Upload folder using huggingface_hub
Browse files- checkpoint-1900/config.json +17 -0
- checkpoint-1900/generation_config.json +6 -0
- checkpoint-1900/model.safetensors +3 -0
- checkpoint-1900/optimizer.pt +3 -0
- checkpoint-1900/rng_state.pth +3 -0
- checkpoint-1900/scaler.pt +3 -0
- checkpoint-1900/scheduler.pt +3 -0
- checkpoint-1900/trainer_state.json +1364 -0
- checkpoint-1900/training_args.bin +3 -0
- checkpoint-1900/vocab_rank_shared_8k.json +0 -0
checkpoint-1900/config.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"FastHyperNet125mForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"dtype": "float32",
|
| 6 |
+
"hidden_size": 768,
|
| 7 |
+
"hyperbolic_dim": 16,
|
| 8 |
+
"initializer_range": 0.02,
|
| 9 |
+
"intermediate_size": 2048,
|
| 10 |
+
"max_position_embeddings": 512,
|
| 11 |
+
"model_type": "fasthypernet_125m_8k",
|
| 12 |
+
"num_attention_heads": 12,
|
| 13 |
+
"num_hidden_layers": 16,
|
| 14 |
+
"transformers_version": "5.13.1",
|
| 15 |
+
"use_cache": false,
|
| 16 |
+
"vocab_size": 8192
|
| 17 |
+
}
|
checkpoint-1900/generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"output_attentions": false,
|
| 4 |
+
"output_hidden_states": false,
|
| 5 |
+
"transformers_version": "5.13.1"
|
| 6 |
+
}
|
checkpoint-1900/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cc331181ead021088cd93450ecac014c7b9e278a9d97756b8fe0f379ddda51a6
|
| 3 |
+
size 492457480
|
checkpoint-1900/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:253e45a9d5e95da63f03063832e83c55ab56e5c0ed86cfe58e80cd83e95039d5
|
| 3 |
+
size 251142177
|
checkpoint-1900/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96a2fdec9cc7d29c2ce7431337de44b1a8d1b359b9fc5d7442d2178c9c09eef6
|
| 3 |
+
size 14645
|
checkpoint-1900/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4d4e47b8462744d9d75a924a19e7a4f18e9dec4f67a3b1ebbd385645482456c0
|
| 3 |
+
size 1383
|
checkpoint-1900/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96b5dfed1c5cb138b96c9a8651b8df2ba6908fd6d41823d8b67e1c7f7e930c42
|
| 3 |
+
size 1465
|
checkpoint-1900/trainer_state.json
ADDED
|
@@ -0,0 +1,1364 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.010498672355391725,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1900,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.00011665191505990805,
|
| 14 |
+
"grad_norm": 36.77709197998047,
|
| 15 |
+
"learning_rate": 4.8e-07,
|
| 16 |
+
"loss": 8.7532,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0002333038301198161,
|
| 21 |
+
"grad_norm": 27.75691795349121,
|
| 22 |
+
"learning_rate": 1.0799999999999998e-06,
|
| 23 |
+
"loss": 8.7454,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.0003499557451797241,
|
| 28 |
+
"grad_norm": 28.70688247680664,
|
| 29 |
+
"learning_rate": 1.6799999999999998e-06,
|
| 30 |
+
"loss": 8.7275,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.0004666076602396322,
|
| 35 |
+
"grad_norm": 30.075843811035156,
|
| 36 |
+
"learning_rate": 2.2799999999999998e-06,
|
| 37 |
+
"loss": 8.7324,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.0005832595752995402,
|
| 42 |
+
"grad_norm": 24.32572364807129,
|
| 43 |
+
"learning_rate": 2.8799999999999995e-06,
|
| 44 |
+
"loss": 8.7301,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.0006999114903594483,
|
| 49 |
+
"grad_norm": 20.42615509033203,
|
| 50 |
+
"learning_rate": 3.4799999999999993e-06,
|
| 51 |
+
"loss": 8.7192,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.0008165634054193564,
|
| 56 |
+
"grad_norm": 21.43255043029785,
|
| 57 |
+
"learning_rate": 4.079999999999999e-06,
|
| 58 |
+
"loss": 8.7223,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.0009332153204792644,
|
| 63 |
+
"grad_norm": 15.917525291442871,
|
| 64 |
+
"learning_rate": 4.679999999999999e-06,
|
| 65 |
+
"loss": 8.7192,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.0010498672355391725,
|
| 70 |
+
"grad_norm": 18.44275665283203,
|
| 71 |
+
"learning_rate": 5.28e-06,
|
| 72 |
+
"loss": 8.7243,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.0011665191505990805,
|
| 77 |
+
"grad_norm": 23.859643936157227,
|
| 78 |
+
"learning_rate": 5.88e-06,
|
| 79 |
+
"loss": 8.7202,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.0012831710656589885,
|
| 84 |
+
"grad_norm": 25.342121124267578,
|
| 85 |
+
"learning_rate": 6.48e-06,
|
| 86 |
+
"loss": 8.7108,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.0013998229807188965,
|
| 91 |
+
"grad_norm": 31.435964584350586,
|
| 92 |
+
"learning_rate": 7.079999999999999e-06,
|
| 93 |
+
"loss": 8.7104,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.0015164748957788047,
|
| 98 |
+
"grad_norm": 28.961843490600586,
|
| 99 |
+
"learning_rate": 7.68e-06,
|
| 100 |
+
"loss": 8.7152,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.0016331268108387127,
|
| 105 |
+
"grad_norm": 34.92390060424805,
|
| 106 |
+
"learning_rate": 8.279999999999999e-06,
|
| 107 |
+
"loss": 8.7094,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.0017497787258986207,
|
| 112 |
+
