Training in progress, step 4500
Browse files- model.safetensors +1 -1
- run-2/checkpoint-1000/model.safetensors +1 -1
- run-2/checkpoint-1000/optimizer.pt +1 -1
- run-2/checkpoint-1000/scheduler.pt +1 -1
- run-2/checkpoint-1000/trainer_state.json +25 -25
- run-2/checkpoint-1000/training_args.bin +1 -1
- run-2/checkpoint-1500/model.safetensors +1 -1
- run-2/checkpoint-1500/optimizer.pt +1 -1
- run-2/checkpoint-1500/scheduler.pt +1 -1
- run-2/checkpoint-1500/trainer_state.json +31 -31
- run-2/checkpoint-1500/training_args.bin +1 -1
- run-2/checkpoint-2000/model.safetensors +1 -1
- run-2/checkpoint-2000/optimizer.pt +1 -1
- run-2/checkpoint-2000/scheduler.pt +1 -1
- run-2/checkpoint-2000/trainer_state.json +45 -45
- run-2/checkpoint-2000/training_args.bin +1 -1
- run-2/checkpoint-2500/model.safetensors +1 -1
- run-2/checkpoint-2500/optimizer.pt +1 -1
- run-2/checkpoint-2500/scheduler.pt +1 -1
- run-2/checkpoint-2500/trainer_state.json +51 -51
- run-2/checkpoint-2500/training_args.bin +1 -1
- run-2/checkpoint-3000/model.safetensors +1 -1
- run-2/checkpoint-3000/optimizer.pt +1 -1
- run-2/checkpoint-3000/scheduler.pt +1 -1
- run-2/checkpoint-3000/trainer_state.json +65 -65
- run-2/checkpoint-3000/training_args.bin +1 -1
- run-2/checkpoint-3500/model.safetensors +1 -1
- run-2/checkpoint-3500/optimizer.pt +1 -1
- run-2/checkpoint-3500/scheduler.pt +1 -1
- run-2/checkpoint-3500/trainer_state.json +71 -71
- run-2/checkpoint-3500/training_args.bin +1 -1
- run-2/checkpoint-4000/model.safetensors +1 -1
- run-2/checkpoint-4000/optimizer.pt +1 -1
- run-2/checkpoint-4000/scheduler.pt +1 -1
- run-2/checkpoint-4000/trainer_state.json +85 -85
- run-2/checkpoint-4000/training_args.bin +1 -1
- run-2/checkpoint-4500/model.safetensors +1 -1
- run-2/checkpoint-4500/optimizer.pt +1 -1
- run-2/checkpoint-4500/scheduler.pt +1 -1
- run-2/checkpoint-4500/trainer_state.json +91 -91
- run-2/checkpoint-4500/training_args.bin +1 -1
- runs/Oct23_11-35-49_ec65ff5e83b6/events.out.tfevents.1761223404.ec65ff5e83b6.4546.3 +2 -2
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bd48eacc76940b767b8081cd817fd1c37797da71518ea157e83924dd4e29138
|
| 3 |
size 598898116
|
run-2/checkpoint-1000/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f28288ac778bb6f024ce4a73fcc5092a2413b9eeb7560420472810b120686f57
|
| 3 |
size 598898116
|
run-2/checkpoint-1000/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28b456041c32b3168b2f02bb4b47d0e388ca1212cf2326f20b01c76d5e280799
|
| 3 |
size 1197886411
|
run-2/checkpoint-1000/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fbbbb32195293dd885e75215e07e363d7c22cc73f1b1d0c7c508696c26691417
|
| 3 |
size 1465
|
run-2/checkpoint-1000/trainer_state.json
CHANGED
|
@@ -11,53 +11,53 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
}
|
| 56 |
],
|
| 57 |
"logging_steps": 200,
|
| 58 |
-
"max_steps":
|
| 59 |
"num_input_tokens_seen": 0,
|
| 60 |
-
"num_train_epochs":
|
| 61 |
"save_steps": 500,
|
| 62 |
"stateful_callbacks": {
|
| 63 |
"TrainerControl": {
|
|
@@ -75,8 +75,8 @@
|
|
| 75 |
"train_batch_size": 16,
|
| 76 |
"trial_name": null,
|
| 77 |
"trial_params": {
|
| 78 |
-
"alpha": 0.
