Upload folder using huggingface_hub
Browse files- checkpoint-2400/config.json +17 -0
- checkpoint-2400/generation_config.json +6 -0
- checkpoint-2400/hybrid_tokenizer_mapping.json +0 -0
- checkpoint-2400/model.safetensors +3 -0
- checkpoint-2400/optimizer.pt +3 -0
- checkpoint-2400/rng_state.pth +3 -0
- checkpoint-2400/scaler.pt +3 -0
- checkpoint-2400/scheduler.pt +3 -0
- checkpoint-2400/tokenizer_config.json +9 -0
- checkpoint-2400/trainer_state.json +1714 -0
- checkpoint-2400/training_args.bin +3 -0
checkpoint-2400/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_tagless",
|
| 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-2400/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-2400/hybrid_tokenizer_mapping.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
checkpoint-2400/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bfef5e9eceeecb0aab6014b8be6303176537d0f361a6f1c11865c2ee5e816c49
|
| 3 |
+
size 501755480
|
checkpoint-2400/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:20e7e8a6240f7cecb913e63d487adf32bf9efc21d88494f187ed733d5e9fba4f
|
| 3 |
+
size 254410983
|
checkpoint-2400/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:deff96684f5e3e4e495036d529d56112ceca86c104dad04a47c7478c552b2202
|
| 3 |
+
size 14645
|
checkpoint-2400/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0213625bdd454c34ffe243020d1693c7405a317aa7e5804773886aa8071db518
|
| 3 |
+
size 1383
|
checkpoint-2400/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:88393b9666b344b64f6dc32c6d1f493a025993601e065480a86e2607afe89b7b
|
| 3 |
+
size 1465
|
checkpoint-2400/tokenizer_config.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"tokenizer_class": "HFCompatibleHybridTokenizer",
|
| 3 |
+
"vocab_size": 8192,
|
| 4 |
+
"model_max_length": 512,
|
| 5 |
+
"bos_token": "<bos>",
|
| 6 |
+
"eos_token": "<eos>",
|
| 7 |
+
"unk_token": "<unk>",
|
| 8 |
+
"pad_token": "<pad>"
|
| 9 |
+
}
|
checkpoint-2400/trainer_state.json
ADDED
|
@@ -0,0 +1,1714 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.36921656859351565,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2400,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0015384023691396484,
|
| 14 |
+
"grad_norm": 15216.6796875,
|
| 15 |
+
"learning_rate": 1.9999999999999996e-07,
|
| 16 |
+
"loss": 7.628,
|
| 17 |
+
"step": 10
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.003076804738279297,
|
| 21 |
+
"grad_norm": 539.3572387695312,
|
| 22 |
+
"learning_rate": 1.2e-06,
|
| 23 |
+
"loss": 7.3644,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.004615207107418946,
|
| 28 |
+
"grad_norm": 496.3940734863281,
|
| 29 |
+
"learning_rate": 2.1999999999999997e-06,
|
| 30 |
+
"loss": 7.0137,
|
| 31 |
+
"step": 30
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.006153609476558594,
|
| 35 |
+
"grad_norm": 110.5931396484375,
|
| 36 |
+
"learning_rate": 3.1999999999999994e-06,
|
| 37 |
+
"loss": 6.9299,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.007692011845698243,
|
| 42 |
+
"grad_norm": 442.7750549316406,
|
| 43 |
+
"learning_rate": 4.2e-06,
|
| 44 |
+
"loss": 6.88,
|
| 45 |
+
"step": 50
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.009230414214837892,
|
| 49 |
+
"grad_norm": 285.75390625,
|
| 50 |
+
"learning_rate": 5.199999999999999e-06,
|
| 51 |
+
"loss": 6.7252,
|
| 52 |
+
"step": 60
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.010768816583977539,
|
| 56 |
+
"grad_norm": 238.201416015625,
|
| 57 |
+
"learning_rate": 6.199999999999999e-06,
|
| 58 |
+
"loss": 6.7641,
|
| 59 |
+
"step": 70
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.012307218953117188,
|
| 63 |
+
"grad_norm": 203.7633514404297,
|
| 64 |
+
"learning_rate": 7.2e-06,
|
| 65 |
+
"loss": 6.6961,
|
| 66 |
+
"step": 80
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.013845621322256836,
|
| 70 |
+
"grad_norm": 537.8616943359375,
|
| 71 |
+
"learning_rate": 8.2e-06,
|
| 72 |
+
"loss": 6.5996,
|
| 73 |
+
"step": 90
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.015384023691396485,
|
| 77 |
+
"grad_norm": 643.6299438476562,
|
| 78 |
+
"learning_rate": 9.199999999999998e-06,
|
| 79 |
+
"loss": 6.5856,
|
| 80 |
+
"step": 100
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.016922426060536132,
|
| 84 |
+
"grad_norm": 679.996337890625,
|
| 85 |
+
"learning_rate": 1.02e-05,
|
| 86 |
+
"loss": 6.599,
|
| 87 |
+
"step": 110
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.018460828429675783,
|
| 91 |
+
"grad_norm": 452.3795471191406,
|
| 92 |
+
"learning_rate": 1.12e-05,
|
| 93 |
+
"loss": 6.6354,
|
| 94 |
+
"step": 120
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.01999923079881543,
|
| 98 |
+
"grad_norm": 117.24376678466797,
|
| 99 |
+
"learning_rate": 1.2199999999999998e-05,
|
| 100 |
+
"loss": 6.5363,
|
| 101 |
+
"step": 130
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.021537633167955077,
|
| 105 |
+
"grad_norm": 145.65036010742188,
|
| 106 |
+
"learning_rate": 1.3199999999999997e-05,
|
| 107 |
+
"loss": 6.5564,
|
| 108 |
+
"step": 140
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.023076035537094728,
|
| 112 |
+
"grad_norm": 178.4263458251953,
|
| 113 |
+
"learning_rate": 1.4199999999999998e-05,
|
| 114 |
+
"loss": 6.5828,
|
| 115 |
+
"step": 150
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.024614437906234375,
|
| 119 |
+
"grad_norm": 174.7183837890625,
|
| 120 |
+
"learning_rate": 1.5199999999999998e-05,
|
| 121 |
+
"loss": 6.4821,
|
| 122 |
+
"step": 160
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.026152840275374026,
|
| 126 |
+
"grad_norm": 145.78704833984375,
|
| 127 |
+
"learning_rate": 1.6199999999999997e-05,
|
| 128 |
+
"loss": 6.4711,
|
| 129 |
+
"step": 170
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.027691242644513673,
|
| 133 |
+
"grad_norm": 217.81866455078125,
|
| 134 |
+
"learning_rate": 1.7199999999999998e-05,
|
| 135 |
+
"loss": 6.4947,
|
| 136 |
+
"step": 180
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.02922964501365332,
|
| 140 |
+
"grad_norm": 105.79635620117188,
|
| 141 |
+
"learning_rate": 1.82e-05,
|
| 142 |
+
"loss": 6.4997,
|
| 143 |
+
"step": 190
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.03076804738279297,
|
| 147 |
+
"grad_norm": 86.61837768554688,
|
| 148 |
+
"learning_rate": 1.92e-05,
|
| 149 |
+
"loss": 6.4604,
|
| 150 |
+
"step": 200
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.03230644975193262,
|
| 154 |
+
"grad_norm": 102.03221893310547,
|
| 155 |
+
"learning_rate": 2.0199999999999996e-05,
|
| 156 |
+
"loss": 6.4451,
|
| 157 |
+
"step": 210
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.033844852121072265,
|
| 161 |
+
"grad_norm": 93.15425109863281,
|
| 162 |
+
"learning_rate": 2.12e-05,
|
| 163 |
+
"loss": 6.4653,
|
| 164 |
+
"step": 220
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.035383254490211916,
|
| 168 |
+
"grad_norm": 120.66065979003906,
|
| 169 |
+
