Training in progress, step 2500, checkpoint
Browse files- checkpoint-2500/config.json +40 -0
- checkpoint-2500/generation_config.json +12 -0
- checkpoint-2500/model.safetensors +3 -0
- checkpoint-2500/optimizer.pt +3 -0
- checkpoint-2500/rng_state.pth +3 -0
- checkpoint-2500/scheduler.pt +3 -0
- checkpoint-2500/trainer_state.json +1783 -0
- checkpoint-2500/training_args.bin +3 -0
checkpoint-2500/config.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "meta-llama/Llama-3.2-1B-Instruct",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlamaForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"bos_token_id": 128000,
|
| 9 |
+
"eos_token_id": [
|
| 10 |
+
128001,
|
| 11 |
+
128008,
|
| 12 |
+
128009
|
| 13 |
+
],
|
| 14 |
+
"head_dim": 64,
|
| 15 |
+
"hidden_act": "silu",
|
| 16 |
+
"hidden_size": 2048,
|
| 17 |
+
"initializer_range": 0.02,
|
| 18 |
+
"intermediate_size": 8192,
|
| 19 |
+
"max_position_embeddings": 131072,
|
| 20 |
+
"mlp_bias": false,
|
| 21 |
+
"model_type": "llama",
|
| 22 |
+
"num_attention_heads": 32,
|
| 23 |
+
"num_hidden_layers": 16,
|
| 24 |
+
"num_key_value_heads": 8,
|
| 25 |
+
"pretraining_tp": 1,
|
| 26 |
+
"rms_norm_eps": 1e-05,
|
| 27 |
+
"rope_scaling": {
|
| 28 |
+
"factor": 32.0,
|
| 29 |
+
"high_freq_factor": 4.0,
|
| 30 |
+
"low_freq_factor": 1.0,
|
| 31 |
+
"original_max_position_embeddings": 8192,
|
| 32 |
+
"rope_type": "llama3"
|
| 33 |
+
},
|
| 34 |
+
"rope_theta": 500000.0,
|
| 35 |
+
"tie_word_embeddings": true,
|
| 36 |
+
"torch_dtype": "bfloat16",
|
| 37 |
+
"transformers_version": "4.46.3",
|
| 38 |
+
"use_cache": true,
|
| 39 |
+
"vocab_size": 128256
|
| 40 |
+
}
|
checkpoint-2500/generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128008,
|
| 7 |
+
128009
|
| 8 |
+
],
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.46.3"
|
| 12 |
+
}
|
checkpoint-2500/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:25510bc11cc73089926dc8e84f7d6133a7d84382888b93073ea1ef6d91d41d35
|
| 3 |
+
size 2471645608
|
checkpoint-2500/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69d49ef49b4c86f98f1b0b8401714613b5791ae7d2d5595bf2ac355c6dd4dac6
|
| 3 |
+
size 3724602
|
checkpoint-2500/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f2fba0be0d5ff6435647ef811e02938bae4a724479797c0e92fc5c9c0c20a0f
|
| 3 |
+
size 14244
|
checkpoint-2500/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3d944f5855dedd93de5810fb83351c896ba8bcf8ae6abf8ac51fb2013da8b6d
|
| 3 |
+
size 1064
|
checkpoint-2500/trainer_state.json
ADDED
|
@@ -0,0 +1,1783 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.23769059814839025,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 2500,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0009507623925935609,
|
| 13 |
+
"grad_norm": 37.25,
|
| 14 |
+
"learning_rate": 8.000000000000001e-07,
|
| 15 |
+
"loss": 0.1712,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0019015247851871218,
|
| 20 |
+
"grad_norm": 161.0,
|
| 21 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 22 |
+
"loss": 0.1789,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.0028522871777806828,
|
| 27 |
+
"grad_norm": 382.0,
|
| 28 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 29 |
+
"loss": 0.2312,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.0038030495703742437,
|
| 34 |
+
"grad_norm": 316.0,
|
| 35 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 36 |
+
"loss": 0.2718,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.004753811962967805,
|
| 41 |
+
"grad_norm": 314.0,
|
| 42 |
+
"learning_rate": 4.000000000000001e-06,
|
| 43 |
+
"loss": 0.3624,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.0057045743555613655,
|
| 48 |
+
"grad_norm": 310.0,
|
| 49 |
+
"learning_rate": 4.800000000000001e-06,
|
| 50 |
+
"loss": 0.3867,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.0066553367481549264,
|
| 55 |
+
"grad_norm": 84.0,
|
| 56 |
+
"learning_rate": 5.600000000000001e-06,
|
| 57 |
+
"loss": 0.6101,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.007606099140748487,
|
| 62 |
+
"grad_norm": 73.0,
|
| 63 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 64 |
+
"loss": 0.5141,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.008556861533342048,
|
| 69 |
+
"grad_norm": 54.75,
|
| 70 |
+
"learning_rate": 7.2000000000000005e-06,
|
| 71 |
+
"loss": 0.5038,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.00950762392593561,
|
| 76 |
+
"grad_norm": 81.5,
|
| 77 |
+
"learning_rate": 8.000000000000001e-06,
|
| 78 |
+
"loss": 0.6746,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.01045838631852917,
|
| 83 |
+
"grad_norm": 31.0,
|
| 84 |
+
"learning_rate": 8.8e-06,
|
| 85 |
+
"loss": 0.4569,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.011409148711122731,
|
| 90 |
+
"grad_norm": 47.0,
|
| 91 |
+
"learning_rate": 9.600000000000001e-06,
|
| 92 |
+
"loss": 0.3131,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.012359911103716292,
|
| 97 |
+
"grad_norm": 250.0,
|
| 98 |
+
"learning_rate": 1.04e-05,
|
| 99 |
+
"loss": 0.5069,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.013310673496309853,
|
| 104 |
+
"grad_norm": 28.375,
|
| 105 |
+
"learning_rate": 1.1200000000000001e-05,
|
| 106 |
+
"loss": 0.3334,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.014261435888903414,
|
| 111 |
+
"grad_norm": 29.875,
|
| 112 |
+
"learning_rate": 1.2e-05,
|
| 113 |
+
"loss": 0.2985,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.015212198281496975,
|
| 118 |
+
"grad_norm": 22.125,
|
| 119 |
+
"learning_rate": 1.2800000000000001e-05,
|
| 120 |
+
"loss": 0.283,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.016162960674090537,
|
| 125 |
+
"grad_norm": 24.5,
|
| 126 |
+
"learning_rate": 1.3600000000000002e-05,
|
| 127 |
+
"loss": 0.3052,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.017113723066684097,
|
| 132 |
+
"grad_norm": 7.5,
|
| 133 |
+
"learning_rate": 1.4400000000000001e-05,
|
| 134 |
+
"loss": 0.276,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.01806448545927766,
|
| 139 |
+
"grad_norm": 33.25,
|
| 140 |
+
"learning_rate": 1.5200000000000002e-05,
|
| 141 |
+
"loss": 0.302,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.01901524785187122,
|
| 146 |
+
"grad_norm": 115.0,
|
| 147 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 148 |
+
"loss": 0.651,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.01996601024446478,
|
| 153 |
+
"grad_norm": 17.125,
|
| 154 |
+
"learning_rate": 1.6800000000000002e-05,
|
| 155 |
+