"grad_norm": 33.02936553955078,
|
| 113 |
+
"learning_rate": 8.88e-06,
|
| 114 |
+
"loss": 8.6974,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.0018664306409585287,
|
| 119 |
+
"grad_norm": 32.488067626953125,
|
| 120 |
+
"learning_rate": 9.48e-06,
|
| 121 |
+
"loss": 8.6996,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0019830825560184367,
|
| 126 |
+
"grad_norm": 23.191457748413086,
|
| 127 |
+
"learning_rate": 1.0079999999999998e-05,
|
| 128 |
+
"loss": 8.7105,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.002099734471078345,
|
| 133 |
+
"grad_norm": 106.55644226074219,
|
| 134 |
+
"learning_rate": 1.068e-05,
|
| 135 |
+
"loss": 8.6988,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.0022163863861382528,
|
| 140 |
+
"grad_norm": 38.405574798583984,
|
| 141 |
+
"learning_rate": 1.1279999999999999e-05,
|
| 142 |
+
"loss": 8.7015,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.002333038301198161,
|
| 147 |
+
"grad_norm": 12.893147468566895,
|
| 148 |
+
"learning_rate": 1.188e-05,
|
| 149 |
+
"loss": 8.6945,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.002449690216258069,
|
| 154 |
+
"grad_norm": 98.10584259033203,
|
| 155 |
+
"learning_rate": 1.2479999999999999e-05,
|
| 156 |
+
"loss": 8.6942,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.002566342131317977,
|
| 161 |
+
"grad_norm": 14.490903854370117,
|
| 162 |
+
"learning_rate": 1.3079999999999998e-05,
|
| 163 |
+
"loss": 8.6946,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.002682994046377885,
|
| 168 |
+
"grad_norm": 13.219086647033691,
|
| 169 |
+
"learning_rate": 1.368e-05,
|
| 170 |
+
"loss": 8.7053,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.002799645961437793,
|
| 175 |
+
"grad_norm": 16.4891357421875,
|
| 176 |
+
"learning_rate": 1.428e-05,
|
| 177 |
+
"loss": 8.6833,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.0029162978764977012,
|
| 182 |
+
"grad_norm": 20.30918312072754,
|
| 183 |
+
"learning_rate": 1.4879999999999998e-05,
|
| 184 |
+
"loss": 8.6788,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.0030329497915576095,
|
| 189 |
+
"grad_norm": 47.12828063964844,
|
| 190 |
+
"learning_rate": 1.5419999999999998e-05,
|
| 191 |
+
"loss": 8.6797,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.0031496017066175172,
|
| 196 |
+
"grad_norm": 13.309244155883789,
|
| 197 |
+
"learning_rate": 1.602e-05,
|
| 198 |
+
"loss": 8.6818,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.0032662536216774255,
|
| 203 |
+
"grad_norm": 20.72736358642578,
|
| 204 |
+
"learning_rate": 1.6619999999999997e-05,
|
| 205 |
+
"loss": 8.6939,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.0033829055367373333,
|
| 210 |
+
"grad_norm": 13.736023902893066,
|
| 211 |
+
"learning_rate": 1.7219999999999998e-05,
|
| 212 |
+
"loss": 8.6678,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.0034995574517972415,
|
| 217 |
+
"grad_norm": 19.503042221069336,
|
| 218 |
+
"learning_rate": 1.782e-05,
|
| 219 |
+
"loss": 8.6681,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.0036162093668571497,
|
| 224 |
+
"grad_norm": 14.752237319946289,
|
| 225 |
+
"learning_rate": 1.842e-05,
|
| 226 |
+
"loss": 8.6768,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0037328612819170575,
|
| 231 |
+
"grad_norm": 25.988895416259766,
|
| 232 |
+
"learning_rate": 1.9019999999999997e-05,
|
| 233 |
+
"loss": 8.6562,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.0038495131969769657,
|
| 238 |
+
"grad_norm": 18.02843475341797,
|
| 239 |
+
"learning_rate": 1.962e-05,
|
| 240 |
+
"loss": 8.6626,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.0039661651120368735,
|
| 245 |
+
"grad_norm": 11.028422355651855,
|
| 246 |
+
"learning_rate": 2.022e-05,
|
| 247 |
+
"loss": 8.6661,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.004082817027096782,
|
| 252 |
+
"grad_norm": 15.294510841369629,
|
| 253 |
+
"learning_rate": 2.082e-05,
|
| 254 |
+
"loss": 8.6603,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.00419946894215669,
|
| 259 |
+
"grad_norm": 19.6163272857666,
|
| 260 |
+
"learning_rate": 2.1419999999999998e-05,
|
| 261 |
+
"loss": 8.6617,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.004316120857216598,
|
| 266 |
+
"grad_norm": 29.14927864074707,
|
| 267 |
+
"learning_rate": 2.202e-05,
|
| 268 |
+
"loss": 8.6559,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.0044327727722765055,
|
| 273 |
+
"grad_norm": 18.308366775512695,
|
| 274 |
+
"learning_rate": 2.2619999999999997e-05,
|
| 275 |
+
"loss": 8.6531,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.004549424687336414,
|
| 280 |
+
"grad_norm": 14.790651321411133,
|
| 281 |
+
"learning_rate": 2.3219999999999998e-05,
|
| 282 |
+
"loss": 8.6526,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.004666076602396322,
|
| 287 |
+
"grad_norm": 22.447404861450195,
|
| 288 |
+
"learning_rate": 2.382e-05,
|
| 289 |
+
"loss": 8.6479,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.00478272851745623,
|
| 294 |
+
"grad_norm": 48.76446533203125,