"learning_rate": 2.2199999999999998e-05,
|
| 170 |
+
"loss": 6.3898,
|
| 171 |
+
"step": 230
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.036921656859351566,
|
| 175 |
+
"grad_norm": 108.00463104248047,
|
| 176 |
+
"learning_rate": 2.3199999999999998e-05,
|
| 177 |
+
"loss": 6.3907,
|
| 178 |
+
"step": 240
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.03846005922849121,
|
| 182 |
+
"grad_norm": 129.6912841796875,
|
| 183 |
+
"learning_rate": 2.42e-05,
|
| 184 |
+
"loss": 6.3551,
|
| 185 |
+
"step": 250
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.03999846159763086,
|
| 189 |
+
"grad_norm": 66.96916961669922,
|
| 190 |
+
"learning_rate": 2.52e-05,
|
| 191 |
+
"loss": 6.3443,
|
| 192 |
+
"step": 260
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.04153686396677051,
|
| 196 |
+
"grad_norm": 44.31541061401367,
|
| 197 |
+
"learning_rate": 2.6199999999999996e-05,
|
| 198 |
+
"loss": 6.3372,
|
| 199 |
+
"step": 270
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.043075266335910155,
|
| 203 |
+
"grad_norm": 21.828279495239258,
|
| 204 |
+
"learning_rate": 2.72e-05,
|
| 205 |
+
"loss": 6.3368,
|
| 206 |
+
"step": 280
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.044613668705049805,
|
| 210 |
+
"grad_norm": 38.62431716918945,
|
| 211 |
+
"learning_rate": 2.8199999999999998e-05,
|
| 212 |
+
"loss": 6.2708,
|
| 213 |
+
"step": 290
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.046152071074189456,
|
| 217 |
+
"grad_norm": 28.7634334564209,
|
| 218 |
+
"learning_rate": 2.9199999999999995e-05,
|
| 219 |
+
"loss": 6.3324,
|
| 220 |
+
"step": 300
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.0476904734433291,
|
| 224 |
+
"grad_norm": 29.64866065979004,
|
| 225 |
+
"learning_rate": 3.02e-05,
|
| 226 |
+
"loss": 6.329,
|
| 227 |
+
"step": 310
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.04922887581246875,
|
| 231 |
+
"grad_norm": 27.20194435119629,
|
| 232 |
+
"learning_rate": 3.119999999999999e-05,
|
| 233 |
+
"loss": 6.2864,
|
| 234 |
+
"step": 320
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.0507672781816084,
|
| 238 |
+
"grad_norm": 24.59855079650879,
|
| 239 |
+
"learning_rate": 3.22e-05,
|
| 240 |
+
"loss": 6.2371,
|
| 241 |
+
"step": 330
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.05230568055074805,
|
| 245 |
+
"grad_norm": 32.6953239440918,
|
| 246 |
+
"learning_rate": 3.3199999999999994e-05,
|
| 247 |
+
"loss": 6.2824,
|
| 248 |
+
"step": 340
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.053844082919887695,
|
| 252 |
+
"grad_norm": 36.308067321777344,
|
| 253 |
+
"learning_rate": 3.42e-05,
|
| 254 |
+
"loss": 6.2977,
|
| 255 |
+
"step": 350
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.055382485289027346,
|
| 259 |
+
"grad_norm": 25.23465347290039,
|
| 260 |
+
"learning_rate": 3.5199999999999995e-05,
|
| 261 |
+
"loss": 6.3222,
|
| 262 |
+
"step": 360
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.056920887658166996,
|
| 266 |
+
"grad_norm": 47.594947814941406,
|
| 267 |
+
"learning_rate": 3.62e-05,
|
| 268 |
+
"loss": 6.2513,
|
| 269 |
+
"step": 370
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.05845929002730664,
|
| 273 |
+
"grad_norm": 44.8812370300293,
|
| 274 |
+
"learning_rate": 3.7199999999999996e-05,
|
| 275 |
+
"loss": 6.2244,
|
| 276 |
+
"step": 380
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.05999769239644629,
|
| 280 |
+
"grad_norm": 22.950286865234375,
|
| 281 |
+
"learning_rate": 3.8199999999999993e-05,
|
| 282 |
+
"loss": 6.1613,
|
| 283 |
+
"step": 390
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.06153609476558594,
|
| 287 |
+
"grad_norm": 23.65997314453125,
|
| 288 |
+
"learning_rate": 3.919999999999999e-05,
|
| 289 |
+
"loss": 6.1839,
|
| 290 |
+
"step": 400
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.06307449713472559,
|
| 294 |
+
"grad_norm": 37.60873031616211,
|
| 295 |
+
"learning_rate": 4.02e-05,
|
| 296 |
+
"loss": 6.1783,
|
| 297 |
+
"step": 410
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.06461289950386524,
|
| 301 |
+
"grad_norm": 30.03960418701172,
|
| 302 |
+
"learning_rate": 4.12e-05,
|
| 303 |
+
"loss": 6.1673,
|
| 304 |
+
"step": 420
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.06615130187300489,
|
| 308 |
+
"grad_norm": 38.63550567626953,
|
| 309 |
+
"learning_rate": 4.2199999999999996e-05,
|
| 310 |
+
"loss": 6.1151,
|
| 311 |
+
"step": 430
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.06768970424214453,
|
| 315 |
+
"grad_norm": 24.747684478759766,
|
| 316 |
+
"learning_rate": 4.319999999999999e-05,
|
| 317 |
+
"loss": 6.0666,
|
| 318 |
+
"step": 440
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.06922810661128419,
|
| 322 |
+
"grad_norm": 22.29100799560547,
|
| 323 |
+
"learning_rate": 4.42e-05,
|
| 324 |
+
"loss": 6.1266,
|
| 325 |
+
"step": 450
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.07076650898042383,
|
| 329 |
+
"grad_norm": 24.165313720703125,
|
| 330 |
+
"learning_rate": 4.5199999999999994e-05,
|
| 331 |
+
"loss": 6.1199,
|
| 332 |
+
"step": 460
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.07230491134956347,
|
| 336 |
+
"grad_norm": 19.26260757446289,
|
| 337 |
+
"learning_rate": 4.62e-05,
|
| 338 |
+
"loss": 6.1002,
|
| 339 |
+
"step": 470
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.07384331371870313,
|
| 343 |
+
"grad_norm": 21.792062759399414,
|
| 344 |
+
"learning_rate": 4.7199999999999995e-05,
|
| 345 |
+
"loss": 6.0919,
|
| 346 |
+
"step": 480
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.07538171608784278,
|
| 350 |
+
"grad_norm": 25.019775390625,
|
| 351 |
+
"learning_rate": 4.82e-05,
|
| 352 |
+
"loss": 6.1171,
|
| 353 |
+
"step": 490
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.07692011845698242,
|
| 357 |
+
"grad_norm": 21.131744384765625,
|
| 358 |
+
"learning_rate": 4.9199999999999997e-05,
|
| 359 |
+
"loss": 6.0676,
|
| 360 |
+
"step": 500
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.07845852082612208,
|
| 364 |
+
"grad_norm": 24.013105392456055,
|
| 365 |
+
"learning_rate": 5.0199999999999994e-05,
|
| 366 |
+
"loss": 6.0463,
|
| 367 |
+
"step": 510
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.07999692319526172,
|
| 371 |
+
"grad_norm": 20.643056869506836,
|
| 372 |
+
"learning_rate": 5.119999999999999e-05,
|
| 373 |
+
"loss": 6.0317,
|
| 374 |
+
"step": 520
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.08153532556440136,
|
| 378 |
+
"grad_norm": 27.55217170715332,
|
| 379 |
+
"learning_rate": 5.2199999999999995e-05,
|
| 380 |
+
"loss": 5.9955,
|
| 381 |
+
"step": 530
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.08307372793354102,
|
| 385 |
+
"grad_norm": 24.888734817504883,
|
| 386 |
+
"learning_rate": 5.32e-05,
|
| 387 |
+
"loss": 6.0481,
|
| 388 |
+
"step": 540
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.08461213030268067,
|
| 392 |
+
"grad_norm": 19.217220306396484,
|
| 393 |
+