"loss": 0.3109,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.02091677263705834,
|
| 160 |
+
"grad_norm": 5.625,
|
| 161 |
+
"learning_rate": 1.76e-05,
|
| 162 |
+
"loss": 0.2692,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.021867535029651903,
|
| 167 |
+
"grad_norm": 25.375,
|
| 168 |
+
"learning_rate": 1.8400000000000003e-05,
|
| 169 |
+
"loss": 0.2983,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.022818297422245462,
|
| 174 |
+
"grad_norm": 15.6875,
|
| 175 |
+
"learning_rate": 1.9200000000000003e-05,
|
| 176 |
+
"loss": 0.2063,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.023769059814839025,
|
| 181 |
+
"grad_norm": 18.375,
|
| 182 |
+
"learning_rate": 2e-05,
|
| 183 |
+
"loss": 0.2113,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.024719822207432584,
|
| 188 |
+
"grad_norm": 16.625,
|
| 189 |
+
"learning_rate": 1.9999025240093045e-05,
|
| 190 |
+
"loss": 0.2393,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.025670584600026147,
|
| 195 |
+
"grad_norm": 3.671875,
|
| 196 |
+
"learning_rate": 1.9996101150403543e-05,
|
| 197 |
+
"loss": 0.2577,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.026621346992619706,
|
| 202 |
+
"grad_norm": 25.5,
|
| 203 |
+
"learning_rate": 1.9991228300988586e-05,
|
| 204 |
+
"loss": 0.26,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.02757210938521327,
|
| 209 |
+
"grad_norm": 25.875,
|
| 210 |
+
"learning_rate": 1.9984407641819812e-05,
|
| 211 |
+
"loss": 0.2123,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.028522871777806828,
|
| 216 |
+
"grad_norm": 18.625,
|
| 217 |
+
"learning_rate": 1.9975640502598243e-05,
|
| 218 |
+
"loss": 0.1998,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.02947363417040039,
|
| 223 |
+
"grad_norm": 11.625,
|
| 224 |
+
"learning_rate": 1.9964928592495046e-05,
|
| 225 |
+
"loss": 0.1834,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.03042439656299395,
|
| 230 |
+
"grad_norm": 14.375,
|
| 231 |
+
"learning_rate": 1.9952273999818312e-05,
|
| 232 |
+
"loss": 0.2481,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.03137515895558751,
|
| 237 |
+
"grad_norm": 17.875,
|
| 238 |
+
"learning_rate": 1.9937679191605964e-05,
|
| 239 |
+
"loss": 0.2155,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.032325921348181075,
|
| 244 |
+
"grad_norm": 18.5,
|
| 245 |
+
"learning_rate": 1.9921147013144782e-05,
|
| 246 |
+
"loss": 0.1976,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.03327668374077463,
|
| 251 |
+
"grad_norm": 14.0625,
|
| 252 |
+
"learning_rate": 1.9902680687415704e-05,
|
| 253 |
+
"loss": 0.1961,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.03422744613336819,
|
| 258 |
+
"grad_norm": 8.4375,
|
| 259 |
+
"learning_rate": 1.988228381446553e-05,
|
| 260 |
+
"loss": 0.1953,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.035178208525961756,
|
| 265 |
+
"grad_norm": 11.5,
|
| 266 |
+
"learning_rate": 1.985996037070505e-05,
|
| 267 |
+
"loss": 0.2067,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.03612897091855532,
|
| 272 |
+
"grad_norm": 12.3125,
|
| 273 |
+
"learning_rate": 1.983571470813386e-05,
|
| 274 |
+
"loss": 0.2216,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.037079733311148874,
|
| 279 |
+
"grad_norm": 13.0625,
|
| 280 |
+
"learning_rate": 1.9809551553491918e-05,
|
| 281 |
+
"loss": 0.1795,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.03803049570374244,
|
| 286 |
+
"grad_norm": 12.75,
|
| 287 |
+
"learning_rate": 1.9781476007338058e-05,
|
| 288 |
+
"loss": 0.1708,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.038981258096336,
|
| 293 |
+
"grad_norm": 14.875,
|
| 294 |
+
"learning_rate": 1.9751493543055634e-05,
|
| 295 |
+
"loss": 0.1729,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.03993202048892956,
|
| 300 |
+
"grad_norm": 12.375,
|
| 301 |
+
"learning_rate": 1.9719610005785466e-05,
|
| 302 |
+
"loss": 0.1695,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.040882782881523125,
|
| 307 |
+
"grad_norm": 5.78125,
|
| 308 |
+
"learning_rate": 1.9685831611286312e-05,
|
| 309 |
+
"loss": 0.1541,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.04183354527411668,
|
| 314 |
+
"grad_norm": 10.125,
|
| 315 |
+
"learning_rate": 1.9650164944723116e-05,
|
| 316 |
+
"loss": 0.222,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.04278430766671024,
|
| 321 |
+
"grad_norm": 7.46875,
|
| 322 |
+
"learning_rate": 1.961261695938319e-05,
|
| 323 |
+
"loss": 0.1897,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.043735070059303806,
|
| 328 |
+
"grad_norm": 8.125,
|
| 329 |
+
"learning_rate": 1.9573194975320672e-05,
|
| 330 |
+
"loss": 0.2019,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.04468583245189737,
|
| 335 |
+
"grad_norm": 7.3125,
|
| 336 |
+
"learning_rate": 1.9531906677929472e-05,
|
| 337 |
+
"loss": 0.1736,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.045636594844490924,
|
| 342 |
+
"grad_norm": 9.6875,
|
| 343 |
+
"learning_rate": 1.9488760116444966e-05,
|
| 344 |
+
"loss": 0.1836,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.04658735723708449,
|
| 349 |
+
"grad_norm": 14.0625,
|
| 350 |
+
"learning_rate": 1.944376370237481e-05,
|
| 351 |
+
"loss": 0.1787,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.04753811962967805,
|
| 356 |
+
"grad_norm": 12.875,
|
| 357 |
+
"learning_rate": 1.9396926207859085e-05,
|
| 358 |
+
"loss": 0.1524,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.04848888202227161,
|
| 363 |
+
"grad_norm": 4.03125,
|
| 364 |
+
"learning_rate": 1.9348256763960146e-05,
|
| 365 |
+
"loss": 0.2198,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.04943964441486517,
|
| 370 |
+
"grad_norm": 10.375,
|
| 371 |
+
"learning_rate": 1.9297764858882516e-05,
|
| 372 |
+
"loss": 0.1712,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.05039040680745873,
|
| 377 |
+
"grad_norm": 10.625,
|
| 378 |
+
"learning_rate": 1.9245460336123136e-05,
|
| 379 |
+
"loss": 0.1702,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.05134116920005229,
|
| 384 |
+
"grad_norm": 12.375,
|
| 385 |
+
"learning_rate": 1.9191353392552346e-05,
|
| 386 |
+
"loss": 0.1558,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.052291931592645856,
|
| 391 |
+
"grad_norm": 4.21875,
|
| 392 |
+
"learning_rate": 1.913545457642601e-05,
|
| 393 |
+
"loss": 0.1701,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 0.05324269398523941,
|
| 398 |
+
"grad_norm": 5.53125,
|
| 399 |
+
"learning_rate": 1.907777478532909e-05,
