End of training
Browse files- README.md +14 -2
- all_results.json +16 -0
- eval_results.json +10 -0
- tokenizer.json +1 -6
- train_results.json +9 -0
- trainer_state.json +560 -0
README.md
CHANGED
|
@@ -3,11 +3,23 @@ license: other
|
|
| 3 |
base_model: Qwen/Qwen1.5-4B
|
| 4 |
tags:
|
| 5 |
- generated_from_trainer
|
|
|
|
|
|
|
| 6 |
metrics:
|
| 7 |
- accuracy
|
| 8 |
model-index:
|
| 9 |
- name: lmind_hotpot_train8000_eval7405_v1_qa_Qwen_Qwen1.5-4B_5e-4_lora2
|
| 10 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
library_name: peft
|
| 12 |
---
|
| 13 |
|
|
@@ -16,7 +28,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 16 |
|
| 17 |
# lmind_hotpot_train8000_eval7405_v1_qa_Qwen_Qwen1.5-4B_5e-4_lora2
|
| 18 |
|
| 19 |
-
This model is a fine-tuned version of [Qwen/Qwen1.5-4B](https://huggingface.co/Qwen/Qwen1.5-4B) on
|
| 20 |
It achieves the following results on the evaluation set:
|
| 21 |
- Loss: 3.7344
|
| 22 |
- Accuracy: 0.4864
|
|
|
|
| 3 |
base_model: Qwen/Qwen1.5-4B
|
| 4 |
tags:
|
| 5 |
- generated_from_trainer
|
| 6 |
+
datasets:
|
| 7 |
+
- tyzhu/lmind_hotpot_train8000_eval7405_v1_qa
|
| 8 |
metrics:
|
| 9 |
- accuracy
|
| 10 |
model-index:
|
| 11 |
- name: lmind_hotpot_train8000_eval7405_v1_qa_Qwen_Qwen1.5-4B_5e-4_lora2
|
| 12 |
+
results:
|
| 13 |
+
- task:
|
| 14 |
+
name: Causal Language Modeling
|
| 15 |
+
type: text-generation
|
| 16 |
+
dataset:
|
| 17 |
+
name: tyzhu/lmind_hotpot_train8000_eval7405_v1_qa
|
| 18 |
+
type: tyzhu/lmind_hotpot_train8000_eval7405_v1_qa
|
| 19 |
+
metrics:
|
| 20 |
+
- name: Accuracy
|
| 21 |
+
type: accuracy
|
| 22 |
+
value: 0.48644444444444446
|
| 23 |
library_name: peft
|
| 24 |
---
|
| 25 |
|
|
|
|
| 28 |
|
| 29 |
# lmind_hotpot_train8000_eval7405_v1_qa_Qwen_Qwen1.5-4B_5e-4_lora2
|
| 30 |
|
| 31 |
+
This model is a fine-tuned version of [Qwen/Qwen1.5-4B](https://huggingface.co/Qwen/Qwen1.5-4B) on the tyzhu/lmind_hotpot_train8000_eval7405_v1_qa dataset.
|
| 32 |
It achieves the following results on the evaluation set:
|
| 33 |
- Loss: 3.7344
|
| 34 |
- Accuracy: 0.4864
|
all_results.json
ADDED
|
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.48644444444444446,
|
| 4 |
+
"eval_loss": 3.7344048023223877,
|
| 5 |
+
"eval_runtime": 6.2207,
|
| 6 |
+
"eval_samples": 500,
|
| 7 |
+
"eval_samples_per_second": 80.377,
|
| 8 |
+
"eval_steps_per_second": 10.127,
|
| 9 |
+
"perplexity": 41.863101325727705,
|
| 10 |
+
"total_flos": 1.293538587312128e+17,
|
| 11 |
+
"train_loss": 0.6887707939147949,
|
| 12 |
+
"train_runtime": 11763.1434,
|
| 13 |
+
"train_samples": 8000,
|
| 14 |
+
"train_samples_per_second": 13.602,
|
| 15 |
+
"train_steps_per_second": 0.425
|
| 16 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"eval_accuracy": 0.48644444444444446,
|
| 4 |
+
"eval_loss": 3.7344048023223877,
|
| 5 |
+
"eval_runtime": 6.2207,
|
| 6 |
+
"eval_samples": 500,
|
| 7 |
+
"eval_samples_per_second": 80.377,
|
| 8 |
+
"eval_steps_per_second": 10.127,
|
| 9 |
+
"perplexity": 41.863101325727705
|
| 10 |
+
}
|
tokenizer.json
CHANGED
|
@@ -1,11 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"version": "1.0",
|
| 3 |
-
"truncation":
|
| 4 |
-
"direction": "Right",
|
| 5 |
-
"max_length": 1024,
|
| 6 |
-
"strategy": "LongestFirst",
|
| 7 |
-
"stride": 0
|
| 8 |
-
},
|
| 9 |
"padding": null,
|
| 10 |
"added_tokens": [
|
| 11 |
{
|
|
|
|
| 1 |
{
|
| 2 |
"version": "1.0",
