 29600
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.05016835016835017,
|
| 1049 |
+
"grad_norm": 0.43009576201438904,
|
| 1050 |
+
"learning_rate": 4.7492508417508416e-05,
|
| 1051 |
+
"loss": 0.4681,
|
| 1052 |
+
"step": 29800
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.050505050505050504,
|
| 1056 |
+
"grad_norm": 0.3901723623275757,
|
| 1057 |
+
"learning_rate": 4.74756734006734e-05,
|
| 1058 |
+
"loss": 0.4509,
|
| 1059 |
+
"step": 30000
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.05084175084175084,
|
| 1063 |
+
"grad_norm": 0.2859044373035431,
|
| 1064 |
+
"learning_rate": 4.745883838383839e-05,
|
| 1065 |
+
"loss": 0.4262,
|
| 1066 |
+
"step": 30200
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.05117845117845118,
|
| 1070 |
+
"grad_norm": 0.29008200764656067,
|
| 1071 |
+
"learning_rate": 4.744208754208754e-05,
|
| 1072 |
+
"loss": 0.434,
|
| 1073 |
+
"step": 30400
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.051515151515151514,
|
| 1077 |
+
"grad_norm": 0.8030261397361755,
|
| 1078 |
+
"learning_rate": 4.742525252525253e-05,
|
| 1079 |
+
"loss": 0.5066,
|
| 1080 |
+
"step": 30600
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.05185185185185185,
|
| 1084 |
+
"grad_norm": 0.6552255153656006,
|
| 1085 |
+
"learning_rate": 4.740841750841751e-05,
|
| 1086 |
+
"loss": 0.4888,
|
| 1087 |
+
"step": 30800
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.05218855218855219,
|
| 1091 |
+
"grad_norm": 0.47479531168937683,
|
| 1092 |
+
"learning_rate": 4.7391582491582496e-05,
|
| 1093 |
+
"loss": 0.4886,
|
| 1094 |
+
"step": 31000
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.052525252525252523,
|
| 1098 |
+
"grad_norm": 0.6294977068901062,
|
| 1099 |
+
"learning_rate": 4.737474747474748e-05,
|
| 1100 |
+
"loss": 0.4144,
|
| 1101 |
+
"step": 31200
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.05286195286195286,
|
| 1105 |
+
"grad_norm": 0.346327006816864,
|
| 1106 |
+
"learning_rate": 4.7357912457912465e-05,
|
| 1107 |
+
"loss": 0.517,
|
| 1108 |
+
"step": 31400
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.0531986531986532,
|
| 1112 |
+
"grad_norm": 0.4449813961982727,
|
| 1113 |
+
"learning_rate": 4.734107744107744e-05,
|
| 1114 |
+
"loss": 0.4681,
|
| 1115 |
+
"step": 31600
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.05353535353535353,
|
| 1119 |
+
"grad_norm": 0.4756247401237488,
|
| 1120 |
+
"learning_rate": 4.7324242424242426e-05,
|
| 1121 |
+
"loss": 0.4772,
|
| 1122 |
+
"step": 31800
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.05387205387205387,
|
| 1126 |
+
"grad_norm": 0.5156170129776001,
|
| 1127 |
+
"learning_rate": 4.730740740740741e-05,
|
| 1128 |
+
"loss": 0.456,
|
| 1129 |
+
"step": 32000
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.054208754208754206,
|
| 1133 |
+
"grad_norm": 1.6009584665298462,
|
| 1134 |
+
"learning_rate": 4.729057239057239e-05,
|
| 1135 |
+
"loss": 0.4483,
|
| 1136 |
+
"step": 32200
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.05454545454545454,
|
| 1140 |
+
"grad_norm": 0.4179598391056061,
|
| 1141 |
+
"learning_rate": 4.727373737373737e-05,
|
| 1142 |
+
"loss": 0.4438,
|
| 1143 |
+
"step": 32400
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.05488215488215488,
|
| 1147 |
+
"grad_norm": 0.09701373428106308,
|
| 1148 |
+
"learning_rate": 4.725690235690236e-05,
|
| 1149 |
+
"loss": 0.4746,
|
| 1150 |
+
"step": 32600
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.055218855218855216,
|
| 1154 |
+
"grad_norm": 0.4205819368362427,
|
| 1155 |
+
"learning_rate": 4.724006734006734e-05,
|
| 1156 |
+
"loss": 0.4459,
|
| 1157 |
+
"step": 32800
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.05555555555555555,
|
| 1161 |
+
"grad_norm": 0.5670439004898071,
|
| 1162 |
+
"learning_rate": 4.7223232323232326e-05,
|
| 1163 |
+
"loss": 0.5025,
|
| 1164 |
+
"step": 33000
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.05589225589225589,
|
| 1168 |
+
"grad_norm": 0.47377878427505493,
|
| 1169 |
+
"learning_rate": 4.720639730639731e-05,
|
| 1170 |
+
"loss": 0.4746,
|
| 1171 |
+
"step": 33200
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.056228956228956226,
|
| 1175 |
+
"grad_norm": 0.254245400428772,
|
| 1176 |
+
"learning_rate": 4.718956228956229e-05,
|
| 1177 |
+
"loss": 0.4926,
|
| 1178 |
+
"step": 33400
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.05656565656565657,
|
| 1182 |
+
"grad_norm": 0.299713671207428,
|
| 1183 |
+
"learning_rate": 4.717272727272727e-05,
|
| 1184 |
+
"loss": 0.4748,
|
| 1185 |
+
"step": 33600
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.056902356902356906,
|
| 1189 |
+
"grad_norm": 0.16089321672916412,
|
| 1190 |
+
"learning_rate": 4.715589225589226e-05,
|
| 1191 |
+
"loss": 0.4813,
|
| 1192 |
+
"step": 33800
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.05723905723905724,
|
| 1196 |
+
"grad_norm": 0.27492621541023254,
|
| 1197 |
+
"learning_rate": 4.713914141414142e-05,
|
| 1198 |
+
"loss": 0.473,
|
| 1199 |
+
"step": 34000
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.05757575757575758,
|
| 1203 |
+
"grad_norm": 0.8170735239982605,
|
| 1204 |
+
"learning_rate": 4.71223063973064e-05,
|
| 1205 |
+
"loss": 0.4251,
|
| 1206 |
+
"step": 34200
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.057912457912457915,
|
| 1210 |
+
"grad_norm": 0.41072168946266174,
|
| 1211 |
+
"learning_rate": 4.710547138047138e-05,
|
| 1212 |
+
"loss": 0.4692,
|
| 1213 |
+
"step": 34400
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.05824915824915825,
|
| 1217 |
+
"grad_norm": 0.37332773208618164,
|
| 1218 |
+
"learning_rate": 4.708863636363637e-05,
|
| 1219 |
+
"loss": 0.4289,
|
| 1220 |
+
"step": 34600
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.05858585858585859,
|
| 1224 |
+
"grad_norm": 0.3257604241371155,
|
| 1225 |
+
"learning_rate": 4.7071801346801345e-05,
|
| 1226 |
+
"loss": 0.4623,
|
| 1227 |
+
"step": 34800
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.058922558922558925,
|
| 1231 |
+
"grad_norm": 0.23426009714603424,
|
| 1232 |
+
"learning_rate": 4.705496632996633e-05,
|
| 1233 |
+
"loss": 0.5082,
|
| 1234 |
+
"step": 35000
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.05925925925925926,
|
| 1238 |
+
"grad_norm": 0.28719109296798706,
|
| 1239 |
+
"learning_rate": 4.703813131313132e-05,
|
| 1240 |
+
"loss": 0.4767,
|
| 1241 |
+
"step": 35200
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.0595959595959596,
|
| 1245 |
+
"grad_norm": 0.35480618476867676,
|
| 1246 |
+
"learning_rate": 4.70212962962963e-05,
|
| 1247 |
+
"loss": 0.5006,
|
| 1248 |
+
"step": 35400
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.059932659932659935,
|
| 1252 |
+
"grad_norm": 0.5050226449966431,
|
| 1253 |
+
"learning_rate": 4.700446127946128e-05,
|
| 1254 |
+
"loss": 0.4809,
|
| 1255 |
+
"step": 35600
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.06026936026936027,
|
| 1259 |
+
"grad_norm": 0.2631937265396118,
|
| 1260 |
+
"learning_rate": 4.698762626262627e-05,
|
| 1261 |
+
"loss": 0.4508,
|
| 1262 |
+
"step": 35800
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.06060606060606061,
|
| 1266 |
+
"grad_norm": 0.32295939326286316,
|
| 1267 |
+
"learning_rate": 4.6970791245791244e-05,
|
| 1268 |
+
"loss": 0.4697,
|
| 1269 |
+
"step": 36000
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.060942760942760944,
|
| 1273 |
+
"grad_norm": 0.34667742252349854,
|
| 1274 |
+
"learning_rate": 4.695404040404041e-05,
|
| 1275 |
+
"loss": 0.4692,
|
| 1276 |
+
"step": 36200
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.06127946127946128,
|
| 1280 |
+
"grad_norm": 0.6987492442131042,
|
| 1281 |
+
"learning_rate": 4.693720538720539e-05,
|
| 1282 |
+
"loss": 0.4596,
|
| 1283 |
+
"step": 36400
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.06161616161616162,
|
| 1287 |
+
"grad_norm": 0.4795779883861542,
|
| 1288 |
+
"learning_rate": 4.692037037037037e-05,
|
| 1289 |
+
"loss": 0.4361,
|
| 1290 |
+
"step": 36600
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.061952861952861954,
|
| 1294 |
+
"grad_norm": 0.5291064381599426,
|
| 1295 |
+
"learning_rate": 4.6903535353535355e-05,
|
| 1296 |
+
"loss": 0.4408,
|
| 1297 |
+
"step": 36800
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.06228956228956229,
|
| 1301 |
+
"grad_norm": 0.46040576696395874,
|
| 1302 |
+
"learning_rate": 4.688670033670034e-05,
|
| 1303 |
+
"loss": 0.4976,
|
| 1304 |
+
"step": 37000
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.06262626262626263,
|
| 1308 |
+
"grad_norm": 2.0511856079101562,
|
| 1309 |
+
"learning_rate": 4.6869865319865324e-05,
|
| 1310 |
+
"loss": 0.4747,
|
| 1311 |
+
"step": 37200
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.06296296296296296,
|
| 1315 |
+