|
| 400 |
+
"loss": 0.1729,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.054193456377832974,
|
| 405 |
+
"grad_norm": 11.5625,
|
| 406 |
+
"learning_rate": 1.901832526405114e-05,
|
| 407 |
+
"loss": 0.1881,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.05514421877042654,
|
| 412 |
+
"grad_norm": 6.40625,
|
| 413 |
+
"learning_rate": 1.895711760239413e-05,
|
| 414 |
+
"loss": 0.1452,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.0560949811630201,
|
| 419 |
+
"grad_norm": 6.3125,
|
| 420 |
+
"learning_rate": 1.889416373291298e-05,
|
| 421 |
+
"loss": 0.1457,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.057045743555613655,
|
| 426 |
+
"grad_norm": 9.625,
|
| 427 |
+
"learning_rate": 1.8829475928589272e-05,
|
| 428 |
+
"loss": 0.1376,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.05799650594820722,
|
| 433 |
+
"grad_norm": 10.6875,
|
| 434 |
+
"learning_rate": 1.8763066800438638e-05,
|
| 435 |
+
"loss": 0.1383,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 0.05894726834080078,
|
| 440 |
+
"grad_norm": 8.1875,
|
| 441 |
+
"learning_rate": 1.869494929505219e-05,
|
| 442 |
+
"loss": 0.1396,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.05989803073339434,
|
| 447 |
+
"grad_norm": 10.5,
|
| 448 |
+
"learning_rate": 1.8625136692072577e-05,
|
| 449 |
+
"loss": 0.1444,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.0608487931259879,
|
| 454 |
+
"grad_norm": 9.0625,
|
| 455 |
+
"learning_rate": 1.855364260160507e-05,
|
| 456 |
+
"loss": 0.1622,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.06179955551858146,
|
| 461 |
+
"grad_norm": 7.84375,
|
| 462 |
+
"learning_rate": 1.848048096156426e-05,
|
| 463 |
+
"loss": 0.1606,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.06275031791117502,
|
| 468 |
+
"grad_norm": 10.25,
|
| 469 |
+
"learning_rate": 1.8405666034956842e-05,
|
| 470 |
+
"loss": 0.1633,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.06370108030376859,
|
| 475 |
+
"grad_norm": 2.390625,
|
| 476 |
+
"learning_rate": 1.8329212407100996e-05,
|
| 477 |
+
"loss": 0.1472,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 0.06465184269636215,
|
| 482 |
+
"grad_norm": 7.5625,
|
| 483 |
+
"learning_rate": 1.8251134982782952e-05,
|
| 484 |
+
"loss": 0.1419,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.06560260508895571,
|
| 489 |
+
"grad_norm": 10.6875,
|
| 490 |
+
"learning_rate": 1.8171448983351284e-05,
|
| 491 |
+
"loss": 0.1259,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.06655336748154926,
|
| 496 |
+
"grad_norm": 4.46875,
|
| 497 |
+
"learning_rate": 1.8090169943749477e-05,
|
| 498 |
+
"loss": 0.1407,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.06750412987414282,
|
| 503 |
+
"grad_norm": 8.8125,
|
| 504 |
+
"learning_rate": 1.8007313709487334e-05,
|
| 505 |
+
"loss": 0.1515,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.06845489226673639,
|
| 510 |
+
"grad_norm": 5.40625,
|
| 511 |
+
"learning_rate": 1.792289643355191e-05,
|
| 512 |
+
"loss": 0.1671,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.06940565465932995,
|
| 517 |
+
"grad_norm": 14.5625,
|
| 518 |
+
"learning_rate": 1.78369345732584e-05,
|
| 519 |
+
"loss": 0.1637,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.07035641705192351,
|
| 524 |
+
"grad_norm": 9.75,
|
| 525 |
+
"learning_rate": 1.7749444887041797e-05,
|
| 526 |
+
"loss": 0.1348,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.07130717944451707,
|
| 531 |
+
"grad_norm": 2.1875,
|
| 532 |
+
"learning_rate": 1.766044443118978e-05,
|
| 533 |
+
"loss": 0.1444,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.07225794183711064,
|
| 538 |
+
"grad_norm": 5.125,
|
| 539 |
+
"learning_rate": 1.7569950556517566e-05,
|
| 540 |
+
"loss": 0.14,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.0732087042297042,
|
| 545 |
+
"grad_norm": 9.0,
|
| 546 |
+
"learning_rate": 1.747798090498532e-05,
|
| 547 |
+
"loss": 0.1445,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.07415946662229775,
|
| 552 |
+
"grad_norm": 9.4375,
|
| 553 |
+
"learning_rate": 1.7384553406258842e-05,
|
| 554 |
+
"loss": 0.1284,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.07511022901489131,
|
| 559 |
+
"grad_norm": 2.59375,
|
| 560 |
+
"learning_rate": 1.7289686274214116e-05,
|
| 561 |
+
"loss": 0.1347,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 0.07606099140748487,
|
| 566 |
+
"grad_norm": 3.890625,
|
| 567 |
+
"learning_rate": 1.7193398003386514e-05,
|
| 568 |
+
"loss": 0.1438,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.07701175380007844,
|
| 573 |
+
"grad_norm": 2.03125,
|
| 574 |
+
"learning_rate": 1.709570736536521e-05,
|
| 575 |
+
"loss": 0.1326,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 0.077962516192672,
|
| 580 |
+
"grad_norm": 7.40625,
|
| 581 |
+
"learning_rate": 1.6996633405133656e-05,
|
| 582 |
+
"loss": 0.1238,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.07891327858526556,
|
| 587 |
+
"grad_norm": 8.125,
|
| 588 |
+
"learning_rate": 1.68961954373567e-05,
|
| 589 |
+
"loss": 0.1181,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.07986404097785912,
|
| 594 |
+
"grad_norm": 3.84375,
|
| 595 |
+
"learning_rate": 1.6794413042615168e-05,
|
| 596 |
+
"loss": 0.1297,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.08081480337045269,
|
| 601 |
+
"grad_norm": 7.78125,
|
| 602 |
+
"learning_rate": 1.6691306063588583e-05,
|
| 603 |
+
"loss": 0.1267,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 0.08176556576304625,
|
| 608 |
+
"grad_norm": 5.78125,
|
| 609 |
+
"learning_rate": 1.6586894601186804e-05,
|
| 610 |
+
"loss": 0.1357,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.0827163281556398,
|
| 615 |
+
"grad_norm": 7.75,
|
| 616 |
+
"learning_rate": 1.6481199010631312e-05,
|
| 617 |
+
"loss": 0.1151,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 0.08366709054823336,
|
| 622 |
+
"grad_norm": 11.0625,
|
| 623 |
+
"learning_rate": 1.63742398974869e-05,
|
| 624 |
+
"loss": 0.1223,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.08461785294082692,
|
| 629 |
+
"grad_norm": 4.65625,
|
| 630 |
+
"learning_rate": 1.6266038113644605e-05,
|
| 631 |
+
"loss": 0.1195,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.08556861533342049,
|
| 636 |
+
"grad_norm": 4.125,
|
| 637 |
+
"learning_rate": 1.6156614753256583e-05,
|
| 638 |
+
"loss": 0.1255,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.08651937772601405,
|
| 643 |
+
"grad_norm": 6.40625,
|
| 644 |
+