"learning_rate": 5.4199999999999996e-05,
|
| 394 |
+
"loss": 5.9471,
|
| 395 |
+
"step": 550
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.08615053267182031,
|
| 399 |
+
"grad_norm": 19.650312423706055,
|
| 400 |
+
"learning_rate": 5.519999999999999e-05,
|
| 401 |
+
"loss": 5.9048,
|
| 402 |
+
"step": 560
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.08768893504095997,
|
| 406 |
+
"grad_norm": 21.437850952148438,
|
| 407 |
+
"learning_rate": 5.619999999999999e-05,
|
| 408 |
+
"loss": 6.0098,
|
| 409 |
+
"step": 570
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.08922733741009961,
|
| 413 |
+
"grad_norm": 23.66594696044922,
|
| 414 |
+
"learning_rate": 5.72e-05,
|
| 415 |
+
"loss": 5.9389,
|
| 416 |
+
"step": 580
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.09076573977923925,
|
| 420 |
+
"grad_norm": 33.00392532348633,
|
| 421 |
+
"learning_rate": 5.82e-05,
|
| 422 |
+
"loss": 5.8897,
|
| 423 |
+
"step": 590
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.09230414214837891,
|
| 427 |
+
"grad_norm": 20.737876892089844,
|
| 428 |
+
"learning_rate": 5.9199999999999996e-05,
|
| 429 |
+
"loss": 5.8598,
|
| 430 |
+
"step": 600
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.09384254451751856,
|
| 434 |
+
"grad_norm": 23.151851654052734,
|
| 435 |
+
"learning_rate": 6.019999999999999e-05,
|
| 436 |
+
"loss": 5.8652,
|
| 437 |
+
"step": 610
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.0953809468866582,
|
| 441 |
+
"grad_norm": 23.21453094482422,
|
| 442 |
+
"learning_rate": 6.12e-05,
|
| 443 |
+
"loss": 5.833,
|
| 444 |
+
"step": 620
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.09691934925579786,
|
| 448 |
+
"grad_norm": 23.856670379638672,
|
| 449 |
+
"learning_rate": 6.22e-05,
|
| 450 |
+
"loss": 5.8686,
|
| 451 |
+
"step": 630
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.0984577516249375,
|
| 455 |
+
"grad_norm": 18.7842960357666,
|
| 456 |
+
"learning_rate": 6.319999999999999e-05,
|
| 457 |
+
"loss": 5.8517,
|
| 458 |
+
"step": 640
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.09999615399407714,
|
| 462 |
+
"grad_norm": 16.733356475830078,
|
| 463 |
+
"learning_rate": 6.419999999999999e-05,
|
| 464 |
+
"loss": 5.8614,
|
| 465 |
+
"step": 650
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.1015345563632168,
|
| 469 |
+
"grad_norm": 18.41499137878418,
|
| 470 |
+
"learning_rate": 6.519999999999999e-05,
|
| 471 |
+
"loss": 5.8441,
|
| 472 |
+
"step": 660
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.10307295873235645,
|
| 476 |
+
"grad_norm": 20.63882827758789,
|
| 477 |
+
"learning_rate": 6.62e-05,
|
| 478 |
+
"loss": 5.7316,
|
| 479 |
+
"step": 670
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.1046113611014961,
|
| 483 |
+
"grad_norm": 17.48029899597168,
|
| 484 |
+
"learning_rate": 6.72e-05,
|
| 485 |
+
"loss": 5.6974,
|
| 486 |
+
"step": 680
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.10614976347063575,
|
| 490 |
+
"grad_norm": 22.546825408935547,
|
| 491 |
+
"learning_rate": 6.819999999999999e-05,
|
| 492 |
+
"loss": 5.7741,
|
| 493 |
+
"step": 690
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.10768816583977539,
|
| 497 |
+
"grad_norm": 16.246665954589844,
|
| 498 |
+
"learning_rate": 6.919999999999999e-05,
|
| 499 |
+
"loss": 5.6912,
|
| 500 |
+
"step": 700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.10922656820891505,
|
| 504 |
+
"grad_norm": 16.516803741455078,
|
| 505 |
+
"learning_rate": 7.02e-05,
|
| 506 |
+
"loss": 5.7397,
|
| 507 |
+
"step": 710
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.11076497057805469,
|
| 511 |
+
"grad_norm": 19.068937301635742,
|
| 512 |
+
"learning_rate": 7.12e-05,
|
| 513 |
+
"loss": 5.7386,
|
| 514 |
+
"step": 720
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 0.11230337294719434,
|
| 518 |
+
"grad_norm": 14.85578727722168,
|
| 519 |
+
"learning_rate": 7.219999999999999e-05,
|
| 520 |
+
"loss": 5.6994,
|
| 521 |
+
"step": 730
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 0.11384177531633399,
|
| 525 |
+
"grad_norm": 18.3763427734375,
|
| 526 |
+
"learning_rate": 7.319999999999999e-05,
|
| 527 |
+
"loss": 5.6505,
|
| 528 |
+
"step": 740
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 0.11538017768547364,
|
| 532 |
+
"grad_norm": 16.224000930786133,
|
| 533 |
+
"learning_rate": 7.419999999999999e-05,
|
| 534 |
+
"loss": 5.7087,
|
| 535 |
+
"step": 750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.11691858005461328,
|
| 539 |
+
"grad_norm": 15.365422248840332,
|
| 540 |
+
"learning_rate": 7.519999999999998e-05,
|
| 541 |
+
"loss": 5.6072,
|
| 542 |
+
"step": 760
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.11845698242375294,
|
| 546 |
+
"grad_norm": 18.633525848388672,
|
| 547 |
+
"learning_rate": 7.62e-05,
|
| 548 |
+
"loss": 5.5342,
|
| 549 |
+
"step": 770
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.11999538479289258,
|
| 553 |
+
"grad_norm": 19.027240753173828,
|
| 554 |
+
"learning_rate": 7.72e-05,
|
| 555 |
+
"loss": 5.5611,
|
| 556 |
+
"step": 780
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 0.12153378716203223,
|
| 560 |
+
"grad_norm": 16.857473373413086,
|
| 561 |
+
"learning_rate": 7.819999999999999e-05,
|
| 562 |
+
"loss": 5.5638,
|
| 563 |
+
"step": 790
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 0.12307218953117188,
|
| 567 |
+
"grad_norm": 15.174002647399902,
|
| 568 |
+
"learning_rate": 7.92e-05,
|
| 569 |
+
"loss": 5.5405,
|
| 570 |
+
"step": 800
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.12461059190031153,
|
| 574 |
+
"grad_norm": 16.1444034576416,
|
| 575 |
+
"learning_rate": 8.019999999999998e-05,
|
| 576 |
+
"loss": 5.5565,
|
| 577 |
+
"step": 810
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 0.12614899426945117,
|
| 581 |
+
"grad_norm": 18.24864959716797,
|
| 582 |
+
"learning_rate": 8.12e-05,
|
| 583 |
+
"loss": 5.4931,
|
| 584 |
+
"step": 820
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.12768739663859083,
|
| 588 |
+
"grad_norm": 18.213571548461914,
|
| 589 |
+
"learning_rate": 8.22e-05,
|
| 590 |
+
"loss": 5.5312,
|
| 591 |
+
"step": 830
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.12922579900773049,
|
| 595 |
+
"grad_norm": 17.174373626708984,
|
| 596 |
+
"learning_rate": 8.319999999999999e-05,
|
| 597 |
+
"loss": 5.4435,
|
| 598 |
+
"step": 840
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 0.13076420137687012,
|
| 602 |
+
"grad_norm": 15.482538223266602,
|
| 603 |
+
"learning_rate": 8.42e-05,
|
| 604 |
+
"loss": 5.4082,
|
| 605 |
+
"step": 850
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 0.13230260374600977,
|
| 609 |
+
"grad_norm": 17.8658504486084,
|
| 610 |
+
"learning_rate": 8.519999999999998e-05,
|
| 611 |
+
"loss": 5.473,
|
| 612 |
+
"step": 860
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.13384100611514943,
|
| 616 |
+
"grad_norm": 15.455946922302246,
|
| 617 |
+