|
| 3 |
+
"truncation": null,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4 |
"padding": null,
|
| 5 |
"added_tokens": [
|
| 6 |
{
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 20.0,
|
| 3 |
+
"total_flos": 1.293538587312128e+17,
|
| 4 |
+
"train_loss": 0.6887707939147949,
|
| 5 |
+
"train_runtime": 11763.1434,
|
| 6 |
+
"train_samples": 8000,
|
| 7 |
+
"train_samples_per_second": 13.602,
|
| 8 |
+
"train_steps_per_second": 0.425
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,560 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.4,
|
| 13 |
+
"grad_norm": 0.5550008416175842,
|
| 14 |
+
"learning_rate": 0.0005,
|
| 15 |
+
"loss": 2.3125,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.8,
|
| 20 |
+
"grad_norm": 0.5415581464767456,
|
| 21 |
+
"learning_rate": 0.0005,
|
| 22 |
+
"loss": 2.252,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 1.0,
|
| 27 |
+
"eval_accuracy": 0.5171428571428571,
|
| 28 |
+
"eval_loss": 2.316453218460083,
|
| 29 |
+
"eval_runtime": 6.073,
|
| 30 |
+
"eval_samples_per_second": 82.332,
|
| 31 |
+
"eval_steps_per_second": 10.374,
|
| 32 |
+
"step": 250
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 1.2,
|
| 36 |
+
"grad_norm": 0.6543728709220886,
|
| 37 |
+
"learning_rate": 0.0005,
|
| 38 |
+
"loss": 2.0311,
|
| 39 |
+
"step": 300
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 1.6,
|
| 43 |
+
"grad_norm": 0.7604817152023315,
|
| 44 |
+
"learning_rate": 0.0005,
|
| 45 |
+
"loss": 1.8327,
|
| 46 |
+
"step": 400
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 2.0,
|
| 50 |
+
"grad_norm": 0.7253739833831787,
|
| 51 |
+
"learning_rate": 0.0005,
|
| 52 |
+
"loss": 1.8363,
|
| 53 |
+
"step": 500
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 2.0,
|
| 57 |
+
"eval_accuracy": 0.5126984126984127,
|
| 58 |
+
"eval_loss": 2.42637038230896,
|
| 59 |
+
"eval_runtime": 6.6931,
|
| 60 |
+
"eval_samples_per_second": 74.704,
|
| 61 |
+
"eval_steps_per_second": 9.413,
|
| 62 |
+
"step": 500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 2.4,
|
| 66 |
+
"grad_norm": 0.9617753624916077,
|
| 67 |
+
"learning_rate": 0.0005,
|
| 68 |
+
"loss": 1.3209,
|
| 69 |
+
"step": 600
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 2.8,
|
| 73 |
+
"grad_norm": 0.9664514064788818,
|
| 74 |
+
"learning_rate": 0.0005,
|
| 75 |
+
"loss": 1.3801,
|
| 76 |
+
"step": 700
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 3.0,
|
| 80 |
+
"eval_accuracy": 0.5059047619047619,
|
| 81 |
+
"eval_loss": 2.612029790878296,
|
| 82 |
+
"eval_runtime": 5.9136,
|
| 83 |
+
"eval_samples_per_second": 84.551,
|
| 84 |
+
"eval_steps_per_second": 10.653,
|
| 85 |
+
"step": 750
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 3.2,
|
| 89 |
+
"grad_norm": 1.1638116836547852,
|
| 90 |
+
"learning_rate": 0.0005,
|
| 91 |
+
"loss": 1.1704,
|
| 92 |
+
"step": 800
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 3.6,
|
| 96 |
+
"grad_norm": 1.1339421272277832,
|
| 97 |
+
"learning_rate": 0.0005,
|
| 98 |
+
"loss": 0.97,
|
| 99 |
+
"step": 900
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 4.0,
|
| 103 |
+
"grad_norm": 1.150955080986023,
|
| 104 |
+
"learning_rate": 0.0005,
|
| 105 |
+
"loss": 1.0246,
|
| 106 |
+
"step": 1000
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 4.0,
|
| 110 |
+
"eval_accuracy": 0.5007936507936508,
|
| 111 |
+