"grad_norm": 0.6845996975898743,
|
| 1316 |
+
"learning_rate": 4.68530303030303e-05,
|
| 1317 |
+
"loss": 0.4145,
|
| 1318 |
+
"step": 37400
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.0632996632996633,
|
| 1322 |
+
"grad_norm": 0.2830463945865631,
|
| 1323 |
+
"learning_rate": 4.6836195286195286e-05,
|
| 1324 |
+
"loss": 0.4888,
|
| 1325 |
+
"step": 37600
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.06363636363636363,
|
| 1329 |
+
"grad_norm": 0.4033803343772888,
|
| 1330 |
+
"learning_rate": 4.681936026936028e-05,
|
| 1331 |
+
"loss": 0.4584,
|
| 1332 |
+
"step": 37800
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.06397306397306397,
|
| 1336 |
+
"grad_norm": 0.26968395709991455,
|
| 1337 |
+
"learning_rate": 4.6802525252525255e-05,
|
| 1338 |
+
"loss": 0.4246,
|
| 1339 |
+
"step": 38000
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.0643097643097643,
|
| 1343 |
+
"grad_norm": 0.14037840068340302,
|
| 1344 |
+
"learning_rate": 4.678569023569024e-05,
|
| 1345 |
+
"loss": 0.4047,
|
| 1346 |
+
"step": 38200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.06464646464646465,
|
| 1350 |
+
"grad_norm": 0.2874729335308075,
|
| 1351 |
+
"learning_rate": 4.676885521885522e-05,
|
| 1352 |
+
"loss": 0.4445,
|
| 1353 |
+
"step": 38400
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.06498316498316499,
|
| 1357 |
+
"grad_norm": 0.2633935213088989,
|
| 1358 |
+
"learning_rate": 4.67520202020202e-05,
|
| 1359 |
+
"loss": 0.4435,
|
| 1360 |
+
"step": 38600
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.06531986531986532,
|
| 1364 |
+
"grad_norm": 0.4510101079940796,
|
| 1365 |
+
"learning_rate": 4.6735185185185185e-05,
|
| 1366 |
+
"loss": 0.4724,
|
| 1367 |
+
"step": 38800
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.06565656565656566,
|
| 1371 |
+
"grad_norm": 0.20095351338386536,
|
| 1372 |
+
"learning_rate": 4.671835016835017e-05,
|
| 1373 |
+
"loss": 0.5128,
|
| 1374 |
+
"step": 39000
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.06599326599326599,
|
| 1378 |
+
"grad_norm": 0.4433535635471344,
|
| 1379 |
+
"learning_rate": 4.670159932659933e-05,
|
| 1380 |
+
"loss": 0.4581,
|
| 1381 |
+
"step": 39200
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.06632996632996634,
|
| 1385 |
+
"grad_norm": 0.5821954607963562,
|
| 1386 |
+
"learning_rate": 4.668476430976431e-05,
|
| 1387 |
+
"loss": 0.4985,
|
| 1388 |
+
"step": 39400
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.06666666666666667,
|
| 1392 |
+
"grad_norm": 1.3577245473861694,
|
| 1393 |
+
"learning_rate": 4.6667929292929296e-05,
|
| 1394 |
+
"loss": 0.5283,
|
| 1395 |
+
"step": 39600
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.06700336700336701,
|
| 1399 |
+
"grad_norm": 0.37699806690216064,
|
| 1400 |
+
"learning_rate": 4.6651094276094274e-05,
|
| 1401 |
+
"loss": 0.4825,
|
| 1402 |
+
"step": 39800
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.06734006734006734,
|
| 1406 |
+
"grad_norm": 0.41804903745651245,
|
| 1407 |
+
"learning_rate": 4.663425925925926e-05,
|
| 1408 |
+
"loss": 0.4907,
|
| 1409 |
+
"step": 40000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.06767676767676768,
|
| 1413 |
+
"grad_norm": 0.243534654378891,
|
| 1414 |
+
"learning_rate": 4.661742424242425e-05,
|
| 1415 |
+
"loss": 0.5059,
|
| 1416 |
+
"step": 40200
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.06801346801346801,
|
| 1420 |
+
"grad_norm": 0.1186649277806282,
|
| 1421 |
+
"learning_rate": 4.660058922558923e-05,
|
| 1422 |
+
"loss": 0.4808,
|
| 1423 |
+
"step": 40400
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.06835016835016836,
|
| 1427 |
+
"grad_norm": 0.30161020159721375,
|
| 1428 |
+
"learning_rate": 4.658375420875421e-05,
|
| 1429 |
+
"loss": 0.4816,
|
| 1430 |
+
"step": 40600
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.06868686868686869,
|
| 1434 |
+
"grad_norm": 0.2418268918991089,
|
| 1435 |
+
"learning_rate": 4.6566919191919195e-05,
|
| 1436 |
+
"loss": 0.4371,
|
| 1437 |
+
"step": 40800
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.06902356902356903,
|
| 1441 |
+
"grad_norm": 0.2833971381187439,
|
| 1442 |
+
"learning_rate": 4.655008417508418e-05,
|
| 1443 |
+
"loss": 0.4686,
|
| 1444 |
+
"step": 41000
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.06936026936026936,
|
| 1448 |
+
"grad_norm": 0.2797035276889801,
|
| 1449 |
+
"learning_rate": 4.653324915824916e-05,
|
| 1450 |
+
"loss": 0.4599,
|
| 1451 |
+
"step": 41200
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.0696969696969697,
|
| 1455 |
+
"grad_norm": 0.3841836154460907,
|
| 1456 |
+
"learning_rate": 4.651641414141414e-05,
|
| 1457 |
+
"loss": 0.4111,
|
| 1458 |
+
"step": 41400
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.07003367003367003,
|
| 1462 |
+
"grad_norm": 0.6590111255645752,
|
| 1463 |
+
"learning_rate": 4.6499579124579126e-05,
|
| 1464 |
+
"loss": 0.4631,
|
| 1465 |
+
"step": 41600
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.07037037037037037,
|
| 1469 |
+
"grad_norm": 0.29527220129966736,
|
| 1470 |
+
"learning_rate": 4.648274410774411e-05,
|
| 1471 |
+
"loss": 0.4963,
|
| 1472 |
+
"step": 41800
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.0707070707070707,
|
| 1476 |
+
"grad_norm": 0.7178300619125366,
|
| 1477 |
+
"learning_rate": 4.6465909090909095e-05,
|
| 1478 |
+
"loss": 0.4694,
|
| 1479 |
+
"step": 42000
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.07104377104377105,
|
| 1483 |
+
"grad_norm": 0.38491058349609375,
|
| 1484 |
+
"learning_rate": 4.644907407407408e-05,
|
| 1485 |
+
"loss": 0.4437,
|
| 1486 |
+
"step": 42200
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.07138047138047138,
|
| 1490 |
+
"grad_norm": 0.3037305772304535,
|
| 1491 |
+
"learning_rate": 4.643223905723906e-05,
|
| 1492 |
+
"loss": 0.4635,
|
| 1493 |
+
"step": 42400
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.07171717171717172,
|
| 1497 |
+
"grad_norm": 0.4430043697357178,
|
| 1498 |
+
"learning_rate": 4.641540404040404e-05,
|
| 1499 |
+
"loss": 0.4623,
|
| 1500 |
+
"step": 42600
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.07205387205387205,
|
| 1504 |
+
"grad_norm": 0.3590750992298126,
|
| 1505 |
+
"learning_rate": 4.6398569023569026e-05,
|
| 1506 |
+
"loss": 0.4827,
|
| 1507 |
+
"step": 42800
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.0723905723905724,
|
| 1511 |
+
"grad_norm": 0.6161913275718689,
|
| 1512 |
+
"learning_rate": 4.638181818181818e-05,
|
| 1513 |
+
"loss": 0.4796,
|
| 1514 |
+
"step": 43000
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.07272727272727272,
|
| 1518 |
+
"grad_norm": 0.3607730269432068,
|
| 1519 |
+
"learning_rate": 4.636498316498317e-05,
|
| 1520 |
+
"loss": 0.4801,
|
| 1521 |
+
"step": 43200
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.07306397306397307,
|
| 1525 |
+
"grad_norm": 0.07056716829538345,
|
| 1526 |
+
"learning_rate": 4.634814814814815e-05,
|
| 1527 |
+
"loss": 0.4738,
|
| 1528 |
+
"step": 43400
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.0734006734006734,
|
| 1532 |
+
"grad_norm": 0.09327512234449387,
|
| 1533 |
+
"learning_rate": 4.633131313131313e-05,
|
| 1534 |
+
"loss": 0.4885,
|
| 1535 |
+
"step": 43600
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.07373737373737374,
|
| 1539 |
+
"grad_norm": 0.2519952952861786,
|
| 1540 |
+
"learning_rate": 4.6314478114478114e-05,
|
| 1541 |
+
"loss": 0.4541,
|
| 1542 |
+
"step": 43800
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.07407407407407407,
|
| 1546 |
+
"grad_norm": 0.4618964195251465,
|
| 1547 |
+
"learning_rate": 4.62976430976431e-05,
|
| 1548 |
+
"loss": 0.4593,
|
| 1549 |
+
"step": 44000
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.07441077441077441,
|
| 1553 |
+
"grad_norm": 0.4683738946914673,
|
| 1554 |
+
"learning_rate": 4.628080808080808e-05,
|
| 1555 |
+
"loss": 0.4536,
|
| 1556 |
+
"step": 44200
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.07474747474747474,
|
| 1560 |
+
"grad_norm": 0.2552854120731354,
|
| 1561 |
+
"learning_rate": 4.626397306397307e-05,
|
| 1562 |
+
"loss": 0.4702,
|
| 1563 |
+
"step": 44400
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.07508417508417509,
|
| 1567 |
+
"grad_norm": 0.33385610580444336,
|
| 1568 |
+
"learning_rate": 4.624713804713805e-05,
|
| 1569 |
+
"loss": 0.4623,
|
| 1570 |
+
"step": 44600
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.07542087542087542,
|
| 1574 |
+
"grad_norm": 0.17833998799324036,
|
| 1575 |
+
"learning_rate": 4.6230303030303036e-05,
|
| 1576 |
+
"loss": 0.419,
|
| 1577 |
+
"step": 44800
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.07575757575757576,
|
| 1581 |
+
"grad_norm": 0.14885468780994415,
|
| 1582 |
+
"learning_rate": 4.621346801346801e-05,
|
| 1583 |
+
"loss": 0.5132,
|
| 1584 |
+