|
| 295 |
+
"learning_rate": 2.4419999999999997e-05,
|
| 296 |
+
"loss": 8.6488,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.004899380432516138,
|
| 301 |
+
"grad_norm": 14.55058479309082,
|
| 302 |
+
"learning_rate": 2.5019999999999998e-05,
|
| 303 |
+
"loss": 8.6483,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.005016032347576046,
|
| 308 |
+
"grad_norm": 14.423152923583984,
|
| 309 |
+
"learning_rate": 2.562e-05,
|
| 310 |
+
"loss": 8.6486,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.005132684262635954,
|
| 315 |
+
"grad_norm": 15.452584266662598,
|
| 316 |
+
"learning_rate": 2.622e-05,
|
| 317 |
+
"loss": 8.6413,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.005249336177695863,
|
| 322 |
+
"grad_norm": 23.887699127197266,
|
| 323 |
+
"learning_rate": 2.6819999999999997e-05,
|
| 324 |
+
"loss": 8.6269,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.00536598809275577,
|
| 329 |
+
"grad_norm": 36.94230651855469,
|
| 330 |
+
"learning_rate": 2.7419999999999995e-05,
|
| 331 |
+
"loss": 8.6358,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.005482640007815678,
|
| 336 |
+
"grad_norm": 12.526345252990723,
|
| 337 |
+
"learning_rate": 2.8019999999999996e-05,
|
| 338 |
+
"loss": 8.6329,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.005599291922875586,
|
| 343 |
+
"grad_norm": 14.956578254699707,
|
| 344 |
+
"learning_rate": 2.8619999999999997e-05,
|
| 345 |
+
"loss": 8.6493,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.005715943837935495,
|
| 350 |
+
"grad_norm": 30.98826026916504,
|
| 351 |
+
"learning_rate": 2.9219999999999998e-05,
|
| 352 |
+
"loss": 8.629,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.0058325957529954025,
|
| 357 |
+
"grad_norm": 17.767681121826172,
|
| 358 |
+
"learning_rate": 2.982e-05,
|
| 359 |
+
"loss": 8.6195,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.00594924766805531,
|
| 364 |
+
"grad_norm": 11.531399726867676,
|
| 365 |
+
"learning_rate": 3.042e-05,
|
| 366 |
+
"loss": 8.6257,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.006065899583115219,
|
| 371 |
+
"grad_norm": 18.6702880859375,
|
| 372 |
+
"learning_rate": 3.102e-05,
|
| 373 |
+
"loss": 8.6313,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.006182551498175127,
|
| 378 |
+
"grad_norm": 17.597599029541016,
|
| 379 |
+
"learning_rate": 3.161999999999999e-05,
|
| 380 |
+
"loss": 8.6267,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.0062992034132350345,
|
| 385 |
+
"grad_norm": 13.608710289001465,
|
| 386 |
+
"learning_rate": 3.221999999999999e-05,
|
| 387 |
+
"loss": 8.6264,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.006415855328294943,
|
| 392 |
+
"grad_norm": 19.217361450195312,
|
| 393 |
+
"learning_rate": 3.2819999999999994e-05,
|
| 394 |
+
"loss": 8.624,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.006532507243354851,
|
| 399 |
+
"grad_norm": 17.15315818786621,
|
| 400 |
+
"learning_rate": 3.3419999999999995e-05,
|
| 401 |
+
"loss": 8.5997,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.006649159158414759,
|
| 406 |
+
"grad_norm": 13.051802635192871,
|
| 407 |
+
"learning_rate": 3.4019999999999996e-05,
|
| 408 |
+
"loss": 8.6134,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.0067658110734746665,
|
| 413 |
+
"grad_norm": 14.354033470153809,
|
| 414 |
+
"learning_rate": 3.462e-05,
|
| 415 |
+
"loss": 8.6168,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.006882462988534575,
|
| 420 |
+
"grad_norm": 11.711945533752441,
|
| 421 |
+
"learning_rate": 3.522e-05,
|
| 422 |
+
"loss": 8.6151,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.006999114903594483,
|
| 427 |
+
"grad_norm": 19.64753532409668,
|
| 428 |
+
"learning_rate": 3.582e-05,
|
| 429 |
+
"loss": 8.6035,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.007115766818654391,
|
| 434 |
+
"grad_norm": 17.521629333496094,
|
| 435 |
+
"learning_rate": 3.6419999999999994e-05,
|
| 436 |
+
"loss": 8.6192,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.007232418733714299,
|
| 441 |
+
"grad_norm": 29.723106384277344,
|
| 442 |
+
"learning_rate": 3.7019999999999995e-05,
|
| 443 |
+
"loss": 8.5963,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.007349070648774207,
|
| 448 |
+
"grad_norm": 14.468299865722656,
|
| 449 |
+
"learning_rate": 3.762e-05,
|
| 450 |
+
"loss": 8.6192,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.007465722563834115,
|
| 455 |
+
"grad_norm": 13.490720748901367,
|
| 456 |
+
"learning_rate": 3.822e-05,
|
| 457 |
+
"loss": 8.6046,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.007582374478894024,
|
| 462 |
+
"grad_norm": 29.188461303710938,
|
| 463 |
+
"learning_rate": 3.881999999999999e-05,
|
| 464 |
+
"loss": 8.6089,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.007699026393953931,
|
| 469 |
+
"grad_norm": 13.968728065490723,
|
| 470 |
+
"learning_rate": 3.941999999999999e-05,
|
| 471 |
+
"loss": 8.6093,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.00781567830901384,
|