"eval_loss": 2.861654758453369,
|
| 112 |
+
"eval_runtime": 5.8839,
|
| 113 |
+
"eval_samples_per_second": 84.978,
|
| 114 |
+
"eval_steps_per_second": 10.707,
|
| 115 |
+
"step": 1000
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 4.4,
|
| 119 |
+
"grad_norm": 1.0390375852584839,
|
| 120 |
+
"learning_rate": 0.0005,
|
| 121 |
+
"loss": 0.6748,
|
| 122 |
+
"step": 1100
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 4.8,
|
| 126 |
+
"grad_norm": 1.198516607284546,
|
| 127 |
+
"learning_rate": 0.0005,
|
| 128 |
+
"loss": 0.7286,
|
| 129 |
+
"step": 1200
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 5.0,
|
| 133 |
+
"eval_accuracy": 0.49587301587301585,
|
| 134 |
+
"eval_loss": 3.095292091369629,
|
| 135 |
+
"eval_runtime": 5.924,
|
| 136 |
+
"eval_samples_per_second": 84.403,
|
| 137 |
+
"eval_steps_per_second": 10.635,
|
| 138 |
+
"step": 1250
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 5.2,
|
| 142 |
+
"grad_norm": 0.8057821989059448,
|
| 143 |
+
"learning_rate": 0.0005,
|
| 144 |
+
"loss": 0.6452,
|
| 145 |
+
"step": 1300
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 5.6,
|
| 149 |
+
"grad_norm": 1.0558875799179077,
|
| 150 |
+
"learning_rate": 0.0005,
|
| 151 |
+
"loss": 0.5639,
|
| 152 |
+
"step": 1400
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 6.0,
|
| 156 |
+
"grad_norm": 1.0491156578063965,
|
| 157 |
+
"learning_rate": 0.0005,
|
| 158 |
+
"loss": 0.601,
|
| 159 |
+
"step": 1500
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 6.0,
|
| 163 |
+
"eval_accuracy": 0.49498412698412697,
|
| 164 |
+
"eval_loss": 3.2138681411743164,
|
| 165 |
+
"eval_runtime": 6.0868,
|
| 166 |
+
"eval_samples_per_second": 82.144,
|
| 167 |
+
"eval_steps_per_second": 10.35,
|
| 168 |
+
"step": 1500
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 6.4,
|
| 172 |
+
"grad_norm": 0.8526061177253723,
|
| 173 |
+
"learning_rate": 0.0005,
|
| 174 |
+
"loss": 0.4771,
|
| 175 |
+
"step": 1600
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 6.8,
|
| 179 |
+
"grad_norm": 1.302850365638733,
|
| 180 |
+
"learning_rate": 0.0005,
|
| 181 |
+
"loss": 0.5138,
|
| 182 |
+
"step": 1700
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 7.0,
|
| 186 |
+
"eval_accuracy": 0.49333333333333335,
|
| 187 |
+
"eval_loss": 3.2912352085113525,
|
| 188 |
+
"eval_runtime": 6.038,
|
| 189 |
+
"eval_samples_per_second": 82.808,
|
| 190 |
+
"eval_steps_per_second": 10.434,
|
| 191 |
+
"step": 1750
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 7.2,
|
| 195 |
+
"grad_norm": 0.7548580169677734,
|
| 196 |
+
"learning_rate": 0.0005,
|
| 197 |
+
"loss": 0.4833,
|
| 198 |
+
"step": 1800
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 7.6,
|
| 202 |
+
"grad_norm": 0.869629442691803,
|
| 203 |
+
"learning_rate": 0.0005,
|
| 204 |
+
"loss": 0.4611,
|
| 205 |
+
"step": 1900
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 8.0,
|
| 209 |
+
"grad_norm": 0.9541189074516296,
|
| 210 |
+
"learning_rate": 0.0005,
|
| 211 |
+
"loss": 0.4837,
|
| 212 |
+
"step": 2000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 8.0,
|
| 216 |
+
"eval_accuracy": 0.49,
|
| 217 |
+
"eval_loss": 3.4516727924346924,
|
| 218 |
+
"eval_runtime": 6.0156,
|
| 219 |
+
"eval_samples_per_second": 83.117,
|
| 220 |
+
"eval_steps_per_second": 10.473,
|
| 221 |
+
"step": 2000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 8.4,
|