"learning_rate": 1.6045991148623752e-05,
|
| 645 |
+
"loss": 0.1257,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 0.08747014011860761,
|
| 650 |
+
"grad_norm": 4.59375,
|
| 651 |
+
"learning_rate": 1.5934188866037017e-05,
|
| 652 |
+
"loss": 0.1175,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.08842090251120117,
|
| 657 |
+
"grad_norm": 6.78125,
|
| 658 |
+
"learning_rate": 1.5821229701572897e-05,
|
| 659 |
+
"loss": 0.109,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.08937166490379474,
|
| 664 |
+
"grad_norm": 5.78125,
|
| 665 |
+
"learning_rate": 1.570713567684432e-05,
|
| 666 |
+
"loss": 0.1239,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.09032242729638829,
|
| 671 |
+
"grad_norm": 6.0625,
|
| 672 |
+
"learning_rate": 1.5591929034707468e-05,
|
| 673 |
+
"loss": 0.1139,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.09127318968898185,
|
| 678 |
+
"grad_norm": 8.5625,
|
| 679 |
+
"learning_rate": 1.5475632234925505e-05,
|
| 680 |
+
"loss": 0.1178,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.09222395208157541,
|
| 685 |
+
"grad_norm": 5.71875,
|
| 686 |
+
"learning_rate": 1.5358267949789968e-05,
|
| 687 |
+
"loss": 0.1135,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.09317471447416897,
|
| 692 |
+
"grad_norm": 9.5,
|
| 693 |
+
"learning_rate": 1.5239859059700794e-05,
|
| 694 |
+
"loss": 0.1115,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.09412547686676254,
|
| 699 |
+
"grad_norm": 4.125,
|
| 700 |
+
"learning_rate": 1.5120428648705716e-05,
|
| 701 |
+
"loss": 0.1131,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 0.0950762392593561,
|
| 706 |
+
"grad_norm": 1.890625,
|
| 707 |
+
"learning_rate": 1.5000000000000002e-05,
|
| 708 |
+
"loss": 0.1182,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.09602700165194966,
|
| 713 |
+
"grad_norm": 6.59375,
|
| 714 |
+
"learning_rate": 1.4878596591387329e-05,
|
| 715 |
+
"loss": 0.1284,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.09697776404454322,
|
| 720 |
+
"grad_norm": 4.0,
|
| 721 |
+
"learning_rate": 1.4756242090702756e-05,
|
| 722 |
+
"loss": 0.11,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.09792852643713677,
|
| 727 |
+
"grad_norm": 3.421875,
|
| 728 |
+
"learning_rate": 1.463296035119862e-05,
|
| 729 |
+
"loss": 0.1243,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.09887928882973034,
|
| 734 |
+
"grad_norm": 10.5625,
|
| 735 |
+
"learning_rate": 1.4508775406894308e-05,
|
| 736 |
+
"loss": 0.1153,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.0998300512223239,
|
| 741 |
+
"grad_norm": 3.953125,
|
| 742 |
+
"learning_rate": 1.4383711467890776e-05,
|
| 743 |
+
"loss": 0.114,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.10078081361491746,
|
| 748 |
+
"grad_norm": 4.15625,
|
| 749 |
+
"learning_rate": 1.4257792915650728e-05,
|
| 750 |
+
"loss": 0.1108,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.10173157600751102,
|
| 755 |
+
"grad_norm": 8.125,
|
| 756 |
+
"learning_rate": 1.413104429824542e-05,
|
| 757 |
+
"loss": 0.1153,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.10268233840010459,
|
| 762 |
+
"grad_norm": 2.921875,
|
| 763 |
+
"learning_rate": 1.4003490325568953e-05,
|
| 764 |
+
"loss": 0.0994,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.10363310079269815,
|
| 769 |
+
"grad_norm": 6.21875,
|
| 770 |
+
"learning_rate": 1.3875155864521031e-05,
|
| 771 |
+
"loss": 0.105,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.10458386318529171,
|
| 776 |
+
"grad_norm": 8.6875,
|
| 777 |
+
"learning_rate": 1.3746065934159123e-05,
|
| 778 |
+
"loss": 0.1083,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.10553462557788526,
|
| 783 |
+
"grad_norm": 3.296875,
|
| 784 |
+
"learning_rate": 1.3616245700820922e-05,
|
| 785 |
+
"loss": 0.1042,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.10648538797047882,
|
| 790 |
+
"grad_norm": 7.5,
|
| 791 |
+
"learning_rate": 1.3485720473218153e-05,
|
| 792 |
+
"loss": 0.1054,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.10743615036307239,
|
| 797 |
+
"grad_norm": 7.375,
|
| 798 |
+
"learning_rate": 1.3354515697502552e-05,
|
| 799 |
+
"loss": 0.1079,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.10838691275566595,
|
| 804 |
+
"grad_norm": 2.75,
|
| 805 |
+
"learning_rate": 1.3222656952305113e-05,
|
| 806 |
+
"loss": 0.1056,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.10933767514825951,
|
| 811 |
+
"grad_norm": 5.5625,
|
| 812 |
+
"learning_rate": 1.3090169943749475e-05,
|
| 813 |
+
"loss": 0.1065,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.11028843754085307,
|
| 818 |
+
"grad_norm": 3.25,
|
| 819 |
+
"learning_rate": 1.2957080500440469e-05,
|
| 820 |
+
"loss": 0.0932,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.11123919993344664,
|
| 825 |
+
"grad_norm": 3.515625,
|
| 826 |
+
"learning_rate": 1.2823414568428767e-05,
|
| 827 |
+
"loss": 0.1043,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.1121899623260402,
|
| 832 |
+
"grad_norm": 4.6875,
|
| 833 |
+
"learning_rate": 1.2689198206152657e-05,
|
| 834 |
+
"loss": 0.1027,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.11314072471863375,
|
| 839 |
+
"grad_norm": 4.03125,
|
| 840 |
+
"learning_rate": 1.2554457579357906e-05,
|
| 841 |
+
"loss": 0.1079,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.11409148711122731,
|
| 846 |
+
"grad_norm": 2.03125,
|
| 847 |
+
"learning_rate": 1.2419218955996677e-05,
|
| 848 |
+
"loss": 0.1074,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.11504224950382087,
|
| 853 |
+
"grad_norm": 3.140625,
|
| 854 |
+
"learning_rate": 1.2283508701106559e-05,
|
| 855 |
+
"loss": 0.1019,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.11599301189641444,
|
| 860 |
+
"grad_norm": 2.03125,
|
| 861 |
+
"learning_rate": 1.2147353271670634e-05,
|
| 862 |
+
"loss": 0.1246,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.116943774289008,
|
| 867 |
+
"grad_norm": 7.09375,
|
| 868 |
+
"learning_rate": 1.2010779211459649e-05,
|
| 869 |
+
"loss": 0.0946,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.11789453668160156,
|
| 874 |
+
"grad_norm": 1.2421875,
|
| 875 |
+
"learning_rate": 1.187381314585725e-05,
|
| 876 |
+
"loss": 0.1007,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.11884529907419512,
|
| 881 |
+
"grad_norm": 4.03125,
|
| 882 |
+
"learning_rate": 1.1736481776669307e-05,
|
| 883 |
+
"loss": 0.105,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.11979606146678869,
|
| 888 |
+