"learning_rate": 8.62e-05,
|
| 618 |
+
"loss": 5.3541,
|
| 619 |
+
"step": 870
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 0.13537940848428906,
|
| 623 |
+
"grad_norm": 15.936956405639648,
|
| 624 |
+
"learning_rate": 8.719999999999999e-05,
|
| 625 |
+
"loss": 5.3488,
|
| 626 |
+
"step": 880
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.13691781085342872,
|
| 630 |
+
"grad_norm": 19.47191619873047,
|
| 631 |
+
"learning_rate": 8.819999999999999e-05,
|
| 632 |
+
"loss": 5.4054,
|
| 633 |
+
"step": 890
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.13845621322256838,
|
| 637 |
+
"grad_norm": 12.858948707580566,
|
| 638 |
+
"learning_rate": 8.92e-05,
|
| 639 |
+
"loss": 5.3355,
|
| 640 |
+
"step": 900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.139994615591708,
|
| 644 |
+
"grad_norm": 15.170941352844238,
|
| 645 |
+
"learning_rate": 9.02e-05,
|
| 646 |
+
"loss": 5.3799,
|
| 647 |
+
"step": 910
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.14153301796084766,
|
| 651 |
+
"grad_norm": 15.187207221984863,
|
| 652 |
+
"learning_rate": 9.12e-05,
|
| 653 |
+
"loss": 5.2994,
|
| 654 |
+
"step": 920
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.14307142032998732,
|
| 658 |
+
"grad_norm": 13.390646934509277,
|
| 659 |
+
"learning_rate": 9.219999999999999e-05,
|
| 660 |
+
"loss": 5.2335,
|
| 661 |
+
"step": 930
|
| 662 |
+
},
|
| 663 |
+
{
|
| 664 |
+
"epoch": 0.14460982269912695,
|
| 665 |
+
"grad_norm": 18.35064697265625,
|
| 666 |
+
"learning_rate": 9.319999999999999e-05,
|
| 667 |
+
"loss": 5.3427,
|
| 668 |
+
"step": 940
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.1461482250682666,
|
| 672 |
+
"grad_norm": 19.055871963500977,
|
| 673 |
+
"learning_rate": 9.419999999999999e-05,
|
| 674 |
+
"loss": 5.2078,
|
| 675 |
+
"step": 950
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.14768662743740626,
|
| 679 |
+
"grad_norm": 16.91037940979004,
|
| 680 |
+
"learning_rate": 9.52e-05,
|
| 681 |
+
"loss": 5.2242,
|
| 682 |
+
"step": 960
|
| 683 |
+
},
|
| 684 |
+
{
|
| 685 |
+
"epoch": 0.1492250298065459,
|
| 686 |
+
"grad_norm": 15.454617500305176,
|
| 687 |
+
"learning_rate": 9.62e-05,
|
| 688 |
+
"loss": 5.2252,
|
| 689 |
+
"step": 970
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.15076343217568555,
|
| 693 |
+
"grad_norm": 13.046433448791504,
|
| 694 |
+
"learning_rate": 9.719999999999999e-05,
|
| 695 |
+
"loss": 5.2289,
|
| 696 |
+
"step": 980
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 0.1523018345448252,
|
| 700 |
+
"grad_norm": 15.343282699584961,
|
| 701 |
+
"learning_rate": 9.819999999999999e-05,
|
| 702 |
+
"loss": 5.1872,
|
| 703 |
+
"step": 990
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.15384023691396484,
|
| 707 |
+
"grad_norm": 13.18584156036377,
|
| 708 |
+
"learning_rate": 9.919999999999999e-05,
|
| 709 |
+
"loss": 5.1641,
|
| 710 |
+
"step": 1000
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.1553786392831045,
|
| 714 |
+
"grad_norm": 16.466337203979492,
|
| 715 |
+
"learning_rate": 0.0001002,
|
| 716 |
+
"loss": 5.0973,
|
| 717 |
+
"step": 1010
|
| 718 |
+
},
|
| 719 |
+
{
|
| 720 |
+
"epoch": 0.15691704165224415,
|
| 721 |
+
"grad_norm": 16.390913009643555,
|
| 722 |
+
"learning_rate": 0.0001012,
|
| 723 |
+
"loss": 5.1066,
|
| 724 |
+
"step": 1020
|
| 725 |
+
},
|
| 726 |
+
{
|
| 727 |
+
"epoch": 0.15845544402138378,
|
| 728 |
+
"grad_norm": 13.89096736907959,
|
| 729 |
+
"learning_rate": 0.00010219999999999999,
|
| 730 |
+
"loss": 5.1375,
|
| 731 |
+
"step": 1030
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.15999384639052344,
|
| 735 |
+
"grad_norm": 14.639664649963379,
|
| 736 |
+
"learning_rate": 0.00010319999999999999,
|
| 737 |
+
"loss": 5.0562,
|
| 738 |
+
"step": 1040
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 0.1615322487596631,
|
| 742 |
+
"grad_norm": 14.390929222106934,
|
| 743 |
+
"learning_rate": 0.00010419999999999998,
|
| 744 |
+
"loss": 5.1309,
|
| 745 |
+
"step": 1050
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.16307065112880273,
|
| 749 |
+
"grad_norm": 15.142280578613281,
|
| 750 |
+
"learning_rate": 0.0001052,
|
| 751 |
+
"loss": 5.0287,
|
| 752 |
+
"step": 1060
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.1646090534979424,
|
| 756 |
+
"grad_norm": 13.173295021057129,
|
| 757 |
+
"learning_rate": 0.00010619999999999998,
|
| 758 |
+
"loss": 4.9568,
|
| 759 |
+
"step": 1070
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.16614745586708204,
|
| 763 |
+
"grad_norm": 15.059986114501953,
|
| 764 |
+
"learning_rate": 0.00010719999999999999,
|
| 765 |
+
"loss": 5.01,
|
| 766 |
+
"step": 1080
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 0.16768585823622167,
|
| 770 |
+
"grad_norm": 13.298657417297363,
|
| 771 |
+
"learning_rate": 0.0001082,
|
| 772 |
+
"loss": 4.9931,
|
| 773 |
+
"step": 1090
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.16922426060536133,
|
| 777 |
+
"grad_norm": 14.215367317199707,
|
| 778 |
+
"learning_rate": 0.00010919999999999998,
|
| 779 |
+
"loss": 4.9235,
|
| 780 |
+
"step": 1100
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 0.170762662974501,
|
| 784 |
+
"grad_norm": 14.457551002502441,
|
| 785 |
+
"learning_rate": 0.0001102,
|
| 786 |
+
"loss": 4.9374,
|
| 787 |
+
"step": 1110
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 0.17230106534364062,
|
| 791 |
+
"grad_norm": 14.105072975158691,
|
| 792 |
+
"learning_rate": 0.00011119999999999998,
|
| 793 |
+
"loss": 4.8857,
|
| 794 |
+
"step": 1120
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.17383946771278028,
|
| 798 |
+
"grad_norm": 16.460472106933594,
|
| 799 |
+
"learning_rate": 0.00011219999999999999,
|
| 800 |
+
"loss": 4.8934,
|
| 801 |
+
"step": 1130
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.17537787008191993,
|
| 805 |
+
"grad_norm": 15.866796493530273,
|
| 806 |
+
"learning_rate": 0.0001132,
|
| 807 |
+
"loss": 4.8335,
|
| 808 |
+
"step": 1140
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.17691627245105956,
|
| 812 |
+
"grad_norm": 14.246673583984375,
|
| 813 |
+
"learning_rate": 0.00011419999999999998,
|
| 814 |
+
"loss": 4.8547,
|
| 815 |
+
"step": 1150
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 0.17845467482019922,
|
| 819 |
+
"grad_norm": 14.18166732788086,
|
| 820 |
+
"learning_rate": 0.0001152,
|
| 821 |
+
"loss": 4.7489,
|
| 822 |
+
"step": 1160
|
| 823 |
+
},
|
| 824 |
+
{
|
| 825 |
+
"epoch": 0.17999307718933888,
|
| 826 |
+
"grad_norm": 14.299484252929688,
|
| 827 |
+
"learning_rate": 0.00011619999999999998,
|
| 828 |
+
"loss": 4.788,
|
| 829 |
+
"step": 1170
|
| 830 |
+
},
|
| 831 |
+
{
|
| 832 |
+
"epoch": 0.1815314795584785,
|
| 833 |
+
"grad_norm": 13.284591674804688,
|
| 834 |
+
"learning_rate": 0.00011719999999999999,
|
| 835 |
+
"loss": 4.7395,
|
| 836 |
+
"step": 1180
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.18306988192761817,
|
| 840 |
+
"grad_norm": 19.244670867919922,