| 476 |
+
"grad_norm": Infinity,
|
| 477 |
+
"learning_rate": 4.001999999999999e-05,
|
| 478 |
+
"loss": 8.6052,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.007932330224073747,
|
| 483 |
+
"grad_norm": 17.822641372680664,
|
| 484 |
+
"learning_rate": 4.055999999999999e-05,
|
| 485 |
+
"loss": 8.6105,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.008048982139133656,
|
| 490 |
+
"grad_norm": 73.21751403808594,
|
| 491 |
+
"learning_rate": 4.115999999999999e-05,
|
| 492 |
+
"loss": 8.5937,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.008165634054193564,
|
| 497 |
+
"grad_norm": 11.047343254089355,
|
| 498 |
+
"learning_rate": 4.175999999999999e-05,
|
| 499 |
+
"loss": 8.5928,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.008282285969253471,
|
| 504 |
+
"grad_norm": 18.778337478637695,
|
| 505 |
+
"learning_rate": 4.2359999999999994e-05,
|
| 506 |
+
"loss": 8.5865,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.00839893788431338,
|
| 511 |
+
"grad_norm": 28.407215118408203,
|
| 512 |
+
"learning_rate": 4.2959999999999995e-05,
|
| 513 |
+
"loss": 8.5852,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.008515589799373287,
|
| 518 |
+
"grad_norm": 8.337827682495117,
|
| 519 |
+
"learning_rate": 4.3559999999999996e-05,
|
| 520 |
+
"loss": 8.5887,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.008632241714433195,
|
| 525 |
+
"grad_norm": 24.844894409179688,
|
| 526 |
+
"learning_rate": 4.416e-05,
|
| 527 |
+
"loss": 8.59,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.008748893629493104,
|
| 532 |
+
"grad_norm": 8.051148414611816,
|
| 533 |
+
"learning_rate": 4.476e-05,
|
| 534 |
+
"loss": 8.5943,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.008865545544553011,
|
| 539 |
+
"grad_norm": 9.884084701538086,
|
| 540 |
+
"learning_rate": 4.536e-05,
|
| 541 |
+
"loss": 8.5933,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.00898219745961292,
|
| 546 |
+
"grad_norm": 14.391705513000488,
|
| 547 |
+
"learning_rate": 4.5959999999999994e-05,
|
| 548 |
+
"loss": 8.5911,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.009098849374672828,
|
| 553 |
+
"grad_norm": 8.84749984741211,
|
| 554 |
+
"learning_rate": 4.6559999999999995e-05,
|
| 555 |
+
"loss": 8.5882,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.009215501289732735,
|
| 560 |
+
"grad_norm": 18.035995483398438,
|
| 561 |
+
"learning_rate": 4.7159999999999996e-05,
|
| 562 |
+
"loss": 8.5933,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.009332153204792644,
|
| 567 |
+
"grad_norm": 18.23465919494629,
|
| 568 |
+
"learning_rate": 4.776e-05,
|
| 569 |
+
"loss": 8.5907,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.009448805119852553,
|
| 574 |
+
"grad_norm": 14.406664848327637,
|
| 575 |
+
"learning_rate": 4.836e-05,
|
| 576 |
+
"loss": 8.5822,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.00956545703491246,
|
| 581 |
+
"grad_norm": 17.29925537109375,
|
| 582 |
+
"learning_rate": 4.896e-05,
|
| 583 |
+
"loss": 8.6041,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.009682108949972368,
|
| 588 |
+
"grad_norm": 13.491657257080078,
|
| 589 |
+
"learning_rate": 4.956e-05,
|
| 590 |
+
"loss": 8.5914,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.009798760865032277,
|
| 595 |
+
"grad_norm": 10.702431678771973,
|
| 596 |
+
"learning_rate": 5.0159999999999994e-05,
|
| 597 |
+
"loss": 8.5695,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.009915412780092184,
|
| 602 |
+
"grad_norm": 21.955284118652344,
|
| 603 |
+
"learning_rate": 5.0759999999999995e-05,
|
| 604 |
+
"loss": 8.576,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.010032064695152092,
|
| 609 |
+
"grad_norm": 40.40593719482422,
|
| 610 |
+
"learning_rate": 5.1359999999999996e-05,
|
| 611 |
+
"loss": 8.5869,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.010148716610212001,
|
| 616 |
+
"grad_norm": 13.60511302947998,
|
| 617 |
+
"learning_rate": 5.195999999999999e-05,
|
| 618 |
+
"loss": 8.5815,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.010265368525271908,
|
| 623 |
+
"grad_norm": 7.648057460784912,
|
| 624 |
+
"learning_rate": 5.255999999999999e-05,
|
| 625 |
+
"loss": 8.5826,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.010382020440331817,
|
| 630 |
+
"grad_norm": 11.52419376373291,
|
| 631 |
+
"learning_rate": 5.315999999999999e-05,
|
| 632 |
+
"loss": 8.5609,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.010498672355391725,
|
| 637 |
+
"grad_norm": 14.577630996704102,
|
| 638 |
+
"learning_rate": 5.3759999999999994e-05,
|
| 639 |
+
"loss": 8.5763,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.010615324270451632,
|
| 644 |
+
"grad_norm": 9.24594497680664,
|
| 645 |
+
"learning_rate": 5.4359999999999995e-05,
|
| 646 |
+
"loss": 8.5722,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.01073197618551154,
|
| 651 |
+
"grad_norm": 10.759099960327148,
|
| 652 |
+
"learning_rate": 5.4959999999999996e-05,
|
| 653 |
+
"loss": 8.583,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.010848628100571448,