| 225 |
+
"grad_norm": 0.7076862454414368,
|
| 226 |
+
"learning_rate": 0.0005,
|
| 227 |
+
"loss": 0.4238,
|
| 228 |
+
"step": 2100
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 8.8,
|
| 232 |
+
"grad_norm": 0.8201280236244202,
|
| 233 |
+
"learning_rate": 0.0005,
|
| 234 |
+
"loss": 0.4506,
|
| 235 |
+
"step": 2200
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 9.0,
|
| 239 |
+
"eval_accuracy": 0.49114285714285716,
|
| 240 |
+
"eval_loss": 3.4106991291046143,
|
| 241 |
+
"eval_runtime": 5.8426,
|
| 242 |
+
"eval_samples_per_second": 85.579,
|
| 243 |
+
"eval_steps_per_second": 10.783,
|
| 244 |
+
"step": 2250
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 9.2,
|
| 248 |
+
"grad_norm": 0.7090291976928711,
|
| 249 |
+
"learning_rate": 0.0005,
|
| 250 |
+
"loss": 0.4432,
|
| 251 |
+
"step": 2300
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 9.6,
|
| 255 |
+
"grad_norm": 0.965961754322052,
|
| 256 |
+
"learning_rate": 0.0005,
|
| 257 |
+
"loss": 0.4297,
|
| 258 |
+
"step": 2400
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 10.0,
|
| 262 |
+
"grad_norm": 1.0484209060668945,
|
| 263 |
+
"learning_rate": 0.0005,
|
| 264 |
+
"loss": 0.4578,
|
| 265 |
+
"step": 2500
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 10.0,
|
| 269 |
+
"eval_accuracy": 0.49053968253968255,
|
| 270 |
+
"eval_loss": 3.47859787940979,
|
| 271 |
+
"eval_runtime": 6.2108,
|
| 272 |
+
"eval_samples_per_second": 80.505,
|
| 273 |
+
"eval_steps_per_second": 10.144,
|
| 274 |
+
"step": 2500
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 10.4,
|
| 278 |
+
"grad_norm": 0.7023801803588867,
|
| 279 |
+
"learning_rate": 0.0005,
|
| 280 |
+
"loss": 0.4151,
|
| 281 |
+
"step": 2600
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 10.8,
|
| 285 |
+
"grad_norm": 1.0452632904052734,
|
| 286 |
+
"learning_rate": 0.0005,
|
| 287 |
+
"loss": 0.4362,
|
| 288 |
+
"step": 2700
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 11.0,
|
| 292 |
+
"eval_accuracy": 0.4899047619047619,
|
| 293 |
+
"eval_loss": 3.5410351753234863,
|
| 294 |
+
"eval_runtime": 6.1027,
|
| 295 |
+
"eval_samples_per_second": 81.931,
|
| 296 |
+
"eval_steps_per_second": 10.323,
|
| 297 |
+
"step": 2750
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 11.2,
|
| 301 |
+
"grad_norm": 0.9101859331130981,
|
| 302 |
+
"learning_rate": 0.0005,
|
| 303 |
+
"loss": 0.4291,
|
| 304 |
+
"step": 2800
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 11.6,
|
| 308 |
+
"grad_norm": 0.9637263417243958,
|
| 309 |
+
"learning_rate": 0.0005,
|
| 310 |
+
"loss": 0.4218,
|
| 311 |
+
"step": 2900
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 12.0,
|
| 315 |
+
"grad_norm": 1.0658060312271118,
|
| 316 |
+
"learning_rate": 0.0005,
|
| 317 |
+
"loss": 0.4429,
|
| 318 |
+
"step": 3000
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 12.0,
|
| 322 |
+
"eval_accuracy": 0.4908571428571429,
|
| 323 |
+
"eval_loss": 3.565582275390625,
|
| 324 |
+
"eval_runtime": 6.1425,
|
| 325 |
+
"eval_samples_per_second": 81.4,
|
| 326 |
+
"eval_steps_per_second": 10.256,
|
| 327 |
+
"step": 3000
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 12.4,
|
| 331 |
+
"grad_norm": 0.7120447158813477,
|
| 332 |
+
"learning_rate": 0.0005,
|
| 333 |
+
"loss": 0.4027,
|
| 334 |
+
"step": 3100
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 12.8,
|
| 338 |
+