|
| 841 |
+
"learning_rate": 0.0001182,
|
| 842 |
+
"loss": 4.7696,
|
| 843 |
+
"step": 1190
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 0.18460828429675782,
|
| 847 |
+
"grad_norm": 12.124654769897461,
|
| 848 |
+
"learning_rate": 0.00011919999999999998,
|
| 849 |
+
"loss": 4.7223,
|
| 850 |
+
"step": 1200
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.18614668666589745,
|
| 854 |
+
"grad_norm": 13.926753044128418,
|
| 855 |
+
"learning_rate": 0.0001202,
|
| 856 |
+
"loss": 4.7244,
|
| 857 |
+
"step": 1210
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.1876850890350371,
|
| 861 |
+
"grad_norm": 14.33655071258545,
|
| 862 |
+
"learning_rate": 0.00012119999999999999,
|
| 863 |
+
"loss": 4.6641,
|
| 864 |
+
"step": 1220
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.18922349140417677,
|
| 868 |
+
"grad_norm": 15.00214672088623,
|
| 869 |
+
"learning_rate": 0.0001222,
|
| 870 |
+
"loss": 4.6653,
|
| 871 |
+
"step": 1230
|
| 872 |
+
},
|
| 873 |
+
{
|
| 874 |
+
"epoch": 0.1907618937733164,
|
| 875 |
+
"grad_norm": 15.199170112609863,
|
| 876 |
+
"learning_rate": 0.00012319999999999999,
|
| 877 |
+
"loss": 4.6486,
|
| 878 |
+
"step": 1240
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.19230029614245606,
|
| 882 |
+
"grad_norm": 15.2691068649292,
|
| 883 |
+
"learning_rate": 0.00012419999999999998,
|
| 884 |
+
"loss": 4.545,
|
| 885 |
+
"step": 1250
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 0.19383869851159571,
|
| 889 |
+
"grad_norm": 13.688949584960938,
|
| 890 |
+
"learning_rate": 0.00012519999999999998,
|
| 891 |
+
"loss": 4.5674,
|
| 892 |
+
"step": 1260
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.19537710088073534,
|
| 896 |
+
"grad_norm": 14.25082015991211,
|
| 897 |
+
"learning_rate": 0.0001262,
|
| 898 |
+
"loss": 4.5894,
|
| 899 |
+
"step": 1270
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.196915503249875,
|
| 903 |
+
"grad_norm": 14.78211498260498,
|
| 904 |
+
"learning_rate": 0.00012719999999999997,
|
| 905 |
+
"loss": 4.5822,
|
| 906 |
+
"step": 1280
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 0.19845390561901466,
|
| 910 |
+
"grad_norm": 15.209827423095703,
|
| 911 |
+
"learning_rate": 0.0001282,
|
| 912 |
+
"loss": 4.5339,
|
| 913 |
+
"step": 1290
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 0.1999923079881543,
|
| 917 |
+
"grad_norm": 13.816431999206543,
|
| 918 |
+
"learning_rate": 0.00012919999999999997,
|
| 919 |
+
"loss": 4.5117,
|
| 920 |
+
"step": 1300
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.20153071035729395,
|
| 924 |
+
"grad_norm": 11.83894157409668,
|
| 925 |
+
"learning_rate": 0.0001302,
|
| 926 |
+
"loss": 4.4655,
|
| 927 |
+
"step": 1310
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.2030691127264336,
|
| 931 |
+
"grad_norm": 12.231562614440918,
|
| 932 |
+
"learning_rate": 0.0001312,
|
| 933 |
+
"loss": 4.4517,
|
| 934 |
+
"step": 1320
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 0.20460751509557323,
|
| 938 |
+
"grad_norm": 14.597062110900879,
|
| 939 |
+
"learning_rate": 0.0001322,
|
| 940 |
+
"loss": 4.4247,
|
| 941 |
+
"step": 1330
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 0.2061459174647129,
|
| 945 |
+
"grad_norm": 14.041385650634766,
|
| 946 |
+
"learning_rate": 0.00013319999999999999,
|
| 947 |
+
"loss": 4.4515,
|
| 948 |
+
"step": 1340
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 0.20768431983385255,
|
| 952 |
+
"grad_norm": 14.399479866027832,
|
| 953 |
+
"learning_rate": 0.0001342,
|
| 954 |
+
"loss": 4.4116,
|
| 955 |
+
"step": 1350
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 0.2092227222029922,
|
| 959 |
+
"grad_norm": 11.779489517211914,
|
| 960 |
+
"learning_rate": 0.00013519999999999998,
|
| 961 |
+
"loss": 4.3499,
|
| 962 |
+
"step": 1360
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.21076112457213184,
|
| 966 |
+
"grad_norm": 16.141752243041992,
|
| 967 |
+
"learning_rate": 0.0001362,
|
| 968 |
+
"loss": 4.263,
|
| 969 |
+
"step": 1370
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.2122995269412715,
|
| 973 |
+
"grad_norm": 12.214835166931152,
|
| 974 |
+
"learning_rate": 0.00013719999999999997,
|
| 975 |
+
"loss": 4.3635,
|
| 976 |
+
"step": 1380
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 0.21383792931041115,
|
| 980 |
+
"grad_norm": 11.679731369018555,
|
| 981 |
+
"learning_rate": 0.0001382,
|
| 982 |
+
"loss": 4.3867,
|
| 983 |
+
"step": 1390
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.21537633167955078,
|
| 987 |
+
"grad_norm": 13.870141983032227,
|
| 988 |
+
"learning_rate": 0.0001392,
|
| 989 |
+
"loss": 4.2572,
|
| 990 |
+
"step": 1400
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 0.21691473404869044,
|
| 994 |
+
"grad_norm": 15.715397834777832,
|
| 995 |
+
"learning_rate": 0.0001402,
|
| 996 |
+
"loss": 4.2449,
|
| 997 |
+
"step": 1410
|
| 998 |
+
},
|
| 999 |
+
{
|
| 1000 |
+
"epoch": 0.2184531364178301,
|
| 1001 |
+
"grad_norm": 13.933563232421875,
|
| 1002 |
+
"learning_rate": 0.0001412,
|
| 1003 |
+
"loss": 4.3044,
|
| 1004 |
+
"step": 1420
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.21999153878696973,
|
| 1008 |
+
"grad_norm": 11.775773048400879,
|
| 1009 |
+
"learning_rate": 0.0001422,
|
| 1010 |
+
"loss": 4.1688,
|
| 1011 |
+
"step": 1430
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.22152994115610938,
|
| 1015 |
+
"grad_norm": 14.641937255859375,
|
| 1016 |
+
"learning_rate": 0.00014319999999999998,
|
| 1017 |
+
"loss": 4.2013,
|
| 1018 |
+
"step": 1440
|
| 1019 |
+
},
|
| 1020 |
+
{
|
| 1021 |
+
"epoch": 0.22306834352524904,
|
| 1022 |
+
"grad_norm": 12.434427261352539,
|
| 1023 |
+
"learning_rate": 0.00014419999999999998,
|
| 1024 |
+
"loss": 4.138,
|
| 1025 |
+
"step": 1450
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 0.22460674589438867,
|
| 1029 |
+
"grad_norm": 13.412193298339844,
|
| 1030 |
+
"learning_rate": 0.00014519999999999998,
|
| 1031 |
+
"loss": 4.2305,
|
| 1032 |
+
"step": 1460
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 0.22614514826352833,
|
| 1036 |
+
"grad_norm": 12.081714630126953,
|
| 1037 |
+
"learning_rate": 0.0001462,
|
| 1038 |
+
"loss": 4.1844,
|
| 1039 |
+
"step": 1470
|
| 1040 |
+
},
|
| 1041 |
+
{
|
| 1042 |
+
"epoch": 0.22768355063266799,
|
| 1043 |
+
"grad_norm": 14.079381942749023,
|
| 1044 |
+
"learning_rate": 0.00014719999999999997,
|
| 1045 |
+
"loss": 4.1119,
|
| 1046 |
+
"step": 1480
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.22922195300180762,
|
| 1050 |
+
"grad_norm": 12.457756996154785,
|
| 1051 |
+
"learning_rate": 0.0001482,
|
| 1052 |
+
"loss": 4.1138,
|
| 1053 |
+
"step": 1490
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.23076035537094727,
|
| 1057 |
+
"grad_norm": 12.655537605285645,
|
| 1058 |
+
"learning_rate": 0.0001492,
|
| 1059 |
+
"loss": 4.0457,
|
| 1060 |
+
"step": 1500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 0.23229875774008693,