|
| 658 |
+
"grad_norm": 14.04236125946045,
|
| 659 |
+
"learning_rate": 5.5559999999999997e-05,
|
| 660 |
+
"loss": 8.5536,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.010965280015631356,
|
| 665 |
+
"grad_norm": 9.82194995880127,
|
| 666 |
+
"learning_rate": 5.616e-05,
|
| 667 |
+
"loss": 8.5698,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.011081931930691265,
|
| 672 |
+
"grad_norm": 8.623485565185547,
|
| 673 |
+
"learning_rate": 5.676e-05,
|
| 674 |
+
"loss": 8.574,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.011198583845751172,
|
| 679 |
+
"grad_norm": 7.429067611694336,
|
| 680 |
+
"learning_rate": 5.736e-05,
|
| 681 |
+
"loss": 8.5729,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.01131523576081108,
|
| 686 |
+
"grad_norm": 15.89907455444336,
|
| 687 |
+
"learning_rate": 5.796e-05,
|
| 688 |
+
"loss": 8.5556,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.01143188767587099,
|
| 693 |
+
"grad_norm": 13.150750160217285,
|
| 694 |
+
"learning_rate": 5.856e-05,
|
| 695 |
+
"loss": 8.5811,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.011548539590930896,
|
| 700 |
+
"grad_norm": 10.560754776000977,
|
| 701 |
+
"learning_rate": 5.915999999999999e-05,
|
| 702 |
+
"loss": 8.5705,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.011665191505990805,
|
| 707 |
+
"grad_norm": 9.445666313171387,
|
| 708 |
+
"learning_rate": 5.975999999999999e-05,
|
| 709 |
+
"loss": 8.5764,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.00011665191505990805,
|
| 714 |
+
"grad_norm": 6.930628299713135,
|
| 715 |
+
"learning_rate": 6.035999999999999e-05,
|
| 716 |
+
"loss": 8.7318,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.0002333038301198161,
|
| 721 |
+
"grad_norm": 8.697470664978027,
|
| 722 |
+
"learning_rate": 6.095999999999999e-05,
|
| 723 |
+
"loss": 8.7328,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.0003499557451797241,
|
| 728 |
+
"grad_norm": 10.648292541503906,
|
| 729 |
+
"learning_rate": 6.155999999999999e-05,
|
| 730 |
+
"loss": 8.7367,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.0004666076602396322,
|
| 735 |
+
"grad_norm": 7.450256824493408,
|
| 736 |
+
"learning_rate": 6.215999999999999e-05,
|
| 737 |
+
"loss": 8.7304,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.0005832595752995402,
|
| 742 |
+
"grad_norm": 6.228799343109131,
|
| 743 |
+
"learning_rate": 6.275999999999999e-05,
|
| 744 |
+
"loss": 8.7352,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.0006999114903594483,
|
| 749 |
+
"grad_norm": 4.954726696014404,
|
| 750 |
+
"learning_rate": 6.335999999999999e-05,
|
| 751 |
+
"loss": 8.7285,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.0008165634054193564,
|
| 756 |
+
"grad_norm": 5.833169460296631,
|
| 757 |
+
"learning_rate": 6.395999999999999e-05,
|
| 758 |
+
"loss": 8.7358,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.0009332153204792644,
|
| 763 |
+
"grad_norm": 7.297743320465088,
|
| 764 |
+
"learning_rate": 6.455999999999999e-05,
|
| 765 |
+
"loss": 8.7242,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.0010498672355391725,
|
| 770 |
+
"grad_norm": 7.619863510131836,
|
| 771 |
+
"learning_rate": 6.515999999999999e-05,
|
| 772 |
+
"loss": 8.7348,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.0011665191505990805,
|
| 777 |
+
"grad_norm": 8.537040710449219,
|
| 778 |
+
"learning_rate": 6.576e-05,
|
| 779 |
+
"loss": 8.7257,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.0012831710656589885,
|
| 784 |
+
"grad_norm": 6.015471935272217,
|
| 785 |
+
"learning_rate": 6.636e-05,
|
| 786 |
+
"loss": 8.714,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.0013998229807188965,
|
| 791 |
+
"grad_norm": 6.178020477294922,
|
| 792 |
+
"learning_rate": 6.696e-05,
|
| 793 |
+
"loss": 8.7229,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.0015164748957788047,
|
| 798 |
+
"grad_norm": 7.1294708251953125,
|
| 799 |
+
"learning_rate": 6.756e-05,
|
| 800 |
+
"loss": 8.7224,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.0016331268108387127,
|
| 805 |
+
"grad_norm": 6.2934041023254395,
|
| 806 |
+
"learning_rate": 6.816e-05,
|
| 807 |
+
"loss": 8.7202,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.0017497787258986207,
|
| 812 |
+
"grad_norm": 5.8199543952941895,
|
| 813 |
+
"learning_rate": 6.875999999999999e-05,
|
| 814 |
+
"loss": 8.7148,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.0018664306409585287,
|
| 819 |
+
"grad_norm": 13.640084266662598,
|
| 820 |
+
"learning_rate": 6.935999999999999e-05,
|
| 821 |
+
"loss": 8.7206,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.0019830825560184367,
|
| 826 |
+
"grad_norm": 8.563558578491211,
|
| 827 |
+
"learning_rate": 6.995999999999999e-05,
|
| 828 |
+
"loss": 8.7312,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.002099734471078345,
|
| 833 |
+
"grad_norm": 8.236286163330078,
|
| 834 |
+
"learning_rate": 7.055999999999999e-05,
|
| 835 |
+