"step": 45000
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.07609427609427609,
|
| 1588 |
+
"grad_norm": 0.4861992597579956,
|
| 1589 |
+
"learning_rate": 4.6196632996633e-05,
|
| 1590 |
+
"loss": 0.4914,
|
| 1591 |
+
"step": 45200
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.07643097643097643,
|
| 1595 |
+
"grad_norm": 0.20314612984657288,
|
| 1596 |
+
"learning_rate": 4.617979797979798e-05,
|
| 1597 |
+
"loss": 0.4535,
|
| 1598 |
+
"step": 45400
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.07676767676767676,
|
| 1602 |
+
"grad_norm": 0.7097423076629639,
|
| 1603 |
+
"learning_rate": 4.616304713804714e-05,
|
| 1604 |
+
"loss": 0.4343,
|
| 1605 |
+
"step": 45600
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.0771043771043771,
|
| 1609 |
+
"grad_norm": 0.23547014594078064,
|
| 1610 |
+
"learning_rate": 4.6146212121212124e-05,
|
| 1611 |
+
"loss": 0.449,
|
| 1612 |
+
"step": 45800
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.07744107744107744,
|
| 1616 |
+
"grad_norm": 0.25944817066192627,
|
| 1617 |
+
"learning_rate": 4.612937710437711e-05,
|
| 1618 |
+
"loss": 0.4535,
|
| 1619 |
+
"step": 46000
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.07777777777777778,
|
| 1623 |
+
"grad_norm": 0.6145304441452026,
|
| 1624 |
+
"learning_rate": 4.6112542087542086e-05,
|
| 1625 |
+
"loss": 0.4536,
|
| 1626 |
+
"step": 46200
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.07811447811447811,
|
| 1630 |
+
"grad_norm": 1.2527995109558105,
|
| 1631 |
+
"learning_rate": 4.609570707070707e-05,
|
| 1632 |
+
"loss": 0.4616,
|
| 1633 |
+
"step": 46400
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.07845117845117845,
|
| 1637 |
+
"grad_norm": 0.9534751772880554,
|
| 1638 |
+
"learning_rate": 4.6078872053872055e-05,
|
| 1639 |
+
"loss": 0.5009,
|
| 1640 |
+
"step": 46600
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.07878787878787878,
|
| 1644 |
+
"grad_norm": 0.552191436290741,
|
| 1645 |
+
"learning_rate": 4.606203703703704e-05,
|
| 1646 |
+
"loss": 0.4738,
|
| 1647 |
+
"step": 46800
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.07912457912457913,
|
| 1651 |
+
"grad_norm": 0.28889888525009155,
|
| 1652 |
+
"learning_rate": 4.6045202020202023e-05,
|
| 1653 |
+
"loss": 0.4721,
|
| 1654 |
+
"step": 47000
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.07946127946127945,
|
| 1658 |
+
"grad_norm": 0.4266869127750397,
|
| 1659 |
+
"learning_rate": 4.602845117845118e-05,
|
| 1660 |
+
"loss": 0.4695,
|
| 1661 |
+
"step": 47200
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.0797979797979798,
|
| 1665 |
+
"grad_norm": 0.5105581879615784,
|
| 1666 |
+
"learning_rate": 4.601161616161616e-05,
|
| 1667 |
+
"loss": 0.4739,
|
| 1668 |
+
"step": 47400
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.08013468013468013,
|
| 1672 |
+
"grad_norm": 0.4175490736961365,
|
| 1673 |
+
"learning_rate": 4.599478114478114e-05,
|
| 1674 |
+
"loss": 0.4456,
|
| 1675 |
+
"step": 47600
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.08047138047138047,
|
| 1679 |
+
"grad_norm": 0.3257778584957123,
|
| 1680 |
+
"learning_rate": 4.5977946127946134e-05,
|
| 1681 |
+
"loss": 0.4808,
|
| 1682 |
+
"step": 47800
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.08080808080808081,
|
| 1686 |
+
"grad_norm": 0.3000372648239136,
|
| 1687 |
+
"learning_rate": 4.596111111111112e-05,
|
| 1688 |
+
"loss": 0.4635,
|
| 1689 |
+
"step": 48000
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.08114478114478114,
|
| 1693 |
+
"grad_norm": 0.32268643379211426,
|
| 1694 |
+
"learning_rate": 4.5944276094276096e-05,
|
| 1695 |
+
"loss": 0.4925,
|
| 1696 |
+
"step": 48200
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.08148148148148149,
|
| 1700 |
+
"grad_norm": 0.5290645956993103,
|
| 1701 |
+
"learning_rate": 4.592744107744108e-05,
|
| 1702 |
+
"loss": 0.4711,
|
| 1703 |
+
"step": 48400
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.08181818181818182,
|
| 1707 |
+
"grad_norm": 0.29082873463630676,
|
| 1708 |
+
"learning_rate": 4.5910606060606065e-05,
|
| 1709 |
+
"loss": 0.4728,
|
| 1710 |
+
"step": 48600
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.08215488215488216,
|
| 1714 |
+
"grad_norm": 0.6704333424568176,
|
| 1715 |
+
"learning_rate": 4.589377104377104e-05,
|
| 1716 |
+
"loss": 0.4779,
|
| 1717 |
+
"step": 48800
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.08249158249158249,
|
| 1721 |
+
"grad_norm": 0.27797549962997437,
|
| 1722 |
+
"learning_rate": 4.587693602693603e-05,
|
| 1723 |
+
"loss": 0.4529,
|
| 1724 |
+
"step": 49000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.08282828282828283,
|
| 1728 |
+
"grad_norm": 0.1398361176252365,
|
| 1729 |
+
"learning_rate": 4.586010101010101e-05,
|
| 1730 |
+
"loss": 0.4379,
|
| 1731 |
+
"step": 49200
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.08316498316498316,
|
| 1735 |
+
"grad_norm": 0.31926196813583374,
|
| 1736 |
+
"learning_rate": 4.5843265993265996e-05,
|
| 1737 |
+
"loss": 0.457,
|
| 1738 |
+
"step": 49400
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.08350168350168351,
|
| 1742 |
+
"grad_norm": 0.17603324353694916,
|
| 1743 |
+
"learning_rate": 4.582643097643098e-05,
|
| 1744 |
+
"loss": 0.452,
|
| 1745 |
+
"step": 49600
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.08383838383838384,
|
| 1749 |
+
"grad_norm": 0.4734348654747009,
|
| 1750 |
+
"learning_rate": 4.5809595959595964e-05,
|
| 1751 |
+
"loss": 0.4489,
|
| 1752 |
+
"step": 49800
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.08417508417508418,
|
| 1756 |
+
"grad_norm": 0.2849540710449219,
|
| 1757 |
+
"learning_rate": 4.579276094276094e-05,
|
| 1758 |
+
"loss": 0.4448,
|
| 1759 |
+
"step": 50000
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.08417508417508418,
|
| 1763 |
+
"eval_loss": 0.4674188494682312,
|
| 1764 |
+
"eval_runtime": 400.4812,
|
| 1765 |
+
"eval_samples_per_second": 14.982,
|
| 1766 |
+
"eval_steps_per_second": 14.982,
|
| 1767 |
+
"step": 50000
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.08451178451178451,
|
| 1771 |
+
"grad_norm": 0.3861866295337677,
|
| 1772 |
+
"learning_rate": 4.5775925925925926e-05,
|
| 1773 |
+
"loss": 0.4691,
|
| 1774 |
+
"step": 50200
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.08484848484848485,
|
| 1778 |
+
"grad_norm": 0.23927472531795502,
|
| 1779 |
+
"learning_rate": 4.575909090909091e-05,
|
| 1780 |
+
"loss": 0.519,
|
| 1781 |
+
"step": 50400
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.08518518518518518,
|
| 1785 |
+
"grad_norm": 0.2663820683956146,
|
| 1786 |
+
"learning_rate": 4.5742255892255895e-05,
|
| 1787 |
+
"loss": 0.4394,
|
| 1788 |
+
"step": 50600
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.08552188552188553,
|
| 1792 |
+
"grad_norm": 0.0782080739736557,
|
| 1793 |
+
"learning_rate": 4.572542087542088e-05,
|
| 1794 |
+
"loss": 0.4181,
|
| 1795 |
+
"step": 50800
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.08585858585858586,
|
| 1799 |
+
"grad_norm": 0.32413387298583984,
|
| 1800 |
+
"learning_rate": 4.5708585858585864e-05,
|
| 1801 |
+
"loss": 0.4643,
|
| 1802 |
+
"step": 51000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.0861952861952862,
|
| 1806 |
+
"grad_norm": 0.18156534433364868,
|
| 1807 |
+
"learning_rate": 4.569175084175085e-05,
|
| 1808 |
+
"loss": 0.4511,
|
| 1809 |
+
"step": 51200
|
| 1810 |
+
},
|
| 1811 |
+
{
|
| 1812 |
+
"epoch": 0.08653198653198653,
|
| 1813 |
+
"grad_norm": 0.298673152923584,
|
| 1814 |
+
"learning_rate": 4.5674915824915826e-05,
|
| 1815 |
+
"loss": 0.4424,
|
| 1816 |
+
"step": 51400
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.08686868686868687,
|
| 1820 |
+
"grad_norm": 0.24452580511569977,
|
| 1821 |
+
"learning_rate": 4.565808080808081e-05,
|
| 1822 |
+
"loss": 0.4556,
|
| 1823 |
+
"step": 51600
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 0.0872053872053872,
|
| 1827 |
+
"grad_norm": 0.2959561049938202,
|
| 1828 |
+
"learning_rate": 4.564132996632997e-05,
|
| 1829 |
+
"loss": 0.4543,
|
| 1830 |
+
"step": 51800
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 0.08754208754208755,
|
| 1834 |
+
"grad_norm": 0.6213822960853577,
|
| 1835 |
+
"learning_rate": 4.5624579124579125e-05,
|
| 1836 |
+
"loss": 0.5136,
|
| 1837 |
+
"step": 52000
|
| 1838 |
+
},
|
| 1839 |
+
{
|
| 1840 |
+
"epoch": 0.08787878787878788,
|
| 1841 |
+
"grad_norm": 0.5385012030601501,
|
| 1842 |
+
"learning_rate": 4.560774410774411e-05,
|
| 1843 |
+
"loss": 0.4602,
|
| 1844 |
+
"step": 52200
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.08821548821548822,
|
| 1848 |
+
"grad_norm": 0.6063356995582581,
|
| 1849 |
+
"learning_rate": 4.5590909090909094e-05,
|
| 1850 |
+
"loss": 0.4367,
|
| 1851 |
+
"step": 52400
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 0.08855218855218855,