"grad_norm": 2.796875,
|
| 889 |
+
"learning_rate": 1.159881187691835e-05,
|
| 890 |
+
"loss": 0.0966,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.12074682385938225,
|
| 895 |
+
"grad_norm": 9.8125,
|
| 896 |
+
"learning_rate": 1.1460830285624119e-05,
|
| 897 |
+
"loss": 0.1009,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.1216975862519758,
|
| 902 |
+
"grad_norm": 2.375,
|
| 903 |
+
"learning_rate": 1.1322563902571227e-05,
|
| 904 |
+
"loss": 0.0893,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.12264834864456936,
|
| 909 |
+
"grad_norm": 2.53125,
|
| 910 |
+
"learning_rate": 1.1184039683065014e-05,
|
| 911 |
+
"loss": 0.0942,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.12359911103716292,
|
| 916 |
+
"grad_norm": 3.28125,
|
| 917 |
+
"learning_rate": 1.1045284632676535e-05,
|
| 918 |
+
"loss": 0.101,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.12454987342975649,
|
| 923 |
+
"grad_norm": 5.84375,
|
| 924 |
+
"learning_rate": 1.0906325801977804e-05,
|
| 925 |
+
"loss": 0.0929,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.12550063582235005,
|
| 930 |
+
"grad_norm": 1.6640625,
|
| 931 |
+
"learning_rate": 1.0767190281268187e-05,
|
| 932 |
+
"loss": 0.0909,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.1264513982149436,
|
| 937 |
+
"grad_norm": 5.125,
|
| 938 |
+
"learning_rate": 1.0627905195293135e-05,
|
| 939 |
+
"loss": 0.1023,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 0.12740216060753717,
|
| 944 |
+
"grad_norm": 5.28125,
|
| 945 |
+
"learning_rate": 1.0488497697956134e-05,
|
| 946 |
+
"loss": 0.0894,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 0.12835292300013074,
|
| 951 |
+
"grad_norm": 5.375,
|
| 952 |
+
"learning_rate": 1.0348994967025012e-05,
|
| 953 |
+
"loss": 0.0912,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.1293036853927243,
|
| 958 |
+
"grad_norm": 3.203125,
|
| 959 |
+
"learning_rate": 1.0209424198833571e-05,
|
| 960 |
+
"loss": 0.0979,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 0.13025444778531786,
|
| 965 |
+
"grad_norm": 5.0,
|
| 966 |
+
"learning_rate": 1.0069812602979617e-05,
|
| 967 |
+
"loss": 0.0881,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.13120521017791142,
|
| 972 |
+
"grad_norm": 1.9921875,
|
| 973 |
+
"learning_rate": 9.930187397020385e-06,
|
| 974 |
+
"loss": 0.0837,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 0.132155972570505,
|
| 979 |
+
"grad_norm": 2.875,
|
| 980 |
+
"learning_rate": 9.790575801166432e-06,
|
| 981 |
+
"loss": 0.0905,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 0.13310673496309852,
|
| 986 |
+
"grad_norm": 2.234375,
|
| 987 |
+
"learning_rate": 9.651005032974994e-06,
|
| 988 |
+
"loss": 0.1473,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.13405749735569208,
|
| 993 |
+
"grad_norm": 3.171875,
|
| 994 |
+
"learning_rate": 9.511502302043867e-06,
|
| 995 |
+
"loss": 0.0834,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 0.13500825974828565,
|
| 1000 |
+
"grad_norm": 1.5234375,
|
| 1001 |
+
"learning_rate": 9.372094804706867e-06,
|
| 1002 |
+
"loss": 0.0865,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 0.1359590221408792,
|
| 1007 |
+
"grad_norm": 3.96875,
|
| 1008 |
+
"learning_rate": 9.232809718731815e-06,
|
| 1009 |
+
"loss": 0.089,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.13690978453347277,
|
| 1014 |
+
"grad_norm": 1.3125,
|
| 1015 |
+
"learning_rate": 9.093674198022201e-06,
|
| 1016 |
+
"loss": 0.0881,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.13786054692606634,
|
| 1021 |
+
"grad_norm": 2.484375,
|
| 1022 |
+
"learning_rate": 8.954715367323468e-06,
|
| 1023 |
+
"loss": 0.0819,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 0.1388113093186599,
|
| 1028 |
+
"grad_norm": 1.96875,
|
| 1029 |
+
"learning_rate": 8.815960316934991e-06,
|
| 1030 |
+
"loss": 0.0876,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 0.13976207171125346,
|
| 1035 |
+
"grad_norm": 5.375,
|
| 1036 |
+
"learning_rate": 8.677436097428775e-06,
|
| 1037 |
+
"loss": 0.0828,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 0.14071283410384702,
|
| 1042 |
+
"grad_norm": 2.015625,
|
| 1043 |
+
"learning_rate": 8.539169714375885e-06,
|
| 1044 |
+
"loss": 0.0827,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.14166359649644059,
|
| 1049 |
+
"grad_norm": 2.046875,
|
| 1050 |
+
"learning_rate": 8.401188123081653e-06,
|
| 1051 |
+
"loss": 0.0863,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.14261435888903415,
|
| 1056 |
+
"grad_norm": 1.625,
|
| 1057 |
+
"learning_rate": 8.263518223330698e-06,
|
| 1058 |
+
"loss": 0.0871,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.1435651212816277,
|
| 1063 |
+
"grad_norm": 2.0,
|
| 1064 |
+
"learning_rate": 8.126186854142752e-06,
|
| 1065 |
+
"loss": 0.0807,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 0.14451588367422127,
|
| 1070 |
+
"grad_norm": 1.59375,
|
| 1071 |
+
"learning_rate": 7.989220788540356e-06,
|
| 1072 |
+
"loss": 0.0788,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 0.14546664606681484,
|
| 1077 |
+
"grad_norm": 2.859375,
|
| 1078 |
+
"learning_rate": 7.852646728329368e-06,
|
| 1079 |
+
"loss": 0.0793,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.1464174084594084,
|
| 1084 |
+
"grad_norm": 2.609375,
|
| 1085 |
+
"learning_rate": 7.716491298893443e-06,
|
| 1086 |
+
"loss": 0.08,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 0.14736817085200196,
|
| 1091 |
+
"grad_norm": 2.65625,
|
| 1092 |
+
"learning_rate": 7.580781044003324e-06,
|
| 1093 |
+
"loss": 0.0749,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.1483189332445955,
|
| 1098 |
+
"grad_norm": 3.140625,
|
| 1099 |
+
"learning_rate": 7.445542420642097e-06,
|
| 1100 |
+
"loss": 0.0913,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 0.14926969563718906,
|
| 1105 |
+
"grad_norm": 2.515625,
|
| 1106 |
+
"learning_rate": 7.310801793847344e-06,
|
| 1107 |
+
"loss": 0.077,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 0.15022045802978262,
|
| 1112 |
+
"grad_norm": 3.40625,
|
| 1113 |
+
"learning_rate": 7.176585431571235e-06,
|
| 1114 |
+
"loss": 0.0787,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 0.15117122042237618,
|
| 1119 |
+
"grad_norm": 1.21875,
|
| 1120 |
+
"learning_rate": 7.042919499559538e-06,
|
| 1121 |
+
"loss": 0.0711,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 0.15212198281496975,
|
| 1126 |
+
"grad_norm": 2.03125,
|
| 1127 |
+
"learning_rate": 6.909830056250527e-06,