"loss": 8.7248,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.0022163863861382528,
|
| 840 |
+
"grad_norm": 10.796252250671387,
|
| 841 |
+
"learning_rate": 7.115999999999999e-05,
|
| 842 |
+
"loss": 8.7228,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.002333038301198161,
|
| 847 |
+
"grad_norm": 9.347583770751953,
|
| 848 |
+
"learning_rate": 7.175999999999999e-05,
|
| 849 |
+
"loss": 8.7131,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.002449690216258069,
|
| 854 |
+
"grad_norm": 7.83629846572876,
|
| 855 |
+
"learning_rate": 7.235999999999999e-05,
|
| 856 |
+
"loss": 8.7167,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.002566342131317977,
|
| 861 |
+
"grad_norm": 5.277032375335693,
|
| 862 |
+
"learning_rate": 7.295999999999999e-05,
|
| 863 |
+
"loss": 8.7153,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.002682994046377885,
|
| 868 |
+
"grad_norm": 7.31398868560791,
|
| 869 |
+
"learning_rate": 7.356e-05,
|
| 870 |
+
"loss": 8.7282,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.002799645961437793,
|
| 875 |
+
"grad_norm": 5.442142486572266,
|
| 876 |
+
"learning_rate": 7.416e-05,
|
| 877 |
+
"loss": 8.7111,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.0029162978764977012,
|
| 882 |
+
"grad_norm": 8.926863670349121,
|
| 883 |
+
"learning_rate": 7.476e-05,
|
| 884 |
+
"loss": 8.704,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.0030329497915576095,
|
| 889 |
+
"grad_norm": 10.089258193969727,
|
| 890 |
+
"learning_rate": 7.535999999999998e-05,
|
| 891 |
+
"loss": 8.7048,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.0031496017066175172,
|
| 896 |
+
"grad_norm": 7.243823528289795,
|
| 897 |
+
"learning_rate": 7.595999999999998e-05,
|
| 898 |
+
"loss": 8.7106,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.0032662536216774255,
|
| 903 |
+
"grad_norm": 8.783164024353027,
|
| 904 |
+
"learning_rate": 7.655999999999999e-05,
|
| 905 |
+
"loss": 8.7206,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.0033829055367373333,
|
| 910 |
+
"grad_norm": 5.882843494415283,
|
| 911 |
+
"learning_rate": 7.715999999999999e-05,
|
| 912 |
+
"loss": 8.6971,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.0034995574517972415,
|
| 917 |
+
"grad_norm": 18.633825302124023,
|
| 918 |
+
"learning_rate": 7.775999999999999e-05,
|
| 919 |
+
"loss": 8.7025,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.0036162093668571497,
|
| 924 |
+
"grad_norm": 5.833542823791504,
|
| 925 |
+
"learning_rate": 7.835999999999999e-05,
|
| 926 |
+
"loss": 8.7141,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.0037328612819170575,
|
| 931 |
+
"grad_norm": 21.012325286865234,
|
| 932 |
+
"learning_rate": 7.895999999999999e-05,
|
| 933 |
+
"loss": 8.6972,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.0038495131969769657,
|
| 938 |
+
"grad_norm": 6.614134311676025,
|
| 939 |
+
"learning_rate": 7.955999999999999e-05,
|
| 940 |
+
"loss": 8.7052,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.0039661651120368735,
|
| 945 |
+
"grad_norm": 5.228976249694824,
|
| 946 |
+
"learning_rate": 8.015999999999999e-05,
|
| 947 |
+
"loss": 8.7096,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.004082817027096782,
|
| 952 |
+
"grad_norm": 7.862325668334961,
|
| 953 |
+
"learning_rate": 8.075999999999999e-05,
|
| 954 |
+
"loss": 8.6996,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.00419946894215669,
|
| 959 |
+
"grad_norm": 6.079459190368652,
|
| 960 |
+
"learning_rate": 8.136e-05,
|
| 961 |
+
"loss": 8.7081,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.004316120857216598,
|
| 966 |
+
"grad_norm": 6.500232696533203,
|
| 967 |
+
"learning_rate": 8.196e-05,
|
| 968 |
+
"loss": 8.7051,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.0044327727722765055,
|
| 973 |
+
"grad_norm": 7.602541446685791,
|
| 974 |
+
"learning_rate": 8.256e-05,
|
| 975 |
+
"loss": 8.6998,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.004549424687336414,
|
| 980 |
+
"grad_norm": 6.5400800704956055,
|
| 981 |
+
"learning_rate": 8.316e-05,
|
| 982 |
+
"loss": 8.7057,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.004666076602396322,
|
| 987 |
+
"grad_norm": 10.45396900177002,
|
| 988 |
+
"learning_rate": 8.376e-05,
|
| 989 |
+
"loss": 8.7008,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.00478272851745623,
|
| 994 |
+
"grad_norm": 11.86343002319336,
|
| 995 |
+
"learning_rate": 8.436e-05,
|
| 996 |
+
"loss": 8.7078,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.004899380432516138,
|
| 1001 |
+
"grad_norm": 6.762137413024902,
|
| 1002 |
+
"learning_rate": 8.496e-05,
|
| 1003 |
+
"loss": 8.7003,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.005016032347576046,
|
| 1008 |
+
"grad_norm": 6.732534408569336,
|
| 1009 |
+
"learning_rate": 8.556e-05,
|
| 1010 |
+
"loss": 8.7064,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.005132684262635954,
|
| 1015 |
+
"grad_norm": 4.935892581939697,
|
| 1016 |
+
"learning_rate": 8.616e-05,
|
| 1017 |
+