"grad_norm": 0.8874293565750122,
|
| 339 |
+
"learning_rate": 0.0005,
|
| 340 |
+
"loss": 0.4366,
|
| 341 |
+
"step": 3200
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 13.0,
|
| 345 |
+
"eval_accuracy": 0.48898412698412697,
|
| 346 |
+
"eval_loss": 3.542457103729248,
|
| 347 |
+
"eval_runtime": 5.9997,
|
| 348 |
+
"eval_samples_per_second": 83.338,
|
| 349 |
+
"eval_steps_per_second": 10.501,
|
| 350 |
+
"step": 3250
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 13.2,
|
| 354 |
+
"grad_norm": 0.8680710792541504,
|
| 355 |
+
"learning_rate": 0.0005,
|
| 356 |
+
"loss": 0.4258,
|
| 357 |
+
"step": 3300
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 13.6,
|
| 361 |
+
"grad_norm": 1.1757447719573975,
|
| 362 |
+
"learning_rate": 0.0005,
|
| 363 |
+
"loss": 0.4163,
|
| 364 |
+
"step": 3400
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 14.0,
|
| 368 |
+
"grad_norm": 1.1933009624481201,
|
| 369 |
+
"learning_rate": 0.0005,
|
| 370 |
+
"loss": 0.4474,
|
| 371 |
+
"step": 3500
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 14.0,
|
| 375 |
+
"eval_accuracy": 0.48996825396825394,
|
| 376 |
+
"eval_loss": 3.5998141765594482,
|
| 377 |
+
"eval_runtime": 6.0716,
|
| 378 |
+
"eval_samples_per_second": 82.351,
|
| 379 |
+
"eval_steps_per_second": 10.376,
|
| 380 |
+
"step": 3500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 14.4,
|
| 384 |
+
"grad_norm": 0.865963339805603,
|
| 385 |
+
"learning_rate": 0.0005,
|
| 386 |
+
"loss": 0.4037,
|
| 387 |
+
"step": 3600
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 14.8,
|
| 391 |
+
"grad_norm": 0.7994192242622375,
|
| 392 |
+
"learning_rate": 0.0005,
|
| 393 |
+
"loss": 0.4283,
|
| 394 |
+
"step": 3700
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 15.0,
|
| 398 |
+
"eval_accuracy": 0.48704761904761906,
|
| 399 |
+
"eval_loss": 3.6043944358825684,
|
| 400 |
+
"eval_runtime": 5.6932,
|
| 401 |
+
"eval_samples_per_second": 87.825,
|
| 402 |
+
"eval_steps_per_second": 11.066,
|
| 403 |
+
"step": 3750
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 15.2,
|
| 407 |
+
"grad_norm": 0.714190661907196,
|
| 408 |
+
"learning_rate": 0.0005,
|
| 409 |
+
"loss": 0.4183,
|
| 410 |
+
"step": 3800
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 15.6,
|
| 414 |
+
"grad_norm": 0.662158727645874,
|
| 415 |
+
"learning_rate": 0.0005,
|
| 416 |
+
"loss": 0.4085,
|
| 417 |
+
"step": 3900
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 16.0,
|
| 421 |
+
"grad_norm": 1.0598238706588745,
|
| 422 |
+
"learning_rate": 0.0005,
|
| 423 |
+
"loss": 0.4299,
|
| 424 |
+
"step": 4000
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 16.0,
|
| 428 |
+
"eval_accuracy": 0.4881904761904762,
|
| 429 |
+
"eval_loss": 3.6720120906829834,
|
| 430 |
+
"eval_runtime": 5.7732,
|
| 431 |
+
"eval_samples_per_second": 86.607,
|
| 432 |
+
"eval_steps_per_second": 10.912,
|
| 433 |
+
"step": 4000
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 16.4,
|
| 437 |
+
"grad_norm": 0.8209133744239807,
|
| 438 |
+
"learning_rate": 0.0005,
|
| 439 |
+
"loss": 0.3942,
|
| 440 |
+
"step": 4100
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 16.8,
|
| 444 |
+
"grad_norm": 0.9213945269584656,
|
| 445 |
+
"learning_rate": 0.0005,
|
| 446 |
+
"loss": 0.4202,
|
| 447 |
+
"step": 4200
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 17.0,
|
| 451 |
+