|
| 1855 |
+
"grad_norm": 0.43720120191574097,
|
| 1856 |
+
"learning_rate": 4.557407407407407e-05,
|
| 1857 |
+
"loss": 0.4902,
|
| 1858 |
+
"step": 52600
|
| 1859 |
+
},
|
| 1860 |
+
{
|
| 1861 |
+
"epoch": 0.08888888888888889,
|
| 1862 |
+
"grad_norm": 0.4334559738636017,
|
| 1863 |
+
"learning_rate": 4.555723905723906e-05,
|
| 1864 |
+
"loss": 0.4585,
|
| 1865 |
+
"step": 52800
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 0.08922558922558922,
|
| 1869 |
+
"grad_norm": 0.2874049246311188,
|
| 1870 |
+
"learning_rate": 4.554040404040405e-05,
|
| 1871 |
+
"loss": 0.4907,
|
| 1872 |
+
"step": 53000
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 0.08956228956228957,
|
| 1876 |
+
"grad_norm": 0.19016990065574646,
|
| 1877 |
+
"learning_rate": 4.5523569023569025e-05,
|
| 1878 |
+
"loss": 0.4561,
|
| 1879 |
+
"step": 53200
|
| 1880 |
+
},
|
| 1881 |
+
{
|
| 1882 |
+
"epoch": 0.0898989898989899,
|
| 1883 |
+
"grad_norm": 0.7278497815132141,
|
| 1884 |
+
"learning_rate": 4.550673400673401e-05,
|
| 1885 |
+
"loss": 0.4702,
|
| 1886 |
+
"step": 53400
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.09023569023569024,
|
| 1890 |
+
"grad_norm": 0.28533700108528137,
|
| 1891 |
+
"learning_rate": 4.5489898989898993e-05,
|
| 1892 |
+
"loss": 0.5139,
|
| 1893 |
+
"step": 53600
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 0.09057239057239057,
|
| 1897 |
+
"grad_norm": 0.6488041281700134,
|
| 1898 |
+
"learning_rate": 4.547306397306397e-05,
|
| 1899 |
+
"loss": 0.4553,
|
| 1900 |
+
"step": 53800
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 0.09090909090909091,
|
| 1904 |
+
"grad_norm": 0.3091227412223816,
|
| 1905 |
+
"learning_rate": 4.5456228956228955e-05,
|
| 1906 |
+
"loss": 0.4779,
|
| 1907 |
+
"step": 54000
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 0.09124579124579124,
|
| 1911 |
+
"grad_norm": 0.3282964825630188,
|
| 1912 |
+
"learning_rate": 4.543939393939394e-05,
|
| 1913 |
+
"loss": 0.5117,
|
| 1914 |
+
"step": 54200
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 0.09158249158249158,
|
| 1918 |
+
"grad_norm": 0.473143070936203,
|
| 1919 |
+
"learning_rate": 4.5422558922558924e-05,
|
| 1920 |
+
"loss": 0.4791,
|
| 1921 |
+
"step": 54400
|
| 1922 |
+
},
|
| 1923 |
+
{
|
| 1924 |
+
"epoch": 0.09191919191919191,
|
| 1925 |
+
"grad_norm": 0.5263796448707581,
|
| 1926 |
+
"learning_rate": 4.540572390572391e-05,
|
| 1927 |
+
"loss": 0.4686,
|
| 1928 |
+
"step": 54600
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.09225589225589226,
|
| 1932 |
+
"grad_norm": 0.4568365216255188,
|
| 1933 |
+
"learning_rate": 4.538888888888889e-05,
|
| 1934 |
+
"loss": 0.4936,
|
| 1935 |
+
"step": 54800
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.09259259259259259,
|
| 1939 |
+
"grad_norm": 0.9846563935279846,
|
| 1940 |
+
"learning_rate": 4.537205387205388e-05,
|
| 1941 |
+
"loss": 0.4968,
|
| 1942 |
+
"step": 55000
|
| 1943 |
+
},
|
| 1944 |
+
{
|
| 1945 |
+
"epoch": 0.09292929292929293,
|
| 1946 |
+
"grad_norm": 0.15145862102508545,
|
| 1947 |
+
"learning_rate": 4.5355218855218855e-05,
|
| 1948 |
+
"loss": 0.5163,
|
| 1949 |
+
"step": 55200
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 0.09326599326599326,
|
| 1953 |
+
"grad_norm": 0.428117573261261,
|
| 1954 |
+
"learning_rate": 4.533838383838384e-05,
|
| 1955 |
+
"loss": 0.4646,
|
| 1956 |
+
"step": 55400
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 0.0936026936026936,
|
| 1960 |
+
"grad_norm": 0.3261561393737793,
|
| 1961 |
+
"learning_rate": 4.5321548821548824e-05,
|
| 1962 |
+
"loss": 0.4697,
|
| 1963 |
+
"step": 55600
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 0.09393939393939393,
|
| 1967 |
+
"grad_norm": 0.34254854917526245,
|
| 1968 |
+
"learning_rate": 4.530471380471381e-05,
|
| 1969 |
+
"loss": 0.4331,
|
| 1970 |
+
"step": 55800
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.09427609427609428,
|
| 1974 |
+
"grad_norm": 0.15681512653827667,
|
| 1975 |
+
"learning_rate": 4.528787878787879e-05,
|
| 1976 |
+
"loss": 0.4765,
|
| 1977 |
+
"step": 56000
|
| 1978 |
+
},
|
| 1979 |
+
{
|
| 1980 |
+
"epoch": 0.0946127946127946,
|
| 1981 |
+
"grad_norm": 0.16159813106060028,
|
| 1982 |
+
"learning_rate": 4.527104377104378e-05,
|
| 1983 |
+
"loss": 0.4775,
|
| 1984 |
+
"step": 56200
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 0.09494949494949495,
|
| 1988 |
+
"grad_norm": 0.6212481260299683,
|
| 1989 |
+
"learning_rate": 4.5254208754208754e-05,
|
| 1990 |
+
"loss": 0.5314,
|
| 1991 |
+
"step": 56400
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 0.09528619528619528,
|
| 1995 |
+
"grad_norm": 0.3396393656730652,
|
| 1996 |
+
"learning_rate": 4.523737373737374e-05,
|
| 1997 |
+
"loss": 0.4898,
|
| 1998 |
+
"step": 56600
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 0.09562289562289562,
|
| 2002 |
+
"grad_norm": 0.32701626420021057,
|
| 2003 |
+
"learning_rate": 4.5220622895622896e-05,
|
| 2004 |
+
"loss": 0.4421,
|
| 2005 |
+
"step": 56800
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 0.09595959595959595,
|
| 2009 |
+
"grad_norm": 0.15523914992809296,
|
| 2010 |
+
"learning_rate": 4.520378787878788e-05,
|
| 2011 |
+
"loss": 0.4415,
|
| 2012 |
+
"step": 57000
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.0962962962962963,
|
| 2016 |
+
"grad_norm": 0.5103595852851868,
|
| 2017 |
+
"learning_rate": 4.5186952861952865e-05,
|
| 2018 |
+
"loss": 0.453,
|
| 2019 |
+
"step": 57200
|
| 2020 |
+
},
|
| 2021 |
+
{
|
| 2022 |
+
"epoch": 0.09663299663299663,
|
| 2023 |
+
"grad_norm": 0.42163121700286865,
|
| 2024 |
+
"learning_rate": 4.517011784511785e-05,
|
| 2025 |
+
"loss": 0.4738,
|
| 2026 |
+
"step": 57400
|
| 2027 |
+
},
|
| 2028 |
+
{
|
| 2029 |
+
"epoch": 0.09696969696969697,
|
| 2030 |
+
"grad_norm": 0.9396620392799377,
|
| 2031 |
+
"learning_rate": 4.515328282828283e-05,
|
| 2032 |
+
"loss": 0.473,
|
| 2033 |
+
"step": 57600
|
| 2034 |
+
},
|
| 2035 |
+
{
|
| 2036 |
+
"epoch": 0.09730639730639731,
|
| 2037 |
+
"grad_norm": 0.4714924991130829,
|
| 2038 |
+
"learning_rate": 4.513644781144781e-05,
|
| 2039 |
+
"loss": 0.5031,
|
| 2040 |
+
"step": 57800
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 0.09764309764309764,
|
| 2044 |
+
"grad_norm": 0.24718382954597473,
|
| 2045 |
+
"learning_rate": 4.5119612794612796e-05,
|
| 2046 |
+
"loss": 0.4592,
|
| 2047 |
+
"step": 58000
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 0.09797979797979799,
|
| 2051 |
+
"grad_norm": 0.3186817169189453,
|
| 2052 |
+
"learning_rate": 4.510277777777778e-05,
|
| 2053 |
+
"loss": 0.4451,
|
| 2054 |
+
"step": 58200
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.09831649831649832,
|
| 2058 |
+
"grad_norm": 0.34213390946388245,
|
| 2059 |
+
"learning_rate": 4.5085942760942764e-05,
|
| 2060 |
+
"loss": 0.4646,
|
| 2061 |
+
"step": 58400
|
| 2062 |
+
},
|
| 2063 |
+
{
|
| 2064 |
+
"epoch": 0.09865319865319866,
|
| 2065 |
+
"grad_norm": 0.29326021671295166,
|
| 2066 |
+
"learning_rate": 4.506910774410775e-05,
|
| 2067 |
+
"loss": 0.4825,
|
| 2068 |
+
"step": 58600
|
| 2069 |
+
},
|
| 2070 |
+
{
|
| 2071 |
+
"epoch": 0.09898989898989899,
|
| 2072 |
+
"grad_norm": 0.8425318598747253,
|
| 2073 |
+
"learning_rate": 4.505227272727273e-05,
|
| 2074 |
+
"loss": 0.4899,
|
| 2075 |
+
"step": 58800
|
| 2076 |
+
},
|
| 2077 |
+
{
|
| 2078 |
+
"epoch": 0.09932659932659933,
|
| 2079 |
+
"grad_norm": 0.23540657758712769,
|
| 2080 |
+
"learning_rate": 4.503543771043771e-05,
|
| 2081 |
+
"loss": 0.4654,
|
| 2082 |
+
"step": 59000
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 0.09966329966329966,
|
| 2086 |
+
"grad_norm": 0.44379663467407227,
|
| 2087 |
+
"learning_rate": 4.5018602693602695e-05,
|
| 2088 |
+
"loss": 0.4722,
|
| 2089 |
+
"step": 59200
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 0.1,
|
| 2093 |
+
"grad_norm": 0.23975303769111633,
|
| 2094 |
+
"learning_rate": 4.500176767676768e-05,
|
| 2095 |
+
"loss": 0.4589,
|
| 2096 |
+
"step": 59400
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.10033670033670034,
|
| 2100 |
+
"grad_norm": 0.29341402649879456,
|
| 2101 |
+
"learning_rate": 4.498501683501684e-05,
|
| 2102 |
+
"loss": 0.5041,
|
| 2103 |
+
"step": 59600
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 0.10067340067340068,
|
| 2107 |
+
"grad_norm": 0.34747016429901123,
|
| 2108 |
+
"learning_rate": 4.496818181818182e-05,
|
| 2109 |
+
"loss": 0.4647,
|
| 2110 |
+
"step": 59800
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 0.10101010101010101,
|
| 2114 |
+
"grad_norm": 0.8880186080932617,
|
| 2115 |
+
"learning_rate": 4.4951346801346806e-05,
|
| 2116 |
+
"loss": 0.4497,
|
| 2117 |
+
"step": 60000
|
| 2118 |
+
},
|
| 2119 |
+
{
|
| 2120 |
+
"epoch": 0.10134680134680135,
|
| 2121 |
+
"grad_norm": 0.6821927428245544,
|