"loss": 8.7006,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.005249336177695863,
|
| 1022 |
+
"grad_norm": 5.864833831787109,
|
| 1023 |
+
"learning_rate": 8.676e-05,
|
| 1024 |
+
"loss": 8.6891,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.00536598809275577,
|
| 1029 |
+
"grad_norm": 9.068333625793457,
|
| 1030 |
+
"learning_rate": 8.735999999999999e-05,
|
| 1031 |
+
"loss": 8.6943,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.005482640007815678,
|
| 1036 |
+
"grad_norm": 6.341752529144287,
|
| 1037 |
+
"learning_rate": 8.795999999999999e-05,
|
| 1038 |
+
"loss": 8.6945,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.005599291922875586,
|
| 1043 |
+
"grad_norm": 6.8090362548828125,
|
| 1044 |
+
"learning_rate": 8.855999999999999e-05,
|
| 1045 |
+
"loss": 8.7117,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.005715943837935495,
|
| 1050 |
+
"grad_norm": 5.294206619262695,
|
| 1051 |
+
"learning_rate": 8.916e-05,
|
| 1052 |
+
"loss": 8.6916,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.0058325957529954025,
|
| 1057 |
+
"grad_norm": 7.946436405181885,
|
| 1058 |
+
"learning_rate": 8.976e-05,
|
| 1059 |
+
"loss": 8.6905,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.00594924766805531,
|
| 1064 |
+
"grad_norm": 8.320716857910156,
|
| 1065 |
+
"learning_rate": 9.036e-05,
|
| 1066 |
+
"loss": 8.6936,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.006065899583115219,
|
| 1071 |
+
"grad_norm": 5.920083522796631,
|
| 1072 |
+
"learning_rate": 9.096e-05,
|
| 1073 |
+
"loss": 8.7025,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.006182551498175127,
|
| 1078 |
+
"grad_norm": 5.532317161560059,
|
| 1079 |
+
"learning_rate": 9.156e-05,
|
| 1080 |
+
"loss": 8.6954,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.0062992034132350345,
|
| 1085 |
+
"grad_norm": 5.155895709991455,
|
| 1086 |
+
"learning_rate": 9.215999999999998e-05,
|
| 1087 |
+
"loss": 8.6979,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.006415855328294943,
|
| 1092 |
+
"grad_norm": 4.799182415008545,
|
| 1093 |
+
"learning_rate": 9.275999999999999e-05,
|
| 1094 |
+
"loss": 8.7019,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.006532507243354851,
|
| 1099 |
+
"grad_norm": 5.707144737243652,
|
| 1100 |
+
"learning_rate": 9.335999999999999e-05,
|
| 1101 |
+
"loss": 8.6782,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.006649159158414759,
|
| 1106 |
+
"grad_norm": 5.395667552947998,
|
| 1107 |
+
"learning_rate": 9.395999999999999e-05,
|
| 1108 |
+
"loss": 8.6948,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.0067658110734746665,
|
| 1113 |
+
"grad_norm": 10.986604690551758,
|
| 1114 |
+
"learning_rate": 9.455999999999999e-05,
|
| 1115 |
+
"loss": 8.695,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.006882462988534575,
|
| 1120 |
+
"grad_norm": 3.649099588394165,
|
| 1121 |
+
"learning_rate": 9.515999999999999e-05,
|
| 1122 |
+
"loss": 8.6909,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.006999114903594483,
|
| 1127 |
+
"grad_norm": 9.591960906982422,
|
| 1128 |
+
"learning_rate": 9.575999999999999e-05,
|
| 1129 |
+
"loss": 8.6781,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.007115766818654391,
|
| 1134 |
+
"grad_norm": 5.056659698486328,
|
| 1135 |
+
"learning_rate": 9.635999999999999e-05,
|
| 1136 |
+
"loss": 8.6943,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.007232418733714299,
|
| 1141 |
+
"grad_norm": 6.632259368896484,
|
| 1142 |
+
"learning_rate": 9.696e-05,
|
| 1143 |
+
"loss": 8.6745,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.007349070648774207,
|
| 1148 |
+
"grad_norm": 10.13035774230957,
|
| 1149 |
+
"learning_rate": 9.756e-05,
|
| 1150 |
+
"loss": 8.6978,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.007465722563834115,
|
| 1155 |
+
"grad_norm": 7.024849891662598,
|
| 1156 |
+
"learning_rate": 9.816e-05,
|
| 1157 |
+
"loss": 8.6836,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.007582374478894024,
|
| 1162 |
+
"grad_norm": 7.4931464195251465,
|
| 1163 |
+
"learning_rate": 9.875999999999998e-05,
|
| 1164 |
+
"loss": 8.6894,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.007699026393953931,
|
| 1169 |
+
"grad_norm": 4.576127529144287,
|
| 1170 |
+
"learning_rate": 9.935999999999998e-05,
|
| 1171 |
+
"loss": 8.6912,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.00781567830901384,
|
| 1176 |
+
"grad_norm": 4.012307643890381,
|
| 1177 |
+
"learning_rate": 9.995999999999998e-05,
|
| 1178 |
+
"loss": 8.6841,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.007932330224073747,
|
| 1183 |
+
"grad_norm": 3.5759994983673096,
|
| 1184 |
+
"learning_rate": 0.00010055999999999999,
|
| 1185 |
+
"loss": 8.6913,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.008048982139133656,
|
| 1190 |
+
"grad_norm": 5.049357891082764,
|
| 1191 |
+
"learning_rate": 0.00010115999999999999,
|
| 1192 |
+
"loss": 8.6837,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.008165634054193564,
|
| 1197 |
+
"grad_norm": 4.881607532501221,
|
| 1198 |
+