|
| 1128 |
+
"loss": 0.0762,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 0.1530727452075633,
|
| 1133 |
+
"grad_norm": 1.7109375,
|
| 1134 |
+
"learning_rate": 6.777343047694891e-06,
|
| 1135 |
+
"loss": 0.0768,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.15402350760015687,
|
| 1140 |
+
"grad_norm": 2.875,
|
| 1141 |
+
"learning_rate": 6.645484302497452e-06,
|
| 1142 |
+
"loss": 0.0798,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.15497426999275044,
|
| 1147 |
+
"grad_norm": 2.578125,
|
| 1148 |
+
"learning_rate": 6.5142795267818505e-06,
|
| 1149 |
+
"loss": 0.081,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 0.155925032385344,
|
| 1154 |
+
"grad_norm": 1.8125,
|
| 1155 |
+
"learning_rate": 6.383754299179079e-06,
|
| 1156 |
+
"loss": 0.0811,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 0.15687579477793756,
|
| 1161 |
+
"grad_norm": 2.9375,
|
| 1162 |
+
"learning_rate": 6.25393406584088e-06,
|
| 1163 |
+
"loss": 0.0754,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.15782655717053112,
|
| 1168 |
+
"grad_norm": 1.2265625,
|
| 1169 |
+
"learning_rate": 6.124844135478971e-06,
|
| 1170 |
+
"loss": 0.0738,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 0.15877731956312469,
|
| 1175 |
+
"grad_norm": 1.9921875,
|
| 1176 |
+
"learning_rate": 5.996509674431053e-06,
|
| 1177 |
+
"loss": 0.0739,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.15972808195571825,
|
| 1182 |
+
"grad_norm": 1.84375,
|
| 1183 |
+
"learning_rate": 5.868955701754584e-06,
|
| 1184 |
+
"loss": 0.0807,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 0.1606788443483118,
|
| 1189 |
+
"grad_norm": 1.3828125,
|
| 1190 |
+
"learning_rate": 5.742207084349274e-06,
|
| 1191 |
+
"loss": 0.0705,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 0.16162960674090537,
|
| 1196 |
+
"grad_norm": 2.796875,
|
| 1197 |
+
"learning_rate": 5.616288532109225e-06,
|
| 1198 |
+
"loss": 0.0782,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 0.16258036913349894,
|
| 1203 |
+
"grad_norm": 1.84375,
|
| 1204 |
+
"learning_rate": 5.491224593105695e-06,
|
| 1205 |
+
"loss": 0.083,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 0.1635311315260925,
|
| 1210 |
+
"grad_norm": 2.296875,
|
| 1211 |
+
"learning_rate": 5.367039648801386e-06,
|
| 1212 |
+
"loss": 0.0753,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 0.16448189391868603,
|
| 1217 |
+
"grad_norm": 1.9453125,
|
| 1218 |
+
"learning_rate": 5.243757909297247e-06,
|
| 1219 |
+
"loss": 0.08,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.1654326563112796,
|
| 1224 |
+
"grad_norm": 3.859375,
|
| 1225 |
+
"learning_rate": 5.121403408612672e-06,
|
| 1226 |
+
"loss": 0.0681,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 0.16638341870387316,
|
| 1231 |
+
"grad_norm": 2.1875,
|
| 1232 |
+
"learning_rate": 5.000000000000003e-06,
|
| 1233 |
+
"loss": 0.0752,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 0.16733418109646672,
|
| 1238 |
+
"grad_norm": 1.9375,
|
| 1239 |
+
"learning_rate": 4.879571351294287e-06,
|
| 1240 |
+
"loss": 0.08,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 0.16828494348906028,
|
| 1245 |
+
"grad_norm": 1.140625,
|
| 1246 |
+
"learning_rate": 4.76014094029921e-06,
|
| 1247 |
+
"loss": 0.0776,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 0.16923570588165385,
|
| 1252 |
+
"grad_norm": 2.453125,
|
| 1253 |
+
"learning_rate": 4.641732050210032e-06,
|
| 1254 |
+
"loss": 0.077,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.1701864682742474,
|
| 1259 |
+
"grad_norm": 1.453125,
|
| 1260 |
+
"learning_rate": 4.524367765074499e-06,
|
| 1261 |
+
"loss": 0.0776,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.17113723066684097,
|
| 1266 |
+
"grad_norm": 1.7890625,
|
| 1267 |
+
"learning_rate": 4.408070965292534e-06,
|
| 1268 |
+
"loss": 0.0677,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.17208799305943454,
|
| 1273 |
+
"grad_norm": 1.7734375,
|
| 1274 |
+
"learning_rate": 4.292864323155684e-06,
|
| 1275 |
+
"loss": 0.0711,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.1730387554520281,
|
| 1280 |
+
"grad_norm": 1.515625,
|
| 1281 |
+
"learning_rate": 4.178770298427107e-06,
|
| 1282 |
+
"loss": 0.0734,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.17398951784462166,
|
| 1287 |
+
"grad_norm": 1.4140625,
|
| 1288 |
+
"learning_rate": 4.065811133962987e-06,
|
| 1289 |
+
"loss": 0.0706,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 0.17494028023721522,
|
| 1294 |
+
"grad_norm": 1.3671875,
|
| 1295 |
+
"learning_rate": 3.954008851376252e-06,
|
| 1296 |
+
"loss": 0.0675,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 0.17589104262980879,
|
| 1301 |
+
"grad_norm": 1.671875,
|
| 1302 |
+
"learning_rate": 3.8433852467434175e-06,
|
| 1303 |
+
"loss": 0.0722,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.17684180502240235,
|
| 1308 |
+
"grad_norm": 1.171875,
|
| 1309 |
+
"learning_rate": 3.7339618863553983e-06,
|
| 1310 |
+
"loss": 0.0728,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 0.1777925674149959,
|
| 1315 |
+
"grad_norm": 1.03125,
|
| 1316 |
+
"learning_rate": 3.625760102513103e-06,
|
| 1317 |
+
"loss": 0.0715,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 0.17874332980758947,
|
| 1322 |
+
"grad_norm": 2.09375,
|
| 1323 |
+
"learning_rate": 3.5188009893686916e-06,
|
| 1324 |
+
"loss": 0.0691,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 0.179694092200183,
|
| 1329 |
+
"grad_norm": 0.9453125,
|
| 1330 |
+
"learning_rate": 3.4131053988131947e-06,
|
| 1331 |
+
"loss": 0.0693,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 0.18064485459277657,
|
| 1336 |
+
"grad_norm": 1.4921875,
|
| 1337 |
+
"learning_rate": 3.308693936411421e-06,
|
| 1338 |
+
"loss": 0.074,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 0.18159561698537013,
|
| 1343 |
+
"grad_norm": 1.2109375,
|
| 1344 |
+
"learning_rate": 3.2055869573848374e-06,
|
| 1345 |
+
"loss": 0.0688,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.1825463793779637,
|
| 1350 |
+
"grad_norm": 1.0625,
|
| 1351 |
+
"learning_rate": 3.103804562643302e-06,
|
| 1352 |
+
"loss": 0.0705,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 0.18349714177055726,
|
| 1357 |
+
"grad_norm": 1.171875,
|
| 1358 |
+
"learning_rate": 3.003366594866345e-06,
|
| 1359 |
+
"loss": 0.073,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.18444790416315082,
|
| 1364 |
+
"grad_norm": 2.4375,
|
| 1365 |
+
"learning_rate": 2.9042926346347932e-06,