| 2122 |
+
"learning_rate": 4.4934511784511783e-05,
|
| 2123 |
+
"loss": 0.4716,
|
| 2124 |
+
"step": 60200
|
| 2125 |
+
},
|
| 2126 |
+
{
|
| 2127 |
+
"epoch": 0.10168350168350168,
|
| 2128 |
+
"grad_norm": 0.4481610953807831,
|
| 2129 |
+
"learning_rate": 4.491767676767677e-05,
|
| 2130 |
+
"loss": 0.4576,
|
| 2131 |
+
"step": 60400
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 0.10202020202020202,
|
| 2135 |
+
"grad_norm": 0.6106315851211548,
|
| 2136 |
+
"learning_rate": 4.490084175084175e-05,
|
| 2137 |
+
"loss": 0.4386,
|
| 2138 |
+
"step": 60600
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.10235690235690235,
|
| 2142 |
+
"grad_norm": 0.3058004081249237,
|
| 2143 |
+
"learning_rate": 4.4884006734006737e-05,
|
| 2144 |
+
"loss": 0.4914,
|
| 2145 |
+
"step": 60800
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 0.1026936026936027,
|
| 2149 |
+
"grad_norm": 0.4685909152030945,
|
| 2150 |
+
"learning_rate": 4.486717171717172e-05,
|
| 2151 |
+
"loss": 0.4707,
|
| 2152 |
+
"step": 61000
|
| 2153 |
+
},
|
| 2154 |
+
{
|
| 2155 |
+
"epoch": 0.10303030303030303,
|
| 2156 |
+
"grad_norm": 0.643690288066864,
|
| 2157 |
+
"learning_rate": 4.4850336700336705e-05,
|
| 2158 |
+
"loss": 0.5428,
|
| 2159 |
+
"step": 61200
|
| 2160 |
+
},
|
| 2161 |
+
{
|
| 2162 |
+
"epoch": 0.10336700336700337,
|
| 2163 |
+
"grad_norm": 0.44329872727394104,
|
| 2164 |
+
"learning_rate": 4.483350168350168e-05,
|
| 2165 |
+
"loss": 0.4279,
|
| 2166 |
+
"step": 61400
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 0.1037037037037037,
|
| 2170 |
+
"grad_norm": 0.21025417745113373,
|
| 2171 |
+
"learning_rate": 4.481666666666667e-05,
|
| 2172 |
+
"loss": 0.458,
|
| 2173 |
+
"step": 61600
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 0.10404040404040404,
|
| 2177 |
+
"grad_norm": 0.24184706807136536,
|
| 2178 |
+
"learning_rate": 4.479983164983165e-05,
|
| 2179 |
+
"loss": 0.4964,
|
| 2180 |
+
"step": 61800
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.10437710437710437,
|
| 2184 |
+
"grad_norm": 0.4054918587207794,
|
| 2185 |
+
"learning_rate": 4.4782996632996636e-05,
|
| 2186 |
+
"loss": 0.4427,
|
| 2187 |
+
"step": 62000
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 0.10471380471380472,
|
| 2191 |
+
"grad_norm": 0.4844823181629181,
|
| 2192 |
+
"learning_rate": 4.476616161616162e-05,
|
| 2193 |
+
"loss": 0.4529,
|
| 2194 |
+
"step": 62200
|
| 2195 |
+
},
|
| 2196 |
+
{
|
| 2197 |
+
"epoch": 0.10505050505050505,
|
| 2198 |
+
"grad_norm": 0.39470815658569336,
|
| 2199 |
+
"learning_rate": 4.4749326599326605e-05,
|
| 2200 |
+
"loss": 0.4441,
|
| 2201 |
+
"step": 62400
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 0.10538720538720539,
|
| 2205 |
+
"grad_norm": 0.47343677282333374,
|
| 2206 |
+
"learning_rate": 4.473249158249159e-05,
|
| 2207 |
+
"loss": 0.4253,
|
| 2208 |
+
"step": 62600
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 0.10572390572390572,
|
| 2212 |
+
"grad_norm": 0.21908357739448547,
|
| 2213 |
+
"learning_rate": 4.471565656565657e-05,
|
| 2214 |
+
"loss": 0.4445,
|
| 2215 |
+
"step": 62800
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 0.10606060606060606,
|
| 2219 |
+
"grad_norm": 0.4983006715774536,
|
| 2220 |
+
"learning_rate": 4.469882154882155e-05,
|
| 2221 |
+
"loss": 0.4567,
|
| 2222 |
+
"step": 63000
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.1063973063973064,
|
| 2226 |
+
"grad_norm": 0.4615258276462555,
|
| 2227 |
+
"learning_rate": 4.4681986531986536e-05,
|
| 2228 |
+
"loss": 0.489,
|
| 2229 |
+
"step": 63200
|
| 2230 |
+
},
|
| 2231 |
+
{
|
| 2232 |
+
"epoch": 0.10673400673400674,
|
| 2233 |
+
"grad_norm": 0.30304470658302307,
|
| 2234 |
+
"learning_rate": 4.466523569023569e-05,
|
| 2235 |
+
"loss": 0.4748,
|
| 2236 |
+
"step": 63400
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 0.10707070707070707,
|
| 2240 |
+
"grad_norm": 0.1221388503909111,
|
| 2241 |
+
"learning_rate": 4.464840067340068e-05,
|
| 2242 |
+
"loss": 0.4529,
|
| 2243 |
+
"step": 63600
|
| 2244 |
+
},
|
| 2245 |
+
{
|
| 2246 |
+
"epoch": 0.10740740740740741,
|
| 2247 |
+
"grad_norm": 0.36816734075546265,
|
| 2248 |
+
"learning_rate": 4.463156565656566e-05,
|
| 2249 |
+
"loss": 0.4518,
|
| 2250 |
+
"step": 63800
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 0.10774410774410774,
|
| 2254 |
+
"grad_norm": 0.13719257712364197,
|
| 2255 |
+
"learning_rate": 4.461473063973064e-05,
|
| 2256 |
+
"loss": 0.4314,
|
| 2257 |
+
"step": 64000
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 0.10808080808080808,
|
| 2261 |
+
"grad_norm": 0.45440635085105896,
|
| 2262 |
+
"learning_rate": 4.4597895622895624e-05,
|
| 2263 |
+
"loss": 0.5005,
|
| 2264 |
+
"step": 64200
|
| 2265 |
+
},
|
| 2266 |
+
{
|
| 2267 |
+
"epoch": 0.10841750841750841,
|
| 2268 |
+
"grad_norm": 0.4109625816345215,
|
| 2269 |
+
"learning_rate": 4.458106060606061e-05,
|
| 2270 |
+
"loss": 0.4847,
|
| 2271 |
+
"step": 64400
|
| 2272 |
+
},
|
| 2273 |
+
{
|
| 2274 |
+
"epoch": 0.10875420875420876,
|
| 2275 |
+
"grad_norm": 0.43106183409690857,
|
| 2276 |
+
"learning_rate": 4.4564225589225586e-05,
|
| 2277 |
+
"loss": 0.4607,
|
| 2278 |
+
"step": 64600
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 0.10909090909090909,
|
| 2282 |
+
"grad_norm": 0.3796352744102478,
|
| 2283 |
+
"learning_rate": 4.454739057239058e-05,
|
| 2284 |
+
"loss": 0.4793,
|
| 2285 |
+
"step": 64800
|
| 2286 |
+
},
|
| 2287 |
+
{
|
| 2288 |
+
"epoch": 0.10942760942760943,
|
| 2289 |
+
"grad_norm": 0.5014599561691284,
|
| 2290 |
+
"learning_rate": 4.453055555555556e-05,
|
| 2291 |
+
"loss": 0.4848,
|
| 2292 |
+
"step": 65000
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.10976430976430976,
|
| 2296 |
+
"grad_norm": 0.3484991192817688,
|
| 2297 |
+
"learning_rate": 4.4513720538720546e-05,
|
| 2298 |
+
"loss": 0.5129,
|
| 2299 |
+
"step": 65200
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 0.1101010101010101,
|
| 2303 |
+
"grad_norm": 0.2991756200790405,
|
| 2304 |
+
"learning_rate": 4.449688552188552e-05,
|
| 2305 |
+
"loss": 0.4927,
|
| 2306 |
+
"step": 65400
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.11043771043771043,
|
| 2310 |
+
"grad_norm": 0.37985363602638245,
|
| 2311 |
+
"learning_rate": 4.448005050505051e-05,
|
| 2312 |
+
"loss": 0.4372,
|
| 2313 |
+
"step": 65600
|
| 2314 |
+
},
|
| 2315 |
+
{
|
| 2316 |
+
"epoch": 0.11077441077441078,
|
| 2317 |
+
"grad_norm": 0.4508950114250183,
|
| 2318 |
+
"learning_rate": 4.446321548821549e-05,
|
| 2319 |
+
"loss": 0.4182,
|
| 2320 |
+
"step": 65800
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 0.1111111111111111,
|
| 2324 |
+
"grad_norm": 0.7306589484214783,
|
| 2325 |
+
"learning_rate": 4.444646464646465e-05,
|
| 2326 |
+
"loss": 0.5217,
|
| 2327 |
+
"step": 66000
|
| 2328 |
+
},
|
| 2329 |
+
{
|
| 2330 |
+
"epoch": 0.11144781144781145,
|
| 2331 |
+
"grad_norm": 0.17345421016216278,
|
| 2332 |
+
"learning_rate": 4.4429629629629634e-05,
|
| 2333 |
+
"loss": 0.4653,
|
| 2334 |
+
"step": 66200
|
| 2335 |
+
},
|
| 2336 |
+
{
|
| 2337 |
+
"epoch": 0.11178451178451178,
|
| 2338 |
+
"grad_norm": 0.36997854709625244,
|
| 2339 |
+
"learning_rate": 4.441279461279462e-05,
|
| 2340 |
+
"loss": 0.4631,
|
| 2341 |
+
"step": 66400
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 0.11212121212121212,
|
| 2345 |
+
"grad_norm": 0.31563735008239746,
|
| 2346 |
+
"learning_rate": 4.4395959595959596e-05,
|
| 2347 |
+
"loss": 0.4606,
|
| 2348 |
+
"step": 66600
|
| 2349 |
+
},
|
| 2350 |
+
{
|
| 2351 |
+
"epoch": 0.11245791245791245,
|
| 2352 |
+
"grad_norm": 0.23173430562019348,
|
| 2353 |
+
"learning_rate": 4.437912457912458e-05,
|
| 2354 |
+
"loss": 0.4868,
|
| 2355 |
+
"step": 66800
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 0.1127946127946128,
|
| 2359 |
+
"grad_norm": 0.336233526468277,
|
| 2360 |
+
"learning_rate": 4.4362289562289565e-05,
|
| 2361 |
+
"loss": 0.4809,
|
| 2362 |
+
"step": 67000
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 0.11313131313131314,
|
| 2366 |
+
"grad_norm": 0.3722301423549652,
|
| 2367 |
+
"learning_rate": 4.434545454545454e-05,
|
| 2368 |
+
"loss": 0.4396,
|
| 2369 |
+
"step": 67200
|
| 2370 |
+
},
|
| 2371 |
+
{
|
| 2372 |
+
"epoch": 0.11346801346801347,
|
| 2373 |
+
"grad_norm": 0.5491744875907898,
|
| 2374 |
+
"learning_rate": 4.432861952861953e-05,
|
| 2375 |
+
"loss": 0.3938,
|
| 2376 |
+
"step": 67400
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 0.11380471380471381,
|
| 2380 |
+
"grad_norm": 0.2742317020893097,
|
| 2381 |
+
"learning_rate": 4.431178451178452e-05,
|
| 2382 |
+
"loss": 0.4782,
|
| 2383 |
+
"step": 67600
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 0.11414141414141414,
|
| 2387 |
+
"grad_norm": 0.22197793424129486,
|
| 2388 |
+
"learning_rate": 4.4294949494949495e-05,
|
| 2389 |