"learning_rate": 0.00010175999999999999,
|
| 1199 |
+
"loss": 8.6786,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.008282285969253471,
|
| 1204 |
+
"grad_norm": 13.859827041625977,
|
| 1205 |
+
"learning_rate": 0.00010235999999999999,
|
| 1206 |
+
"loss": 8.6743,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.00839893788431338,
|
| 1211 |
+
"grad_norm": 6.686564922332764,
|
| 1212 |
+
"learning_rate": 0.00010295999999999999,
|
| 1213 |
+
"loss": 8.6698,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.008515589799373287,
|
| 1218 |
+
"grad_norm": 4.381744861602783,
|
| 1219 |
+
"learning_rate": 0.00010355999999999999,
|
| 1220 |
+
"loss": 8.6765,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.008632241714433195,
|
| 1225 |
+
"grad_norm": 5.210996627807617,
|
| 1226 |
+
"learning_rate": 0.00010415999999999999,
|
| 1227 |
+
"loss": 8.673,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.008748893629493104,
|
| 1232 |
+
"grad_norm": 4.282205104827881,
|
| 1233 |
+
"learning_rate": 0.00010475999999999999,
|
| 1234 |
+
"loss": 8.6817,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.008865545544553011,
|
| 1239 |
+
"grad_norm": 6.5844316482543945,
|
| 1240 |
+
"learning_rate": 0.00010536,
|
| 1241 |
+
"loss": 8.68,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.00898219745961292,
|
| 1246 |
+
"grad_norm": 4.946044445037842,
|
| 1247 |
+
"learning_rate": 0.00010596,
|
| 1248 |
+
"loss": 8.6818,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.009098849374672828,
|
| 1253 |
+
"grad_norm": 4.784733295440674,
|
| 1254 |
+
"learning_rate": 0.00010656,
|
| 1255 |
+
"loss": 8.6767,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.009215501289732735,
|
| 1260 |
+
"grad_norm": 7.529031276702881,
|
| 1261 |
+
"learning_rate": 0.00010716,
|
| 1262 |
+
"loss": 8.6785,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.009332153204792644,
|
| 1267 |
+
"grad_norm": 5.897704601287842,
|
| 1268 |
+
"learning_rate": 0.00010776,
|
| 1269 |
+
"loss": 8.681,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.009448805119852553,
|
| 1274 |
+
"grad_norm": 5.240443706512451,
|
| 1275 |
+
"learning_rate": 0.00010836,
|
| 1276 |
+
"loss": 8.668,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.00956545703491246,
|
| 1281 |
+
"grad_norm": 6.995890140533447,
|
| 1282 |
+
"learning_rate": 0.00010896,
|
| 1283 |
+
"loss": 8.6891,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.009682108949972368,
|
| 1288 |
+
"grad_norm": 4.577095031738281,
|
| 1289 |
+
"learning_rate": 0.00010956,
|
| 1290 |
+
"loss": 8.682,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.009798760865032277,
|
| 1295 |
+
"grad_norm": 5.011054039001465,
|
| 1296 |
+
"learning_rate": 0.00011016,
|
| 1297 |
+
"loss": 8.6627,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.009915412780092184,
|
| 1302 |
+
"grad_norm": 5.332274913787842,
|
| 1303 |
+
"learning_rate": 0.00011075999999999998,
|
| 1304 |
+
"loss": 8.674,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.010032064695152092,
|
| 1309 |
+
"grad_norm": 4.363228797912598,
|
| 1310 |
+
"learning_rate": 0.00011135999999999998,
|
| 1311 |
+
"loss": 8.6738,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.010148716610212001,
|
| 1316 |
+
"grad_norm": 6.195799350738525,
|
| 1317 |
+
"learning_rate": 0.00011195999999999998,
|
| 1318 |
+
"loss": 8.6687,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.010265368525271908,
|
| 1323 |
+
"grad_norm": 5.533764839172363,
|
| 1324 |
+
"learning_rate": 0.00011255999999999998,
|
| 1325 |
+
"loss": 8.6742,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.010382020440331817,
|
| 1330 |
+
"grad_norm": 5.503600597381592,
|
| 1331 |
+
"learning_rate": 0.00011315999999999998,
|
| 1332 |
+
"loss": 8.6589,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.010498672355391725,
|
| 1337 |
+
"grad_norm": 5.144274711608887,
|
| 1338 |
+
"learning_rate": 0.00011375999999999998,
|
| 1339 |
+
"loss": 8.6663,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
}
|
| 1342 |
+
],
|
| 1343 |
+
"logging_steps": 10,
|
| 1344 |
+
"max_steps": 1000000,
|
| 1345 |
+
"num_input_tokens_seen": 0,
|
| 1346 |
+
"num_train_epochs": 12,
|
| 1347 |
+
"save_steps": 100,
|
| 1348 |
+
"stateful_callbacks": {
|
| 1349 |
+
"TrainerControl": {
|
| 1350 |
+
"args": {
|
| 1351 |
+
"should_epoch_stop": false,
|
| 1352 |
+
"should_evaluate": false,
|
| 1353 |
+
"should_log": false,
|
| 1354 |
+
"should_save": true,
|
| 1355 |
+
"should_training_stop": false
|
| 1356 |
+
},
|
| 1357 |
+
"attributes": {}
|
| 1358 |
+
}
|
| 1359 |
+
},
|
| 1360 |
+
"total_flos": 1.7347323186266112e+16,
|
| 1361 |
+
"train_batch_size": 8,
|
| 1362 |
+
"trial_name": null,
|
| 1363 |
+
"trial_params": null
|
| 1364 |
+
}
|
checkpoint-1900/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68b170d18889101cba0d6a5dd19192f31549fbb1b1fc60a4d9f3703761083c42
|
| 3 |
+
size 5201
|
checkpoint-1900/vocab_rank_shared_8k.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|