"eval_accuracy": 0.48603174603174604,
|
| 452 |
+
"eval_loss": 3.622042417526245,
|
| 453 |
+
"eval_runtime": 6.2522,
|
| 454 |
+
"eval_samples_per_second": 79.972,
|
| 455 |
+
"eval_steps_per_second": 10.076,
|
| 456 |
+
"step": 4250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 17.2,
|
| 460 |
+
"grad_norm": 0.7904919981956482,
|
| 461 |
+
"learning_rate": 0.0005,
|
| 462 |
+
"loss": 0.4163,
|
| 463 |
+
"step": 4300
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 17.6,
|
| 467 |
+
"grad_norm": 0.9449414014816284,
|
| 468 |
+
"learning_rate": 0.0005,
|
| 469 |
+
"loss": 0.4073,
|
| 470 |
+
"step": 4400
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 18.0,
|
| 474 |
+
"grad_norm": 0.8779242038726807,
|
| 475 |
+
"learning_rate": 0.0005,
|
| 476 |
+
"loss": 0.4318,
|
| 477 |
+
"step": 4500
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 18.0,
|
| 481 |
+
"eval_accuracy": 0.4875238095238095,
|
| 482 |
+
"eval_loss": 3.668213129043579,
|
| 483 |
+
"eval_runtime": 6.0372,
|
| 484 |
+
"eval_samples_per_second": 82.82,
|
| 485 |
+
"eval_steps_per_second": 10.435,
|
| 486 |
+
"step": 4500
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 18.4,
|
| 490 |
+
"grad_norm": 0.748566746711731,
|
| 491 |
+
"learning_rate": 0.0005,
|
| 492 |
+
"loss": 0.3935,
|
| 493 |
+
"step": 4600
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 18.8,
|
| 497 |
+
"grad_norm": 0.9708538055419922,
|
| 498 |
+
"learning_rate": 0.0005,
|
| 499 |
+
"loss": 0.4151,
|
| 500 |
+
"step": 4700
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 19.0,
|
| 504 |
+
"eval_accuracy": 0.4857142857142857,
|
| 505 |
+
"eval_loss": 3.710489273071289,
|
| 506 |
+
"eval_runtime": 5.9293,
|
| 507 |
+
"eval_samples_per_second": 84.326,
|
| 508 |
+
"eval_steps_per_second": 10.625,
|
| 509 |
+
"step": 4750
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 19.2,
|
| 513 |
+
"grad_norm": 0.7500284314155579,
|
| 514 |
+
"learning_rate": 0.0005,
|
| 515 |
+
"loss": 0.4043,
|
| 516 |
+
"step": 4800
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 19.6,
|
| 520 |
+
"grad_norm": 0.8508574366569519,
|
| 521 |
+
"learning_rate": 0.0005,
|
| 522 |
+
"loss": 0.4025,
|
| 523 |
+
"step": 4900
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 20.0,
|
| 527 |
+
"grad_norm": 0.9081528782844543,
|
| 528 |
+
"learning_rate": 0.0005,
|
| 529 |
+
"loss": 0.4227,
|
| 530 |
+
"step": 5000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 20.0,
|
| 534 |
+
"eval_accuracy": 0.48644444444444446,
|
| 535 |
+
"eval_loss": 3.7344048023223877,
|
| 536 |
+
"eval_runtime": 6.0765,
|
| 537 |
+
"eval_samples_per_second": 82.284,
|
| 538 |
+
"eval_steps_per_second": 10.368,
|
| 539 |
+
"step": 5000
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 20.0,
|
| 543 |
+
"step": 5000,
|
| 544 |
+
"total_flos": 1.293538587312128e+17,
|
| 545 |
+
"train_loss": 0.6887707939147949,
|
| 546 |
+
"train_runtime": 11763.1434,
|
| 547 |
+
"train_samples_per_second": 13.602,
|
| 548 |
+
"train_steps_per_second": 0.425
|
| 549 |
+
}
|
| 550 |
+
],
|
| 551 |
+
"logging_steps": 100,
|
| 552 |
+
"max_steps": 5000,
|
| 553 |
+
"num_input_tokens_seen": 0,
|
| 554 |
+
"num_train_epochs": 20,
|
| 555 |
+
"save_steps": 500,
|
| 556 |
+
"total_flos": 1.293538587312128e+17,
|
| 557 |
+
"train_batch_size": 1,
|
| 558 |
+
"trial_name": null,
|
| 559 |
+
"trial_params": null
|
| 560 |
+
}
|