+
"loss": 0.4233,
|
| 2390 |
+
"step": 67800
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.11447811447811448,
|
| 2394 |
+
"grad_norm": 0.2449079006910324,
|
| 2395 |
+
"learning_rate": 4.427811447811448e-05,
|
| 2396 |
+
"loss": 0.4889,
|
| 2397 |
+
"step": 68000
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 0.11481481481481481,
|
| 2401 |
+
"grad_norm": 0.25308141112327576,
|
| 2402 |
+
"learning_rate": 4.4261279461279464e-05,
|
| 2403 |
+
"loss": 0.4275,
|
| 2404 |
+
"step": 68200
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 0.11515151515151516,
|
| 2408 |
+
"grad_norm": 0.22254426777362823,
|
| 2409 |
+
"learning_rate": 4.424444444444444e-05,
|
| 2410 |
+
"loss": 0.487,
|
| 2411 |
+
"step": 68400
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 0.11548821548821549,
|
| 2415 |
+
"grad_norm": 0.4978940784931183,
|
| 2416 |
+
"learning_rate": 4.4227609427609426e-05,
|
| 2417 |
+
"loss": 0.4709,
|
| 2418 |
+
"step": 68600
|
| 2419 |
+
},
|
| 2420 |
+
{
|
| 2421 |
+
"epoch": 0.11582491582491583,
|
| 2422 |
+
"grad_norm": 0.2366330623626709,
|
| 2423 |
+
"learning_rate": 4.421077441077441e-05,
|
| 2424 |
+
"loss": 0.487,
|
| 2425 |
+
"step": 68800
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 0.11616161616161616,
|
| 2429 |
+
"grad_norm": 0.38192349672317505,
|
| 2430 |
+
"learning_rate": 4.41939393939394e-05,
|
| 2431 |
+
"loss": 0.4751,
|
| 2432 |
+
"step": 69000
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.1164983164983165,
|
| 2436 |
+
"grad_norm": 0.4711579382419586,
|
| 2437 |
+
"learning_rate": 4.417710437710438e-05,
|
| 2438 |
+
"loss": 0.4694,
|
| 2439 |
+
"step": 69200
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.11683501683501683,
|
| 2443 |
+
"grad_norm": 0.7776811122894287,
|
| 2444 |
+
"learning_rate": 4.4160269360269364e-05,
|
| 2445 |
+
"loss": 0.505,
|
| 2446 |
+
"step": 69400
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 0.11717171717171718,
|
| 2450 |
+
"grad_norm": 0.22125215828418732,
|
| 2451 |
+
"learning_rate": 4.414343434343435e-05,
|
| 2452 |
+
"loss": 0.4599,
|
| 2453 |
+
"step": 69600
|
| 2454 |
+
},
|
| 2455 |
+
{
|
| 2456 |
+
"epoch": 0.1175084175084175,
|
| 2457 |
+
"grad_norm": 0.3384982943534851,
|
| 2458 |
+
"learning_rate": 4.4126683501683505e-05,
|
| 2459 |
+
"loss": 0.4825,
|
| 2460 |
+
"step": 69800
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 0.11784511784511785,
|
| 2464 |
+
"grad_norm": 0.35308724641799927,
|
| 2465 |
+
"learning_rate": 4.410984848484849e-05,
|
| 2466 |
+
"loss": 0.4698,
|
| 2467 |
+
"step": 70000
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 0.11818181818181818,
|
| 2471 |
+
"grad_norm": 0.3890261650085449,
|
| 2472 |
+
"learning_rate": 4.4093013468013474e-05,
|
| 2473 |
+
"loss": 0.4414,
|
| 2474 |
+
"step": 70200
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.11851851851851852,
|
| 2478 |
+
"grad_norm": 0.6729969382286072,
|
| 2479 |
+
"learning_rate": 4.407617845117845e-05,
|
| 2480 |
+
"loss": 0.5287,
|
| 2481 |
+
"step": 70400
|
| 2482 |
+
},
|
| 2483 |
+
{
|
| 2484 |
+
"epoch": 0.11885521885521885,
|
| 2485 |
+
"grad_norm": 0.18775266408920288,
|
| 2486 |
+
"learning_rate": 4.4059343434343436e-05,
|
| 2487 |
+
"loss": 0.4809,
|
| 2488 |
+
"step": 70600
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 0.1191919191919192,
|
| 2492 |
+
"grad_norm": 0.13612866401672363,
|
| 2493 |
+
"learning_rate": 4.404250841750842e-05,
|
| 2494 |
+
"loss": 0.4591,
|
| 2495 |
+
"step": 70800
|
| 2496 |
+
},
|
| 2497 |
+
{
|
| 2498 |
+
"epoch": 0.11952861952861953,
|
| 2499 |
+
"grad_norm": 0.25239524245262146,
|
| 2500 |
+
"learning_rate": 4.40256734006734e-05,
|
| 2501 |
+
"loss": 0.4514,
|
| 2502 |
+
"step": 71000
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 0.11986531986531987,
|
| 2506 |
+
"grad_norm": 0.5928908586502075,
|
| 2507 |
+
"learning_rate": 4.400883838383838e-05,
|
| 2508 |
+
"loss": 0.4437,
|
| 2509 |
+
"step": 71200
|
| 2510 |
+
},
|
| 2511 |
+
{
|
| 2512 |
+
"epoch": 0.1202020202020202,
|
| 2513 |
+
"grad_norm": 0.3206656575202942,
|
| 2514 |
+
"learning_rate": 4.3992003367003374e-05,
|
| 2515 |
+
"loss": 0.4193,
|
| 2516 |
+
"step": 71400
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.12053872053872054,
|
| 2520 |
+
"grad_norm": 0.1375039964914322,
|
| 2521 |
+
"learning_rate": 4.397516835016835e-05,
|
| 2522 |
+
"loss": 0.4642,
|
| 2523 |
+
"step": 71600
|
| 2524 |
+
},
|
| 2525 |
+
{
|
| 2526 |
+
"epoch": 0.12087542087542087,
|
| 2527 |
+
"grad_norm": 0.31986692547798157,
|
| 2528 |
+
"learning_rate": 4.3958333333333336e-05,
|
| 2529 |
+
"loss": 0.4771,
|
| 2530 |
+
"step": 71800
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.12121212121212122,
|
| 2534 |
+
"grad_norm": 0.3976145088672638,
|
| 2535 |
+
"learning_rate": 4.394158249158249e-05,
|
| 2536 |
+
"loss": 0.5199,
|
| 2537 |
+
"step": 72000
|
| 2538 |
+
},
|
| 2539 |
+
{
|
| 2540 |
+
"epoch": 0.12154882154882155,
|
| 2541 |
+
"grad_norm": 0.5115092992782593,
|
| 2542 |
+
"learning_rate": 4.392474747474747e-05,
|
| 2543 |
+
"loss": 0.4639,
|
| 2544 |
+
"step": 72200
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 0.12188552188552189,
|
| 2548 |
+
"grad_norm": 0.6884472370147705,
|
| 2549 |
+
"learning_rate": 4.390791245791246e-05,
|
| 2550 |
+
"loss": 0.5004,
|
| 2551 |
+
"step": 72400
|
| 2552 |
+
},
|
| 2553 |
+
{
|
| 2554 |
+
"epoch": 0.12222222222222222,
|
| 2555 |
+
"grad_norm": 0.35599565505981445,
|
| 2556 |
+
"learning_rate": 4.3891077441077446e-05,
|
| 2557 |
+
"loss": 0.4125,
|
| 2558 |
+
"step": 72600
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.12255892255892256,
|
| 2562 |
+
"grad_norm": 0.31085407733917236,
|
| 2563 |
+
"learning_rate": 4.387424242424243e-05,
|
| 2564 |
+
"loss": 0.4276,
|
| 2565 |
+
"step": 72800
|
| 2566 |
+
},
|
| 2567 |
+
{
|
| 2568 |
+
"epoch": 0.12289562289562289,
|
| 2569 |
+
"grad_norm": 0.4173491299152374,
|
| 2570 |
+
"learning_rate": 4.385740740740741e-05,
|
| 2571 |
+
"loss": 0.5051,
|
| 2572 |
+
"step": 73000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 0.12323232323232323,
|
| 2576 |
+
"grad_norm": 0.857441782951355,
|
| 2577 |
+
"learning_rate": 4.384057239057239e-05,
|
| 2578 |
+
"loss": 0.4494,
|
| 2579 |
+
"step": 73200
|
| 2580 |
+
},
|
| 2581 |
+
{
|
| 2582 |
+
"epoch": 0.12356902356902356,
|
| 2583 |
+
"grad_norm": 0.2960607707500458,
|
| 2584 |
+
"learning_rate": 4.382382154882155e-05,
|
| 2585 |
+
"loss": 0.5233,
|
| 2586 |
+
"step": 73400
|
| 2587 |
+
},
|
| 2588 |
+
{
|
| 2589 |
+
"epoch": 0.12390572390572391,
|
| 2590 |
+
"grad_norm": 0.38231462240219116,
|
| 2591 |
+
"learning_rate": 4.3806986531986535e-05,
|
| 2592 |
+
"loss": 0.5003,
|
| 2593 |
+
"step": 73600
|
| 2594 |
+
},
|
| 2595 |
+
{
|
| 2596 |
+
"epoch": 0.12424242424242424,
|
| 2597 |
+
"grad_norm": 0.20170505344867706,
|
| 2598 |
+
"learning_rate": 4.379015151515152e-05,
|
| 2599 |
+
"loss": 0.4543,
|
| 2600 |
+
"step": 73800
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.12457912457912458,
|
| 2604 |
+
"grad_norm": 0.49870565533638,
|
| 2605 |
+
"learning_rate": 4.37733164983165e-05,
|
| 2606 |
+
"loss": 0.4931,
|
| 2607 |
+
"step": 74000
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 0.12491582491582491,
|
| 2611 |
+
"grad_norm": 0.21166172623634338,
|
| 2612 |
+
"learning_rate": 4.375648148148148e-05,
|
| 2613 |
+
"loss": 0.4956,
|
| 2614 |
+
"step": 74200
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 0.12525252525252525,
|
| 2618 |
+
"grad_norm": 1.7191145420074463,
|
| 2619 |
+
"learning_rate": 4.3739646464646465e-05,
|
| 2620 |
+
"loss": 0.5209,
|
| 2621 |
+
"step": 74400
|
| 2622 |
+
},
|
| 2623 |
+
{
|
| 2624 |
+
"epoch": 0.12558922558922558,
|
| 2625 |
+
"grad_norm": 0.15442191064357758,
|
| 2626 |
+
"learning_rate": 4.372281144781145e-05,
|
| 2627 |
+
"loss": 0.4723,
|
| 2628 |
+
"step": 74600
|
| 2629 |
+
},
|
| 2630 |
+
{
|
| 2631 |
+
"epoch": 0.1259259259259259,
|
| 2632 |
+
"grad_norm": 0.6088646650314331,
|
| 2633 |
+
"learning_rate": 4.3705976430976434e-05,
|
| 2634 |
+
"loss": 0.4756,
|
| 2635 |
+
"step": 74800
|
| 2636 |
+
},
|
| 2637 |
+
{
|
| 2638 |
+
"epoch": 0.12626262626262627,
|
| 2639 |
+
"grad_norm": 0.20357204973697662,
|
| 2640 |
+
"learning_rate": 4.368914141414142e-05,
|
| 2641 |
+
"loss": 0.4358,
|
| 2642 |
+
"step": 75000
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.1265993265993266,
|
| 2646 |
+
"grad_norm": 0.23374512791633606,
|
| 2647 |
+
"learning_rate": 4.36723063973064e-05,
|
| 2648 |
+
"loss": 0.4873,
|
| 2649 |
+
"step": 75200
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.12693602693602693,
|
| 2653 |
+
"grad_norm": 0.28036201000213623,
|
| 2654 |
+
"learning_rate": 4.365547138047138e-05,
|
| 2655 |
+
"loss": 0.4627,
|
| 2656 |
+
"step": 75400