|
| 1064 |
+
"grad_norm": 14.801631927490234,
|
| 1065 |
+
"learning_rate": 0.0001502,
|
| 1066 |
+
"loss": 3.972,
|
| 1067 |
+
"step": 1510
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 0.23383716010922656,
|
| 1071 |
+
"grad_norm": 12.558164596557617,
|
| 1072 |
+
"learning_rate": 0.0001512,
|
| 1073 |
+
"loss": 4.0401,
|
| 1074 |
+
"step": 1520
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.23537556247836622,
|
| 1078 |
+
"grad_norm": 12.809494972229004,
|
| 1079 |
+
"learning_rate": 0.00015219999999999999,
|
| 1080 |
+
"loss": 3.9904,
|
| 1081 |
+
"step": 1530
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 0.23691396484750588,
|
| 1085 |
+
"grad_norm": 11.842789649963379,
|
| 1086 |
+
"learning_rate": 0.0001532,
|
| 1087 |
+
"loss": 4.0351,
|
| 1088 |
+
"step": 1540
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.2384523672166455,
|
| 1092 |
+
"grad_norm": 10.249466896057129,
|
| 1093 |
+
"learning_rate": 0.00015419999999999998,
|
| 1094 |
+
"loss": 3.9697,
|
| 1095 |
+
"step": 1550
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.23999076958578516,
|
| 1099 |
+
"grad_norm": 11.96493148803711,
|
| 1100 |
+
"learning_rate": 0.00015519999999999998,
|
| 1101 |
+
"loss": 3.891,
|
| 1102 |
+
"step": 1560
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.24152917195492482,
|
| 1106 |
+
"grad_norm": 13.920326232910156,
|
| 1107 |
+
"learning_rate": 0.0001562,
|
| 1108 |
+
"loss": 3.9165,
|
| 1109 |
+
"step": 1570
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 0.24306757432406445,
|
| 1113 |
+
"grad_norm": 10.648415565490723,
|
| 1114 |
+
"learning_rate": 0.0001572,
|
| 1115 |
+
"loss": 3.905,
|
| 1116 |
+
"step": 1580
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 0.2446059766932041,
|
| 1120 |
+
"grad_norm": 12.895554542541504,
|
| 1121 |
+
"learning_rate": 0.00015819999999999997,
|
| 1122 |
+
"loss": 3.8309,
|
| 1123 |
+
"step": 1590
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 0.24614437906234377,
|
| 1127 |
+
"grad_norm": 12.320569038391113,
|
| 1128 |
+
"learning_rate": 0.00015919999999999997,
|
| 1129 |
+
"loss": 3.8677,
|
| 1130 |
+
"step": 1600
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.2476827814314834,
|
| 1134 |
+
"grad_norm": 12.256221771240234,
|
| 1135 |
+
"learning_rate": 0.0001602,
|
| 1136 |
+
"loss": 3.809,
|
| 1137 |
+
"step": 1610
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 0.24922118380062305,
|
| 1141 |
+
"grad_norm": 12.378941535949707,
|
| 1142 |
+
"learning_rate": 0.0001612,
|
| 1143 |
+
"loss": 3.8226,
|
| 1144 |
+
"step": 1620
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 0.2507595861697627,
|
| 1148 |
+
"grad_norm": 11.537256240844727,
|
| 1149 |
+
"learning_rate": 0.00016219999999999999,
|
| 1150 |
+
"loss": 3.7577,
|
| 1151 |
+
"step": 1630
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 0.25229798853890234,
|
| 1155 |
+
"grad_norm": 11.635543823242188,
|
| 1156 |
+
"learning_rate": 0.0001632,
|
| 1157 |
+
"loss": 3.692,
|
| 1158 |
+
"step": 1640
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 0.253836390908042,
|
| 1162 |
+
"grad_norm": 11.149394035339355,
|
| 1163 |
+
"learning_rate": 0.00016419999999999998,
|
| 1164 |
+
"loss": 3.7158,
|
| 1165 |
+
"step": 1650
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 0.25537479327718166,
|
| 1169 |
+
"grad_norm": 12.258176803588867,
|
| 1170 |
+
"learning_rate": 0.00016519999999999998,
|
| 1171 |
+
"loss": 3.72,
|
| 1172 |
+
"step": 1660
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.2569131956463213,
|
| 1176 |
+
"grad_norm": 11.55327033996582,
|
| 1177 |
+
"learning_rate": 0.0001662,
|
| 1178 |
+
"loss": 3.6468,
|
| 1179 |
+
"step": 1670
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.25845159801546097,
|
| 1183 |
+
"grad_norm": 11.125475883483887,
|
| 1184 |
+
"learning_rate": 0.0001672,
|
| 1185 |
+
"loss": 3.6283,
|
| 1186 |
+
"step": 1680
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.2599900003846006,
|
| 1190 |
+
"grad_norm": 11.03936767578125,
|
| 1191 |
+
"learning_rate": 0.00016819999999999997,
|
| 1192 |
+
"loss": 3.6279,
|
| 1193 |
+
"step": 1690
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 0.26152840275374023,
|
| 1197 |
+
"grad_norm": 11.331442832946777,
|
| 1198 |
+
"learning_rate": 0.00016919999999999997,
|
| 1199 |
+
"loss": 3.5886,
|
| 1200 |
+
"step": 1700
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 0.2630668051228799,
|
| 1204 |
+
"grad_norm": 10.849939346313477,
|
| 1205 |
+
"learning_rate": 0.0001702,
|
| 1206 |
+
"loss": 3.6896,
|
| 1207 |
+
"step": 1710
|
| 1208 |
+
},
|
| 1209 |
+
{
|
| 1210 |
+
"epoch": 0.26460520749201955,
|
| 1211 |
+
"grad_norm": 9.935115814208984,
|
| 1212 |
+
"learning_rate": 0.0001712,
|
| 1213 |
+
"loss": 3.5194,
|
| 1214 |
+
"step": 1720
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.2661436098611592,
|
| 1218 |
+
"grad_norm": 11.35429859161377,
|
| 1219 |
+
"learning_rate": 0.00017219999999999998,
|
| 1220 |
+
"loss": 3.5311,
|
| 1221 |
+
"step": 1730
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.26768201223029886,
|
| 1225 |
+
"grad_norm": 11.712488174438477,
|
| 1226 |
+
"learning_rate": 0.00017319999999999998,
|
| 1227 |
+
"loss": 3.6083,
|
| 1228 |
+
"step": 1740
|
| 1229 |
+
},
|
| 1230 |
+
{
|
| 1231 |
+
"epoch": 0.26922041459943846,
|
| 1232 |
+
"grad_norm": 13.349652290344238,
|
| 1233 |
+
"learning_rate": 0.00017419999999999998,
|
| 1234 |
+
"loss": 3.6181,
|
| 1235 |
+
"step": 1750
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 0.2707588169685781,
|
| 1239 |
+
"grad_norm": 10.547552108764648,
|
| 1240 |
+
"learning_rate": 0.00017519999999999998,
|
| 1241 |
+
"loss": 3.6202,
|
| 1242 |
+
"step": 1760
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 0.2722972193377178,
|
| 1246 |
+
"grad_norm": 11.436111450195312,
|
| 1247 |
+
"learning_rate": 0.0001762,
|
| 1248 |
+
"loss": 3.5857,
|
| 1249 |
+
"step": 1770
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.27383562170685743,
|
| 1253 |
+
"grad_norm": 10.192728996276855,
|
| 1254 |
+
"learning_rate": 0.0001772,
|
| 1255 |
+
"loss": 3.3816,
|
| 1256 |
+
"step": 1780
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.2753740240759971,
|
| 1260 |
+
"grad_norm": 9.88837718963623,
|
| 1261 |
+
"learning_rate": 0.00017819999999999997,
|
| 1262 |
+
"loss": 3.4138,
|
| 1263 |
+
"step": 1790
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 0.27691242644513675,
|
| 1267 |
+
"grad_norm": 9.523514747619629,
|
| 1268 |
+
"learning_rate": 0.0001792,
|
| 1269 |
+
"loss": 3.4505,
|
| 1270 |
+
"step": 1800
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.27845082881427635,
|
| 1274 |
+
"grad_norm": 9.86983585357666,
|
| 1275 |
+
"learning_rate": 0.0001802,
|
| 1276 |
+
"loss": 3.4563,
|
| 1277 |
+
"step": 1810
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 0.279989231183416,
|
| 1281 |
+
"grad_norm": 10.267000198364258,