|
| 1366 |
+
"loss": 0.0718,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.18539866655574438,
|
| 1371 |
+
"grad_norm": 1.25,
|
| 1372 |
+
"learning_rate": 2.8066019966134907e-06,
|
| 1373 |
+
"loss": 0.0702,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.18634942894833795,
|
| 1378 |
+
"grad_norm": 1.1015625,
|
| 1379 |
+
"learning_rate": 2.7103137257858867e-06,
|
| 1380 |
+
"loss": 0.0746,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 0.1873001913409315,
|
| 1385 |
+
"grad_norm": 2.328125,
|
| 1386 |
+
"learning_rate": 2.615446593741161e-06,
|
| 1387 |
+
"loss": 0.0691,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.18825095373352507,
|
| 1392 |
+
"grad_norm": 2.03125,
|
| 1393 |
+
"learning_rate": 2.522019095014683e-06,
|
| 1394 |
+
"loss": 0.0695,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.18920171612611864,
|
| 1399 |
+
"grad_norm": 1.515625,
|
| 1400 |
+
"learning_rate": 2.4300494434824373e-06,
|
| 1401 |
+
"loss": 0.0709,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.1901524785187122,
|
| 1406 |
+
"grad_norm": 1.34375,
|
| 1407 |
+
"learning_rate": 2.339555568810221e-06,
|
| 1408 |
+
"loss": 0.0683,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 0.19110324091130576,
|
| 1413 |
+
"grad_norm": 2.125,
|
| 1414 |
+
"learning_rate": 2.2505551129582047e-06,
|
| 1415 |
+
"loss": 0.0673,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 0.19205400330389932,
|
| 1420 |
+
"grad_norm": 0.78125,
|
| 1421 |
+
"learning_rate": 2.163065426741603e-06,
|
| 1422 |
+
"loss": 0.0715,
|
| 1423 |
+
"step": 2020
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 0.19300476569649289,
|
| 1427 |
+
"grad_norm": 1.34375,
|
| 1428 |
+
"learning_rate": 2.0771035664480944e-06,
|
| 1429 |
+
"loss": 0.0675,
|
| 1430 |
+
"step": 2030
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.19395552808908645,
|
| 1434 |
+
"grad_norm": 1.2421875,
|
| 1435 |
+
"learning_rate": 1.9926862905126663e-06,
|
| 1436 |
+
"loss": 0.0757,
|
| 1437 |
+
"step": 2040
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 0.19490629048167998,
|
| 1441 |
+
"grad_norm": 1.1953125,
|
| 1442 |
+
"learning_rate": 1.9098300562505266e-06,
|
| 1443 |
+
"loss": 0.0682,
|
| 1444 |
+
"step": 2050
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 0.19585705287427355,
|
| 1448 |
+
"grad_norm": 1.4375,
|
| 1449 |
+
"learning_rate": 1.8285510166487154e-06,
|
| 1450 |
+
"loss": 0.0684,
|
| 1451 |
+
"step": 2060
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 0.1968078152668671,
|
| 1455 |
+
"grad_norm": 0.9296875,
|
| 1456 |
+
"learning_rate": 1.7488650172170496e-06,
|
| 1457 |
+
"loss": 0.0719,
|
| 1458 |
+
"step": 2070
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 0.19775857765946067,
|
| 1462 |
+
"grad_norm": 1.5,
|
| 1463 |
+
"learning_rate": 1.6707875928990059e-06,
|
| 1464 |
+
"loss": 0.0695,
|
| 1465 |
+
"step": 2080
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 0.19870934005205423,
|
| 1469 |
+
"grad_norm": 0.98828125,
|
| 1470 |
+
"learning_rate": 1.5943339650431578e-06,
|
| 1471 |
+
"loss": 0.0754,
|
| 1472 |
+
"step": 2090
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.1996601024446478,
|
| 1476 |
+
"grad_norm": 1.2109375,
|
| 1477 |
+
"learning_rate": 1.5195190384357405e-06,
|
| 1478 |
+
"loss": 0.071,
|
| 1479 |
+
"step": 2100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 0.20061086483724136,
|
| 1483 |
+
"grad_norm": 0.85546875,
|
| 1484 |
+
"learning_rate": 1.446357398394934e-06,
|
| 1485 |
+
"loss": 0.0721,
|
| 1486 |
+
"step": 2110
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 0.20156162722983492,
|
| 1490 |
+
"grad_norm": 1.0390625,
|
| 1491 |
+
"learning_rate": 1.3748633079274254e-06,
|
| 1492 |
+
"loss": 0.0687,
|
| 1493 |
+
"step": 2120
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.20251238962242848,
|
| 1497 |
+
"grad_norm": 1.203125,
|
| 1498 |
+
"learning_rate": 1.30505070494781e-06,
|
| 1499 |
+
"loss": 0.0706,
|
| 1500 |
+
"step": 2130
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 0.20346315201502205,
|
| 1504 |
+
"grad_norm": 1.0859375,
|
| 1505 |
+
"learning_rate": 1.2369331995613664e-06,
|
| 1506 |
+
"loss": 0.0736,
|
| 1507 |
+
"step": 2140
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 0.2044139144076156,
|
| 1511 |
+
"grad_norm": 1.3359375,
|
| 1512 |
+
"learning_rate": 1.1705240714107301e-06,
|
| 1513 |
+
"loss": 0.0676,
|
| 1514 |
+
"step": 2150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.20536467680020917,
|
| 1518 |
+
"grad_norm": 1.078125,
|
| 1519 |
+
"learning_rate": 1.1058362670870248e-06,
|
| 1520 |
+
"loss": 0.0671,
|
| 1521 |
+
"step": 2160
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.20631543919280274,
|
| 1525 |
+
"grad_norm": 1.515625,
|
| 1526 |
+
"learning_rate": 1.042882397605871e-06,
|
| 1527 |
+
"loss": 0.0674,
|
| 1528 |
+
"step": 2170
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 0.2072662015853963,
|
| 1532 |
+
"grad_norm": 0.99609375,
|
| 1533 |
+
"learning_rate": 9.816747359488632e-07,
|
| 1534 |
+
"loss": 0.0674,
|
| 1535 |
+
"step": 2180
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 0.20821696397798986,
|
| 1539 |
+
"grad_norm": 1.4609375,
|
| 1540 |
+
"learning_rate": 9.222252146709143e-07,
|
| 1541 |
+
"loss": 0.0654,
|
| 1542 |
+
"step": 2190
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 0.20916772637058342,
|
| 1546 |
+
"grad_norm": 1.203125,
|
| 1547 |
+
"learning_rate": 8.645454235739903e-07,
|
| 1548 |
+
"loss": 0.0662,
|
| 1549 |
+
"step": 2200
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 0.21011848876317699,
|
| 1553 |
+
"grad_norm": 1.46875,
|
| 1554 |
+
"learning_rate": 8.086466074476562e-07,
|
| 1555 |
+
"loss": 0.0682,
|
| 1556 |
+
"step": 2210
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.21106925115577052,
|
| 1560 |
+
"grad_norm": 1.3515625,
|
| 1561 |
+
"learning_rate": 7.545396638768698e-07,
|
| 1562 |
+
"loss": 0.0701,
|
| 1563 |
+
"step": 2220
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.21202001354836408,
|
| 1567 |
+
"grad_norm": 1.703125,
|
| 1568 |
+
"learning_rate": 7.022351411174866e-07,
|
| 1569 |
+
"loss": 0.0654,
|
| 1570 |
+
"step": 2230
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 0.21297077594095765,
|
| 1574 |
+
"grad_norm": 1.140625,
|
| 1575 |
+
"learning_rate": 6.517432360398556e-07,
|
| 1576 |
+
"loss": 0.0722,
|
| 1577 |
+
"step": 2240
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 0.2139215383335512,
|