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 0.12727272727272726,
|
| 2660 |
+
"grad_norm": 0.4876658320426941,
|
| 2661 |
+
"learning_rate": 4.3638636363636365e-05,
|
| 2662 |
+
"loss": 0.4865,
|
| 2663 |
+
"step": 75600
|
| 2664 |
+
},
|
| 2665 |
+
{
|
| 2666 |
+
"epoch": 0.12760942760942762,
|
| 2667 |
+
"grad_norm": 0.24980993568897247,
|
| 2668 |
+
"learning_rate": 4.362180134680135e-05,
|
| 2669 |
+
"loss": 0.4715,
|
| 2670 |
+
"step": 75800
|
| 2671 |
+
},
|
| 2672 |
+
{
|
| 2673 |
+
"epoch": 0.12794612794612795,
|
| 2674 |
+
"grad_norm": 0.5056689977645874,
|
| 2675 |
+
"learning_rate": 4.3604966329966334e-05,
|
| 2676 |
+
"loss": 0.5066,
|
| 2677 |
+
"step": 76000
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 0.12828282828282828,
|
| 2681 |
+
"grad_norm": 0.3562251329421997,
|
| 2682 |
+
"learning_rate": 4.358813131313131e-05,
|
| 2683 |
+
"loss": 0.4561,
|
| 2684 |
+
"step": 76200
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.1286195286195286,
|
| 2688 |
+
"grad_norm": 0.5188980102539062,
|
| 2689 |
+
"learning_rate": 4.35712962962963e-05,
|
| 2690 |
+
"loss": 0.4451,
|
| 2691 |
+
"step": 76400
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 0.12895622895622896,
|
| 2695 |
+
"grad_norm": 0.24979503452777863,
|
| 2696 |
+
"learning_rate": 4.355446127946129e-05,
|
| 2697 |
+
"loss": 0.4977,
|
| 2698 |
+
"step": 76600
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 0.1292929292929293,
|
| 2702 |
+
"grad_norm": 0.2918744385242462,
|
| 2703 |
+
"learning_rate": 4.3537626262626264e-05,
|
| 2704 |
+
"loss": 0.4465,
|
| 2705 |
+
"step": 76800
|
| 2706 |
+
},
|
| 2707 |
+
{
|
| 2708 |
+
"epoch": 0.12962962962962962,
|
| 2709 |
+
"grad_norm": 0.11484523117542267,
|
| 2710 |
+
"learning_rate": 4.352079124579125e-05,
|
| 2711 |
+
"loss": 0.5008,
|
| 2712 |
+
"step": 77000
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 0.12996632996632998,
|
| 2716 |
+
"grad_norm": 0.13835379481315613,
|
| 2717 |
+
"learning_rate": 4.350395622895623e-05,
|
| 2718 |
+
"loss": 0.4469,
|
| 2719 |
+
"step": 77200
|
| 2720 |
+
},
|
| 2721 |
+
{
|
| 2722 |
+
"epoch": 0.1303030303030303,
|
| 2723 |
+
"grad_norm": 0.17499831318855286,
|
| 2724 |
+
"learning_rate": 4.348712121212121e-05,
|
| 2725 |
+
"loss": 0.4558,
|
| 2726 |
+
"step": 77400
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.13063973063973064,
|
| 2730 |
+
"grad_norm": 0.14153020083904266,
|
| 2731 |
+
"learning_rate": 4.3470286195286195e-05,
|
| 2732 |
+
"loss": 0.4764,
|
| 2733 |
+
"step": 77600
|
| 2734 |
+
},
|
| 2735 |
+
{
|
| 2736 |
+
"epoch": 0.13097643097643097,
|
| 2737 |
+
"grad_norm": 0.32110026478767395,
|
| 2738 |
+
"learning_rate": 4.345353535353536e-05,
|
| 2739 |
+
"loss": 0.4496,
|
| 2740 |
+
"step": 77800
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 0.13131313131313133,
|
| 2744 |
+
"grad_norm": 0.347741961479187,
|
| 2745 |
+
"learning_rate": 4.343670033670034e-05,
|
| 2746 |
+
"loss": 0.4365,
|
| 2747 |
+
"step": 78000
|
| 2748 |
+
},
|
| 2749 |
+
{
|
| 2750 |
+
"epoch": 0.13164983164983166,
|
| 2751 |
+
"grad_norm": 1.8142364025115967,
|
| 2752 |
+
"learning_rate": 4.341986531986532e-05,
|
| 2753 |
+
"loss": 0.4963,
|
| 2754 |
+
"step": 78200
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 0.13198653198653199,
|
| 2758 |
+
"grad_norm": 0.18966235220432281,
|
| 2759 |
+
"learning_rate": 4.3403030303030306e-05,
|
| 2760 |
+
"loss": 0.4641,
|
| 2761 |
+
"step": 78400
|
| 2762 |
+
},
|
| 2763 |
+
{
|
| 2764 |
+
"epoch": 0.13232323232323231,
|
| 2765 |
+
"grad_norm": 0.7899078726768494,
|
| 2766 |
+
"learning_rate": 4.338619528619528e-05,
|
| 2767 |
+
"loss": 0.42,
|
| 2768 |
+
"step": 78600
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.13265993265993267,
|
| 2772 |
+
"grad_norm": 0.11921744793653488,
|
| 2773 |
+
"learning_rate": 4.336936026936027e-05,
|
| 2774 |
+
"loss": 0.4489,
|
| 2775 |
+
"step": 78800
|
| 2776 |
+
},
|
| 2777 |
+
{
|
| 2778 |
+
"epoch": 0.132996632996633,
|
| 2779 |
+
"grad_norm": 0.3183203935623169,
|
| 2780 |
+
"learning_rate": 4.335252525252526e-05,
|
| 2781 |
+
"loss": 0.4128,
|
| 2782 |
+
"step": 79000
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 0.13333333333333333,
|
| 2786 |
+
"grad_norm": 0.9126468896865845,
|
| 2787 |
+
"learning_rate": 4.3335690235690236e-05,
|
| 2788 |
+
"loss": 0.4063,
|
| 2789 |
+
"step": 79200
|
| 2790 |
+
},
|
| 2791 |
+
{
|
| 2792 |
+
"epoch": 0.13367003367003366,
|
| 2793 |
+
"grad_norm": 0.30544015765190125,
|
| 2794 |
+
"learning_rate": 4.331885521885522e-05,
|
| 2795 |
+
"loss": 0.4363,
|
| 2796 |
+
"step": 79400
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.13400673400673402,
|
| 2800 |
+
"grad_norm": 0.32495343685150146,
|
| 2801 |
+
"learning_rate": 4.3302020202020205e-05,
|
| 2802 |
+
"loss": 0.4311,
|
| 2803 |
+
"step": 79600
|
| 2804 |
+
},
|
| 2805 |
+
{
|
| 2806 |
+
"epoch": 0.13434343434343435,
|
| 2807 |
+
"grad_norm": 0.5995136499404907,
|
| 2808 |
+
"learning_rate": 4.328518518518519e-05,
|
| 2809 |
+
"loss": 0.4304,
|
| 2810 |
+
"step": 79800
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.13468013468013468,
|
| 2814 |
+
"grad_norm": 0.09273191541433334,
|
| 2815 |
+
"learning_rate": 4.326835016835017e-05,
|
| 2816 |
+
"loss": 0.44,
|
| 2817 |
+
"step": 80000
|
| 2818 |
+
},
|
| 2819 |
+
{
|
| 2820 |
+
"epoch": 0.135016835016835,
|
| 2821 |
+
"grad_norm": 0.5083215832710266,
|
| 2822 |
+
"learning_rate": 4.325151515151515e-05,
|
| 2823 |
+
"loss": 0.4484,
|
| 2824 |
+
"step": 80200
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 0.13535353535353536,
|
| 2828 |
+
"grad_norm": 0.4086732566356659,
|
| 2829 |
+
"learning_rate": 4.3234680134680136e-05,
|
| 2830 |
+
"loss": 0.4431,
|
| 2831 |
+
"step": 80400
|
| 2832 |
+
},
|
| 2833 |
+
{
|
| 2834 |
+
"epoch": 0.1356902356902357,
|
| 2835 |
+
"grad_norm": 0.25825831294059753,
|
| 2836 |
+
"learning_rate": 4.321784511784512e-05,
|
| 2837 |
+
"loss": 0.4903,
|
| 2838 |
+
"step": 80600
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 0.13602693602693602,
|
| 2842 |
+
"grad_norm": 0.22938190400600433,
|
| 2843 |
+
"learning_rate": 4.3201010101010105e-05,
|
| 2844 |
+
"loss": 0.446,
|
| 2845 |
+
"step": 80800
|
| 2846 |
+
},
|
| 2847 |
+
{
|
| 2848 |
+
"epoch": 0.13636363636363635,
|
| 2849 |
+
"grad_norm": 4.104939937591553,
|
| 2850 |
+
"learning_rate": 4.318417508417509e-05,
|
| 2851 |
+
"loss": 0.5158,
|
| 2852 |
+
"step": 81000
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.1367003367003367,
|
| 2856 |
+
"grad_norm": 0.6533791422843933,
|
| 2857 |
+
"learning_rate": 4.3167340067340067e-05,
|
| 2858 |
+
"loss": 0.4411,
|
| 2859 |
+
"step": 81200
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 0.13703703703703704,
|
| 2863 |
+
"grad_norm": 0.3710763156414032,
|
| 2864 |
+
"learning_rate": 4.315050505050505e-05,
|
| 2865 |
+
"loss": 0.4504,
|
| 2866 |
+
"step": 81400
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 0.13737373737373737,
|
| 2870 |
+
"grad_norm": 0.32174888253211975,
|
| 2871 |
+
"learning_rate": 4.3133670033670035e-05,
|
| 2872 |
+
"loss": 0.4494,
|
| 2873 |
+
"step": 81600
|
| 2874 |
+
},
|
| 2875 |
+
{
|
| 2876 |
+
"epoch": 0.1377104377104377,
|
| 2877 |
+
"grad_norm": 0.47536543011665344,
|
| 2878 |
+
"learning_rate": 4.311683501683502e-05,
|
| 2879 |
+
"loss": 0.4547,
|
| 2880 |
+
"step": 81800
|
| 2881 |
+
},
|
| 2882 |
+
{
|
| 2883 |
+
"epoch": 0.13804713804713806,
|
| 2884 |
+
"grad_norm": 0.5591254830360413,
|
| 2885 |
+
"learning_rate": 4.310008417508418e-05,
|
| 2886 |
+
"loss": 0.4306,
|
| 2887 |
+
"step": 82000
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 0.1383838383838384,
|
| 2891 |
+
"grad_norm": 0.26102516055107117,
|
| 2892 |
+
"learning_rate": 4.308324915824916e-05,
|
| 2893 |
+
"loss": 0.4377,
|
| 2894 |
+
"step": 82200
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.13872053872053872,
|
| 2898 |
+
"grad_norm": 0.540073812007904,
|
| 2899 |
+
"learning_rate": 4.306641414141414e-05,
|
| 2900 |
+
"loss": 0.4913,
|
| 2901 |
+
"step": 82400
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 0.13905723905723905,
|
| 2905 |
+
"grad_norm": 0.8017529249191284,
|
| 2906 |
+
"learning_rate": 4.3049579124579124e-05,
|
| 2907 |
+
"loss": 0.4382,
|
| 2908 |
+
"step": 82600
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 0.1393939393939394,
|
| 2912 |
+
"grad_norm": 0.6620075106620789,
|
| 2913 |
+
"learning_rate": 4.303274410774411e-05,
|
| 2914 |
+
"loss": 0.3944,
|
| 2915 |
+
"step": 82800
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.13973063973063973,
|
| 2919 |
+
"grad_norm": 0.42346033453941345,
|
| 2920 |
+
"learning_rate": 4.301590909090909e-05,
|
| 2921 |
+
"loss": 0.4149,
|
| 2922 |
+
"step": 83000
|
| 2923 |
+
},
|
| 2924 |
+
{
|
| 2925 |
+
"epoch": 0.14006734006734006,
|
| 2926 |
+
"grad_norm": 0.259355366230011,