|
| 1282 |
+
"learning_rate": 0.00018119999999999999,
|
| 1283 |
+
"loss": 3.2901,
|
| 1284 |
+
"step": 1820
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 0.28152763355255567,
|
| 1288 |
+
"grad_norm": 9.708111763000488,
|
| 1289 |
+
"learning_rate": 0.00018219999999999996,
|
| 1290 |
+
"loss": 3.4291,
|
| 1291 |
+
"step": 1830
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.2830660359216953,
|
| 1295 |
+
"grad_norm": 16.580785751342773,
|
| 1296 |
+
"learning_rate": 0.00018319999999999998,
|
| 1297 |
+
"loss": 3.3824,
|
| 1298 |
+
"step": 1840
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.284604438290835,
|
| 1302 |
+
"grad_norm": 10.267104148864746,
|
| 1303 |
+
"learning_rate": 0.00018419999999999998,
|
| 1304 |
+
"loss": 3.3334,
|
| 1305 |
+
"step": 1850
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.28614284065997464,
|
| 1309 |
+
"grad_norm": 9.529206275939941,
|
| 1310 |
+
"learning_rate": 0.00018519999999999998,
|
| 1311 |
+
"loss": 3.3357,
|
| 1312 |
+
"step": 1860
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 0.28768124302911424,
|
| 1316 |
+
"grad_norm": 9.670571327209473,
|
| 1317 |
+
"learning_rate": 0.0001862,
|
| 1318 |
+
"loss": 3.2536,
|
| 1319 |
+
"step": 1870
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 0.2892196453982539,
|
| 1323 |
+
"grad_norm": 9.460128784179688,
|
| 1324 |
+
"learning_rate": 0.0001872,
|
| 1325 |
+
"loss": 3.3416,
|
| 1326 |
+
"step": 1880
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 0.29075804776739356,
|
| 1330 |
+
"grad_norm": 10.193305015563965,
|
| 1331 |
+
"learning_rate": 0.00018819999999999997,
|
| 1332 |
+
"loss": 3.2714,
|
| 1333 |
+
"step": 1890
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 0.2922964501365332,
|
| 1337 |
+
"grad_norm": 9.581829071044922,
|
| 1338 |
+
"learning_rate": 0.0001892,
|
| 1339 |
+
"loss": 3.2382,
|
| 1340 |
+
"step": 1900
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.29383485250567287,
|
| 1344 |
+
"grad_norm": 10.401164054870605,
|
| 1345 |
+
"learning_rate": 0.0001902,
|
| 1346 |
+
"loss": 3.2288,
|
| 1347 |
+
"step": 1910
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 0.29537325487481253,
|
| 1351 |
+
"grad_norm": 10.252750396728516,
|
| 1352 |
+
"learning_rate": 0.00019119999999999999,
|
| 1353 |
+
"loss": 3.2718,
|
| 1354 |
+
"step": 1920
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 0.29691165724395213,
|
| 1358 |
+
"grad_norm": 9.973576545715332,
|
| 1359 |
+
"learning_rate": 0.0001922,
|
| 1360 |
+
"loss": 3.2742,
|
| 1361 |
+
"step": 1930
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 0.2984500596130918,
|
| 1365 |
+
"grad_norm": 9.254643440246582,
|
| 1366 |
+
"learning_rate": 0.00019319999999999998,
|
| 1367 |
+
"loss": 3.2849,
|
| 1368 |
+
"step": 1940
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.29998846198223145,
|
| 1372 |
+
"grad_norm": 9.65290641784668,
|
| 1373 |
+
"learning_rate": 0.00019419999999999998,
|
| 1374 |
+
"loss": 3.2829,
|
| 1375 |
+
"step": 1950
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.3015268643513711,
|
| 1379 |
+
"grad_norm": 10.02481746673584,
|
| 1380 |
+
"learning_rate": 0.00019519999999999997,
|
| 1381 |
+
"loss": 3.168,
|
| 1382 |
+
"step": 1960
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.30306526672051076,
|
| 1386 |
+
"grad_norm": 9.108298301696777,
|
| 1387 |
+
"learning_rate": 0.0001962,
|
| 1388 |
+
"loss": 3.1571,
|
| 1389 |
+
"step": 1970
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 0.3046036690896504,
|
| 1393 |
+
"grad_norm": 9.827427864074707,
|
| 1394 |
+
"learning_rate": 0.00019719999999999997,
|
| 1395 |
+
"loss": 3.0885,
|
| 1396 |
+
"step": 1980
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 0.30614207145879,
|
| 1400 |
+
"grad_norm": 10.132247924804688,
|
| 1401 |
+
"learning_rate": 0.00019819999999999997,
|
| 1402 |
+
"loss": 3.116,
|
| 1403 |
+
"step": 1990
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 0.3076804738279297,
|
| 1407 |
+
"grad_norm": 8.130690574645996,
|
| 1408 |
+
"learning_rate": 0.0001992,
|
| 1409 |
+
"loss": 3.11,
|
| 1410 |
+
"step": 2000
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 0.30921887619706934,
|
| 1414 |
+
"grad_norm": 8.482885360717773,
|
| 1415 |
+
"learning_rate": 0.0002002,
|
| 1416 |
+
"loss": 3.1642,
|
| 1417 |
+
"step": 2010
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 0.310757278566209,
|
| 1421 |
+
"grad_norm": 9.590483665466309,
|
| 1422 |
+
"learning_rate": 0.00020119999999999998,
|
| 1423 |
+
"loss": 3.1351,
|
| 1424 |
+
"step": 2020
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.31229568093534865,
|
| 1428 |
+
"grad_norm": 9.35024642944336,
|
| 1429 |
+
"learning_rate": 0.0002022,
|
| 1430 |
+
"loss": 3.1109,
|
| 1431 |
+
"step": 2030
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.3138340833044883,
|
| 1435 |
+
"grad_norm": 8.985564231872559,
|
| 1436 |
+
"learning_rate": 0.00020319999999999998,
|
| 1437 |
+
"loss": 3.1119,
|
| 1438 |
+
"step": 2040
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 0.3153724856736279,
|
| 1442 |
+
"grad_norm": 9.026268005371094,
|
| 1443 |
+
"learning_rate": 0.00020419999999999998,
|
| 1444 |
+
"loss": 3.02,
|
| 1445 |
+
"step": 2050
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.31691088804276757,
|
| 1449 |
+
"grad_norm": 10.748119354248047,
|
| 1450 |
+
"learning_rate": 0.0002052,
|
| 1451 |
+
"loss": 2.9356,
|
| 1452 |
+
"step": 2060
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.3184492904119072,
|
| 1456 |
+
"grad_norm": 9.285269737243652,
|
| 1457 |
+
"learning_rate": 0.0002062,
|
| 1458 |
+
"loss": 3.1128,
|
| 1459 |
+
"step": 2070
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.3199876927810469,
|
| 1463 |
+
"grad_norm": 8.402143478393555,
|
| 1464 |
+
"learning_rate": 0.00020719999999999997,
|
| 1465 |
+
"loss": 2.9105,
|
| 1466 |
+
"step": 2080
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.32152609515018654,
|
| 1470 |
+
"grad_norm": 8.802371978759766,
|
| 1471 |
+
"learning_rate": 0.00020819999999999996,
|
| 1472 |
+
"loss": 2.9336,
|
| 1473 |
+
"step": 2090
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.3230644975193262,
|
| 1477 |
+
"grad_norm": 12.419921875,
|
| 1478 |
+
"learning_rate": 0.0002092,
|
| 1479 |
+
"loss": 3.0309,
|
| 1480 |
+
"step": 2100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.3246028998884658,
|
| 1484 |
+
"grad_norm": 8.162283897399902,
|
| 1485 |
+
"learning_rate": 0.0002102,
|
| 1486 |
+
"loss": 3.0597,
|
| 1487 |
+
"step": 2110
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.32614130225760546,
|
| 1491 |
+
"grad_norm": 9.805724143981934,
|
| 1492 |
+
"learning_rate": 0.00021119999999999996,
|
| 1493 |
+
"loss": 2.8969,
|
| 1494 |
+
"step": 2120
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.3276797046267451,
|
| 1498 |
+
"grad_norm": 8.767786979675293,
|
| 1499 |
+
"learning_rate": 0.00021219999999999998,