| 1581 |
+
"grad_norm": 1.7421875,
|
| 1582 |
+
"learning_rate": 6.030737921409169e-07,
|
| 1583 |
+
"loss": 0.073,
|
| 1584 |
+
"step": 2250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 0.21487230072614477,
|
| 1588 |
+
"grad_norm": 1.3125,
|
| 1589 |
+
"learning_rate": 5.562362976251901e-07,
|
| 1590 |
+
"loss": 0.0743,
|
| 1591 |
+
"step": 2260
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 0.21582306311873833,
|
| 1595 |
+
"grad_norm": 1.6015625,
|
| 1596 |
+
"learning_rate": 5.112398835550348e-07,
|
| 1597 |
+
"loss": 0.0689,
|
| 1598 |
+
"step": 2270
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.2167738255113319,
|
| 1602 |
+
"grad_norm": 2.078125,
|
| 1603 |
+
"learning_rate": 4.6809332207053083e-07,
|
| 1604 |
+
"loss": 0.0725,
|
| 1605 |
+
"step": 2280
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 0.21772458790392546,
|
| 1609 |
+
"grad_norm": 1.5546875,
|
| 1610 |
+
"learning_rate": 4.268050246793276e-07,
|
| 1611 |
+
"loss": 0.0724,
|
| 1612 |
+
"step": 2290
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.21867535029651902,
|
| 1616 |
+
"grad_norm": 1.3203125,
|
| 1617 |
+
"learning_rate": 3.8738304061681107e-07,
|
| 1618 |
+
"loss": 0.0725,
|
| 1619 |
+
"step": 2300
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 0.21962611268911258,
|
| 1623 |
+
"grad_norm": 1.7265625,
|
| 1624 |
+
"learning_rate": 3.498350552768859e-07,
|
| 1625 |
+
"loss": 0.072,
|
| 1626 |
+
"step": 2310
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 0.22057687508170615,
|
| 1630 |
+
"grad_norm": 1.0703125,
|
| 1631 |
+
"learning_rate": 3.1416838871368925e-07,
|
| 1632 |
+
"loss": 0.0708,
|
| 1633 |
+
"step": 2320
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 0.2215276374742997,
|
| 1637 |
+
"grad_norm": 1.1328125,
|
| 1638 |
+
"learning_rate": 2.8038999421453827e-07,
|
| 1639 |
+
"loss": 0.0699,
|
| 1640 |
+
"step": 2330
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.22247839986689327,
|
| 1644 |
+
"grad_norm": 1.140625,
|
| 1645 |
+
"learning_rate": 2.4850645694436736e-07,
|
| 1646 |
+
"loss": 0.0701,
|
| 1647 |
+
"step": 2340
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.22342916225948684,
|
| 1651 |
+
"grad_norm": 1.109375,
|
| 1652 |
+
"learning_rate": 2.1852399266194312e-07,
|
| 1653 |
+
"loss": 0.0681,
|
| 1654 |
+
"step": 2350
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 0.2243799246520804,
|
| 1658 |
+
"grad_norm": 1.0625,
|
| 1659 |
+
"learning_rate": 1.9044844650808468e-07,
|
| 1660 |
+
"loss": 0.0761,
|
| 1661 |
+
"step": 2360
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 0.22533068704467396,
|
| 1665 |
+
"grad_norm": 1.3125,
|
| 1666 |
+
"learning_rate": 1.6428529186614195e-07,
|
| 1667 |
+
"loss": 0.0712,
|
| 1668 |
+
"step": 2370
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 0.2262814494372675,
|
| 1672 |
+
"grad_norm": 1.7734375,
|
| 1673 |
+
"learning_rate": 1.400396292949513e-07,
|
| 1674 |
+
"loss": 0.0718,
|
| 1675 |
+
"step": 2380
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 0.22723221182986106,
|
| 1679 |
+
"grad_norm": 0.8203125,
|
| 1680 |
+
"learning_rate": 1.1771618553447217e-07,
|
| 1681 |
+
"loss": 0.0685,
|
| 1682 |
+
"step": 2390
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.22818297422245462,
|
| 1686 |
+
"grad_norm": 1.28125,
|
| 1687 |
+
"learning_rate": 9.731931258429638e-08,
|
| 1688 |
+
"loss": 0.0678,
|
| 1689 |
+
"step": 2400
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 0.22913373661504818,
|
| 1693 |
+
"grad_norm": 1.0625,
|
| 1694 |
+
"learning_rate": 7.885298685522235e-08,
|
| 1695 |
+
"loss": 0.07,
|
| 1696 |
+
"step": 2410
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 0.23008449900764175,
|
| 1700 |
+
"grad_norm": 1.2578125,
|
| 1701 |
+
"learning_rate": 6.232080839403631e-08,
|
| 1702 |
+
"loss": 0.0676,
|
| 1703 |
+
"step": 2420
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 0.2310352614002353,
|
| 1707 |
+
"grad_norm": 1.078125,
|
| 1708 |
+
"learning_rate": 4.772600018168816e-08,
|
| 1709 |
+
"loss": 0.0725,
|
| 1710 |
+
"step": 2430
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 0.23198602379282887,
|
| 1714 |
+
"grad_norm": 0.984375,
|
| 1715 |
+
"learning_rate": 3.50714075049563e-08,
|
| 1716 |
+
"loss": 0.0744,
|
| 1717 |
+
"step": 2440
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 0.23293678618542243,
|
| 1721 |
+
"grad_norm": 1.1875,
|
| 1722 |
+
"learning_rate": 2.4359497401758026e-08,
|
| 1723 |
+
"loss": 0.0712,
|
| 1724 |
+
"step": 2450
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.233887548578016,
|
| 1728 |
+
"grad_norm": 1.390625,
|
| 1729 |
+
"learning_rate": 1.5592358180189782e-08,
|
| 1730 |
+
"loss": 0.0724,
|
| 1731 |
+
"step": 2460
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.23483831097060956,
|
| 1735 |
+
"grad_norm": 1.3359375,
|
| 1736 |
+
"learning_rate": 8.771699011416169e-09,
|
| 1737 |
+
"loss": 0.0679,
|
| 1738 |
+
"step": 2470
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 0.23578907336320312,
|
| 1742 |
+
"grad_norm": 1.359375,
|
| 1743 |
+
"learning_rate": 3.898849596456477e-09,
|
| 1744 |
+
"loss": 0.0688,
|
| 1745 |
+
"step": 2480
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 0.23673983575579668,
|
| 1749 |
+
"grad_norm": 1.4296875,
|
| 1750 |
+
"learning_rate": 9.74759906957612e-10,
|
| 1751 |
+
"loss": 0.0701,
|
| 1752 |
+
"step": 2490
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 0.23769059814839025,
|
| 1756 |
+
"grad_norm": 1.8984375,
|
| 1757 |
+
"learning_rate": 0.0,
|
| 1758 |
+
"loss": 0.0727,
|
| 1759 |
+
"step": 2500
|
| 1760 |
+
}
|
| 1761 |
+
],
|
| 1762 |
+
"logging_steps": 10,
|
| 1763 |
+
"max_steps": 2500,
|
| 1764 |
+
"num_input_tokens_seen": 0,
|
| 1765 |
+
"num_train_epochs": 1,
|
| 1766 |
+
"save_steps": 500,
|
| 1767 |
+
"stateful_callbacks": {
|
| 1768 |
+
"TrainerControl": {
|
| 1769 |
+
"args": {
|
| 1770 |
+
"should_epoch_stop": false,
|
| 1771 |
+
"should_evaluate": false,
|
| 1772 |
+
"should_log": false,
|
| 1773 |
+
"should_save": true,
|
| 1774 |
+
"should_training_stop": true
|
| 1775 |
+
},
|
| 1776 |
+
"attributes": {}
|
| 1777 |
+
}
|
| 1778 |
+
},
|
| 1779 |
+
"total_flos": 9.5664155394048e+17,
|
| 1780 |
+
"train_batch_size": 4,
|
| 1781 |
+
"trial_name": null,
|
| 1782 |
+
"trial_params": null
|
| 1783 |
+
}
|
checkpoint-2500/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c70756fa796c5b8b678d3416e6432fe054129c477229591823e8c98733d24754
|
| 3 |
+
size 5368
|