|
| 2927 |
+
"learning_rate": 4.299907407407408e-05,
|
| 2928 |
+
"loss": 0.4737,
|
| 2929 |
+
"step": 83200
|
| 2930 |
+
},
|
| 2931 |
+
{
|
| 2932 |
+
"epoch": 0.1404040404040404,
|
| 2933 |
+
"grad_norm": 0.10005613416433334,
|
| 2934 |
+
"learning_rate": 4.298223905723906e-05,
|
| 2935 |
+
"loss": 0.4642,
|
| 2936 |
+
"step": 83400
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 0.14074074074074075,
|
| 2940 |
+
"grad_norm": 0.40637290477752686,
|
| 2941 |
+
"learning_rate": 4.2965404040404045e-05,
|
| 2942 |
+
"loss": 0.4251,
|
| 2943 |
+
"step": 83600
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 0.14107744107744108,
|
| 2947 |
+
"grad_norm": 0.5488855838775635,
|
| 2948 |
+
"learning_rate": 4.294856902356902e-05,
|
| 2949 |
+
"loss": 0.4669,
|
| 2950 |
+
"step": 83800
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 0.1414141414141414,
|
| 2954 |
+
"grad_norm": 0.36019712686538696,
|
| 2955 |
+
"learning_rate": 4.293173400673401e-05,
|
| 2956 |
+
"loss": 0.4493,
|
| 2957 |
+
"step": 84000
|
| 2958 |
+
},
|
| 2959 |
+
{
|
| 2960 |
+
"epoch": 0.14175084175084174,
|
| 2961 |
+
"grad_norm": 0.20330995321273804,
|
| 2962 |
+
"learning_rate": 4.291498316498317e-05,
|
| 2963 |
+
"loss": 0.4969,
|
| 2964 |
+
"step": 84200
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 0.1420875420875421,
|
| 2968 |
+
"grad_norm": 0.23681996762752533,
|
| 2969 |
+
"learning_rate": 4.289814814814815e-05,
|
| 2970 |
+
"loss": 0.4774,
|
| 2971 |
+
"step": 84400
|
| 2972 |
+
},
|
| 2973 |
+
{
|
| 2974 |
+
"epoch": 0.14242424242424243,
|
| 2975 |
+
"grad_norm": 0.1740342527627945,
|
| 2976 |
+
"learning_rate": 4.2881313131313134e-05,
|
| 2977 |
+
"loss": 0.4915,
|
| 2978 |
+
"step": 84600
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 0.14276094276094276,
|
| 2982 |
+
"grad_norm": 0.4355227053165436,
|
| 2983 |
+
"learning_rate": 4.286447811447812e-05,
|
| 2984 |
+
"loss": 0.4681,
|
| 2985 |
+
"step": 84800
|
| 2986 |
+
},
|
| 2987 |
+
{
|
| 2988 |
+
"epoch": 0.14309764309764308,
|
| 2989 |
+
"grad_norm": 0.295913964509964,
|
| 2990 |
+
"learning_rate": 4.2847643097643096e-05,
|
| 2991 |
+
"loss": 0.4818,
|
| 2992 |
+
"step": 85000
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 0.14343434343434344,
|
| 2996 |
+
"grad_norm": 0.17617417871952057,
|
| 2997 |
+
"learning_rate": 4.283080808080808e-05,
|
| 2998 |
+
"loss": 0.4661,
|
| 2999 |
+
"step": 85200
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 0.14377104377104377,
|
| 3003 |
+
"grad_norm": 0.36346206068992615,
|
| 3004 |
+
"learning_rate": 4.2813973063973064e-05,
|
| 3005 |
+
"loss": 0.456,
|
| 3006 |
+
"step": 85400
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.1441077441077441,
|
| 3010 |
+
"grad_norm": 0.5108135342597961,
|
| 3011 |
+
"learning_rate": 4.279713804713805e-05,
|
| 3012 |
+
"loss": 0.4311,
|
| 3013 |
+
"step": 85600
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 0.14444444444444443,
|
| 3017 |
+
"grad_norm": 0.6109139919281006,
|
| 3018 |
+
"learning_rate": 4.278030303030303e-05,
|
| 3019 |
+
"loss": 0.4398,
|
| 3020 |
+
"step": 85800
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 0.1447811447811448,
|
| 3024 |
+
"grad_norm": 0.35451897978782654,
|
| 3025 |
+
"learning_rate": 4.276346801346802e-05,
|
| 3026 |
+
"loss": 0.4674,
|
| 3027 |
+
"step": 86000
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 0.14511784511784512,
|
| 3031 |
+
"grad_norm": 0.4733004570007324,
|
| 3032 |
+
"learning_rate": 4.2746632996632995e-05,
|
| 3033 |
+
"loss": 0.4275,
|
| 3034 |
+
"step": 86200
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.14545454545454545,
|
| 3038 |
+
"grad_norm": 0.30051594972610474,
|
| 3039 |
+
"learning_rate": 4.272979797979798e-05,
|
| 3040 |
+
"loss": 0.4493,
|
| 3041 |
+
"step": 86400
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 0.1457912457912458,
|
| 3045 |
+
"grad_norm": 0.5513753294944763,
|
| 3046 |
+
"learning_rate": 4.2712962962962964e-05,
|
| 3047 |
+
"loss": 0.4803,
|
| 3048 |
+
"step": 86600
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 0.14612794612794613,
|
| 3052 |
+
"grad_norm": 0.5906115770339966,
|
| 3053 |
+
"learning_rate": 4.269612794612795e-05,
|
| 3054 |
+
"loss": 0.4502,
|
| 3055 |
+
"step": 86800
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 0.14646464646464646,
|
| 3059 |
+
"grad_norm": 0.662507176399231,
|
| 3060 |
+
"learning_rate": 4.267929292929293e-05,
|
| 3061 |
+
"loss": 0.4247,
|
| 3062 |
+
"step": 87000
|
| 3063 |
+
},
|
| 3064 |
+
{
|
| 3065 |
+
"epoch": 0.1468013468013468,
|
| 3066 |
+
"grad_norm": 0.5167519450187683,
|
| 3067 |
+
"learning_rate": 4.266254208754209e-05,
|
| 3068 |
+
"loss": 0.477,
|
| 3069 |
+
"step": 87200
|
| 3070 |
+
},
|
| 3071 |
+
{
|
| 3072 |
+
"epoch": 0.14713804713804715,
|
| 3073 |
+
"grad_norm": 0.3927953243255615,
|
| 3074 |
+
"learning_rate": 4.2645707070707075e-05,
|
| 3075 |
+
"loss": 0.461,
|
| 3076 |
+
"step": 87400
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 0.14747474747474748,
|
| 3080 |
+
"grad_norm": 0.3797866106033325,
|
| 3081 |
+
"learning_rate": 4.262895622895623e-05,
|
| 3082 |
+
"loss": 0.4923,
|
| 3083 |
+
"step": 87600
|
| 3084 |
+
},
|
| 3085 |
+
{
|
| 3086 |
+
"epoch": 0.1478114478114478,
|
| 3087 |
+
"grad_norm": 0.3852689862251282,
|
| 3088 |
+
"learning_rate": 4.2612121212121216e-05,
|
| 3089 |
+
"loss": 0.4254,
|
| 3090 |
+
"step": 87800
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 0.14814814814814814,
|
| 3094 |
+
"grad_norm": 0.4846220016479492,
|
| 3095 |
+
"learning_rate": 4.25952861952862e-05,
|
| 3096 |
+
"loss": 0.4837,
|
| 3097 |
+
"step": 88000
|
| 3098 |
+
},
|
| 3099 |
+
{
|
| 3100 |
+
"epoch": 0.1484848484848485,
|
| 3101 |
+
"grad_norm": 0.7036873698234558,
|
| 3102 |
+
"learning_rate": 4.257845117845118e-05,
|
| 3103 |
+
"loss": 0.4854,
|
| 3104 |
+
"step": 88200
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 0.14882154882154883,
|
| 3108 |
+
"grad_norm": 0.5932942032814026,
|
| 3109 |
+
"learning_rate": 4.256161616161616e-05,
|
| 3110 |
+
"loss": 0.4509,
|
| 3111 |
+
"step": 88400
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 0.14915824915824916,
|
| 3115 |
+
"grad_norm": 0.2926032841205597,
|
| 3116 |
+
"learning_rate": 4.254478114478115e-05,
|
| 3117 |
+
"loss": 0.4805,
|
| 3118 |
+
"step": 88600
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 0.1494949494949495,
|
| 3122 |
+
"grad_norm": 0.16851143538951874,
|
| 3123 |
+
"learning_rate": 4.2527946127946125e-05,
|
| 3124 |
+
"loss": 0.419,
|
| 3125 |
+
"step": 88800
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 0.14983164983164984,
|
| 3129 |
+
"grad_norm": 0.1768457293510437,
|
| 3130 |
+
"learning_rate": 4.2511111111111116e-05,
|
| 3131 |
+
"loss": 0.4432,
|
| 3132 |
+
"step": 89000
|
| 3133 |
+
},
|
| 3134 |
+
{
|
| 3135 |
+
"epoch": 0.15016835016835017,
|
| 3136 |
+
"grad_norm": 1.1971328258514404,
|
| 3137 |
+
"learning_rate": 4.24942760942761e-05,
|
| 3138 |
+
"loss": 0.542,
|
| 3139 |
+
"step": 89200
|
| 3140 |
+
},
|
| 3141 |
+
{
|
| 3142 |
+
"epoch": 0.1505050505050505,
|
| 3143 |
+
"grad_norm": 0.19443285465240479,
|
| 3144 |
+
"learning_rate": 4.247744107744108e-05,
|
| 3145 |
+
"loss": 0.4485,
|
| 3146 |
+
"step": 89400
|
| 3147 |
+
},
|
| 3148 |
+
{
|
| 3149 |
+
"epoch": 0.15084175084175083,
|
| 3150 |
+
"grad_norm": 0.2966189682483673,
|
| 3151 |
+
"learning_rate": 4.246060606060606e-05,
|
| 3152 |
+
"loss": 0.4376,
|
| 3153 |
+
"step": 89600
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 0.1511784511784512,
|
| 3157 |
+
"grad_norm": 0.3715890645980835,
|
| 3158 |
+
"learning_rate": 4.244385521885522e-05,
|
| 3159 |
+
"loss": 0.4862,
|
| 3160 |
+
"step": 89800
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 0.15151515151515152,
|
| 3164 |
+
"grad_norm": 0.2697141468524933,
|
| 3165 |
+
"learning_rate": 4.2427020202020204e-05,
|
| 3166 |
+
"loss": 0.4746,
|
| 3167 |
+
"step": 90000
|
| 3168 |
+
}
|
| 3169 |
+
],
|
| 3170 |
+
"logging_steps": 200,
|
| 3171 |
+
"max_steps": 594000,
|
| 3172 |
+
"num_input_tokens_seen": 0,
|
| 3173 |
+
"num_train_epochs": 1,
|
| 3174 |
+
"save_steps": 10000,
|
| 3175 |
+
"stateful_callbacks": {
|
| 3176 |
+
"TrainerControl": {
|
| 3177 |
+
"args": {
|
| 3178 |
+
"should_epoch_stop": false,
|
| 3179 |
+
"should_evaluate": false,
|
| 3180 |
+
"should_log": false,
|
| 3181 |
+
"should_save": true,
|
| 3182 |
+
"should_training_stop": false
|
| 3183 |
+
},
|
| 3184 |
+
"attributes": {}
|
| 3185 |
+
}
|
| 3186 |
+
},
|
| 3187 |
+
"total_flos": 6.8214306373632e+17,
|
| 3188 |
+
"train_batch_size": 1,
|
| 3189 |
+
"trial_name": null,
|
| 3190 |
+
"trial_params": null
|
| 3191 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b1948218513adbf973cc207e8fa1e02c821194d4fa349dad5f52a5a29e57121a
|
| 3 |
+
size 5048
|