|
| 1500 |
+
"loss": 2.9634,
|
| 1501 |
+
"step": 2130
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.3292181069958848,
|
| 1505 |
+
"grad_norm": 9.984760284423828,
|
| 1506 |
+
"learning_rate": 0.00021319999999999998,
|
| 1507 |
+
"loss": 2.7813,
|
| 1508 |
+
"step": 2140
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.33075650936502443,
|
| 1512 |
+
"grad_norm": 9.307220458984375,
|
| 1513 |
+
"learning_rate": 0.00021419999999999998,
|
| 1514 |
+
"loss": 2.9534,
|
| 1515 |
+
"step": 2150
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.3322949117341641,
|
| 1519 |
+
"grad_norm": 8.813392639160156,
|
| 1520 |
+
"learning_rate": 0.0002152,
|
| 1521 |
+
"loss": 2.8916,
|
| 1522 |
+
"step": 2160
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.33383331410330375,
|
| 1526 |
+
"grad_norm": 9.606163024902344,
|
| 1527 |
+
"learning_rate": 0.0002162,
|
| 1528 |
+
"loss": 2.8932,
|
| 1529 |
+
"step": 2170
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.33537171647244335,
|
| 1533 |
+
"grad_norm": 8.489242553710938,
|
| 1534 |
+
"learning_rate": 0.00021719999999999997,
|
| 1535 |
+
"loss": 2.7864,
|
| 1536 |
+
"step": 2180
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.336910118841583,
|
| 1540 |
+
"grad_norm": 8.720660209655762,
|
| 1541 |
+
"learning_rate": 0.0002182,
|
| 1542 |
+
"loss": 2.8711,
|
| 1543 |
+
"step": 2190
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.33844852121072266,
|
| 1547 |
+
"grad_norm": 10.3447265625,
|
| 1548 |
+
"learning_rate": 0.0002192,
|
| 1549 |
+
"loss": 2.9784,
|
| 1550 |
+
"step": 2200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.3399869235798623,
|
| 1554 |
+
"grad_norm": 9.338818550109863,
|
| 1555 |
+
"learning_rate": 0.00022019999999999999,
|
| 1556 |
+
"loss": 2.8191,
|
| 1557 |
+
"step": 2210
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.341525325949002,
|
| 1561 |
+
"grad_norm": 9.70388126373291,
|
| 1562 |
+
"learning_rate": 0.00022119999999999996,
|
| 1563 |
+
"loss": 2.7959,
|
| 1564 |
+
"step": 2220
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.34306372831814164,
|
| 1568 |
+
"grad_norm": 8.778887748718262,
|
| 1569 |
+
"learning_rate": 0.00022219999999999998,
|
| 1570 |
+
"loss": 2.7949,
|
| 1571 |
+
"step": 2230
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.34460213068728124,
|
| 1575 |
+
"grad_norm": 28.05166244506836,
|
| 1576 |
+
"learning_rate": 0.00022319999999999998,
|
| 1577 |
+
"loss": 2.809,
|
| 1578 |
+
"step": 2240
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.3461405330564209,
|
| 1582 |
+
"grad_norm": 8.896856307983398,
|
| 1583 |
+
"learning_rate": 0.00022419999999999997,
|
| 1584 |
+
"loss": 2.7963,
|
| 1585 |
+
"step": 2250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.34767893542556055,
|
| 1589 |
+
"grad_norm": 9.01641845703125,
|
| 1590 |
+
"learning_rate": 0.0002252,
|
| 1591 |
+
"loss": 2.8156,
|
| 1592 |
+
"step": 2260
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.3492173377947002,
|
| 1596 |
+
"grad_norm": 9.186058044433594,
|
| 1597 |
+
"learning_rate": 0.00022619999999999997,
|
| 1598 |
+
"loss": 2.8044,
|
| 1599 |
+
"step": 2270
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.35075574016383987,
|
| 1603 |
+
"grad_norm": 8.313657760620117,
|
| 1604 |
+
"learning_rate": 0.00022719999999999997,
|
| 1605 |
+
"loss": 2.7988,
|
| 1606 |
+
"step": 2280
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.3522941425329795,
|
| 1610 |
+
"grad_norm": 9.183436393737793,
|
| 1611 |
+
"learning_rate": 0.0002282,
|
| 1612 |
+
"loss": 2.7274,
|
| 1613 |
+
"step": 2290
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.35383254490211913,
|
| 1617 |
+
"grad_norm": 8.578018188476562,
|
| 1618 |
+
"learning_rate": 0.0002292,
|
| 1619 |
+
"loss": 2.6665,
|
| 1620 |
+
"step": 2300
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.3553709472712588,
|
| 1624 |
+
"grad_norm": 9.016053199768066,
|
| 1625 |
+
"learning_rate": 0.00023019999999999998,
|
| 1626 |
+
"loss": 2.7776,
|
| 1627 |
+
"step": 2310
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.35690934964039844,
|
| 1631 |
+
"grad_norm": 8.603869438171387,
|
| 1632 |
+
"learning_rate": 0.0002312,
|
| 1633 |
+
"loss": 2.7173,
|
| 1634 |
+
"step": 2320
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.3584477520095381,
|
| 1638 |
+
"grad_norm": 9.482624053955078,
|
| 1639 |
+
"learning_rate": 0.00023219999999999998,
|
| 1640 |
+
"loss": 2.7423,
|
| 1641 |
+
"step": 2330
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.35998615437867776,
|
| 1645 |
+
"grad_norm": 10.839862823486328,
|
| 1646 |
+
"learning_rate": 0.00023319999999999998,
|
| 1647 |
+
"loss": 2.7134,
|
| 1648 |
+
"step": 2340
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.3615245567478174,
|
| 1652 |
+
"grad_norm": 9.699636459350586,
|
| 1653 |
+
"learning_rate": 0.00023419999999999997,
|
| 1654 |
+
"loss": 2.7307,
|
| 1655 |
+
"step": 2350
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.363062959116957,
|
| 1659 |
+
"grad_norm": 8.4449462890625,
|
| 1660 |
+
"learning_rate": 0.0002352,
|
| 1661 |
+
"loss": 2.6609,
|
| 1662 |
+
"step": 2360
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.3646013614860967,
|
| 1666 |
+
"grad_norm": 8.534226417541504,
|
| 1667 |
+
"learning_rate": 0.00023619999999999997,
|
| 1668 |
+
"loss": 2.6645,
|
| 1669 |
+
"step": 2370
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.36613976385523633,
|
| 1673 |
+
"grad_norm": 10.571688652038574,
|
| 1674 |
+
"learning_rate": 0.00023719999999999997,
|
| 1675 |
+
"loss": 2.6355,
|
| 1676 |
+
"step": 2380
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.367678166224376,
|
| 1680 |
+
"grad_norm": 9.755474090576172,
|
| 1681 |
+
"learning_rate": 0.0002382,
|
| 1682 |
+
"loss": 2.6799,
|
| 1683 |
+
"step": 2390
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.36921656859351565,
|
| 1687 |
+
"grad_norm": 9.571345329284668,
|
| 1688 |
+
"learning_rate": 0.0002392,
|
| 1689 |
+
"loss": 2.6043,
|
| 1690 |
+
"step": 2400
|
| 1691 |
+
}
|
| 1692 |
+
],
|
| 1693 |
+
"logging_steps": 10,
|
| 1694 |
+
"max_steps": 100000,
|
| 1695 |
+
"num_input_tokens_seen": 0,
|
| 1696 |
+
"num_train_epochs": 16,
|
| 1697 |
+
"save_steps": 100,
|
| 1698 |
+
"stateful_callbacks": {
|
| 1699 |
+
"TrainerControl": {
|
| 1700 |
+
"args": {
|
| 1701 |
+
"should_epoch_stop": false,
|
| 1702 |
+
"should_evaluate": false,
|
| 1703 |
+
"should_log": false,
|
| 1704 |
+
"should_save": true,
|
| 1705 |
+
"should_training_stop": false
|
| 1706 |
+
},
|
| 1707 |
+
"attributes": {}
|
| 1708 |
+
}
|
| 1709 |
+
},
|
| 1710 |
+
"total_flos": 4.5944195352576e+16,
|
| 1711 |
+
"train_batch_size": 8,
|
| 1712 |
+
"trial_name": null,
|
| 1713 |
+
"trial_params": null
|
| 1714 |
+
}
|
checkpoint-2400/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62417cc687938b852957f7012ea833e7979cfe3f532b1c8c4b1c41687231ccf6
|
| 3 |
+
size 5201
|