|
| 79 |
-
"num_train_epochs":
|
| 80 |
-
"temperature":
|
| 81 |
}
|
| 82 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
}
|
| 56 |
],
|
| 57 |
"logging_steps": 200,
|
| 58 |
+
"max_steps": 4770,
|
| 59 |
"num_input_tokens_seen": 0,
|
| 60 |
+
"num_train_epochs": 5,
|
| 61 |
"save_steps": 500,
|
| 62 |
"stateful_callbacks": {
|
| 63 |
"TrainerControl": {
|
|
|
|
| 75 |
"train_batch_size": 16,
|
| 76 |
"trial_name": null,
|
| 77 |
"trial_params": {
|
| 78 |
+
"alpha": 0.7896637522578164,
|
| 79 |
+
"num_train_epochs": 5,
|
| 80 |
+
"temperature": 4
|
| 81 |
}
|
| 82 |
}
|
run-2/checkpoint-1000/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-1500/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1905fe41fb81fe4b5d38b9e652c318ffbd46e23986515a5e820ed291c6f988d0
|
| 3 |
size 598898116
|
run-2/checkpoint-1500/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e73e459986df9de94cad4322a98423855bfa5bfddae92b272623ac8d4e3f3e0
|
| 3 |
size 1197886411
|
run-2/checkpoint-1500/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:abdaff917f8d0d0fa51c87b92a586474c34f6688f66a58df19a2d3f2ab5de74c
|
| 3 |
size 1465
|
run-2/checkpoint-1500/trainer_state.json
CHANGED
|
@@ -11,67 +11,67 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
}
|
| 70 |
],
|
| 71 |
"logging_steps": 200,
|
| 72 |
-
"max_steps":
|
| 73 |
"num_input_tokens_seen": 0,
|
| 74 |
-
"num_train_epochs":
|
| 75 |
"save_steps": 500,
|
| 76 |
"stateful_callbacks": {
|
| 77 |
"TrainerControl": {
|
|
@@ -89,8 +89,8 @@
|
|
| 89 |
"train_batch_size": 16,
|
| 90 |
"trial_name": null,
|
| 91 |
"trial_params": {
|
| 92 |
-
"alpha": 0.
|
| 93 |
-
"num_train_epochs":
|
| 94 |
-
"temperature":
|
| 95 |
}
|
| 96 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
}
|
| 70 |
],
|
| 71 |
"logging_steps": 200,
|
| 72 |
+
"max_steps": 4770,
|
| 73 |
"num_input_tokens_seen": 0,
|
| 74 |
+
"num_train_epochs": 5,
|
| 75 |
"save_steps": 500,
|
| 76 |
"stateful_callbacks": {
|
| 77 |
"TrainerControl": {
|
|
|
|
| 89 |
"train_batch_size": 16,
|
| 90 |
"trial_name": null,
|
| 91 |
"trial_params": {
|
| 92 |
+
"alpha": 0.7896637522578164,
|
| 93 |
+
"num_train_epochs": 5,
|
| 94 |
+
"temperature": 4
|
| 95 |
}
|
| 96 |
}
|
run-2/checkpoint-1500/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-2000/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5089196f4a13dded45220594cc91ac3c2e254c5da9950d2fb32f1376046ee70f
|
| 3 |
size 598898116
|
run-2/checkpoint-2000/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b3b800a7fd50bb06d4ac7cc561fbd4b01e9318508644ceca01fec8f9d5b9a2c
|
| 3 |
size 1197886411
|
run-2/checkpoint-2000/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ad6299911574d785b66d71f02acef1c78edd67929fbb86d1e12dcb5b017c4eae
|
| 3 |
size 1465
|
run-2/checkpoint-2000/trainer_state.json
CHANGED
|
@@ -11,97 +11,97 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
-
"grad_norm": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"learning_rate":
|
| 81 |
-
"loss": 0.
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
-
"eval_accuracy": 0.
|
| 87 |
-
"eval_loss": 0.
|
| 88 |
-
"eval_runtime":
|
| 89 |
-
"eval_samples_per_second":
|
| 90 |
-
"eval_steps_per_second":
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
-
"grad_norm": 0.
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 2000
|
| 99 |
}
|
| 100 |
],
|
| 101 |
"logging_steps": 200,
|
| 102 |
-
"max_steps":
|
| 103 |
"num_input_tokens_seen": 0,
|
| 104 |
-
"num_train_epochs":
|
| 105 |
"save_steps": 500,
|
| 106 |
"stateful_callbacks": {
|
| 107 |
"TrainerControl": {
|
|
@@ -119,8 +119,8 @@
|
|
| 119 |
"train_batch_size": 16,
|
| 120 |
"trial_name": null,
|
| 121 |
"trial_params": {
|
| 122 |
-
"alpha": 0.
|
| 123 |
-
"num_train_epochs":
|
| 124 |
-
"temperature":
|
| 125 |
}
|
| 126 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
+
"grad_norm": 0.17968730628490448,
|
| 73 |
+
"learning_rate": 1.9943396226415094e-05,
|
| 74 |
+
"loss": 0.0487,
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
+
"grad_norm": 17.465627670288086,
|
| 80 |
+
"learning_rate": 1.868553459119497e-05,
|
| 81 |
+
"loss": 0.0556,
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
+
"eval_accuracy": 0.9496774193548387,
|
| 87 |
+
"eval_loss": 0.23183350265026093,
|
| 88 |
+
"eval_runtime": 6.1991,
|
| 89 |
+
"eval_samples_per_second": 500.07,
|
| 90 |
+
"eval_steps_per_second": 31.295,
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
+
"grad_norm": 0.020424585789442062,
|
| 96 |
+
"learning_rate": 1.742767295597484e-05,
|
| 97 |
+
"loss": 0.0409,
|
| 98 |
"step": 2000
|
| 99 |
}
|
| 100 |
],
|
| 101 |
"logging_steps": 200,
|
| 102 |
+
"max_steps": 4770,
|
| 103 |
"num_input_tokens_seen": 0,
|
| 104 |
+
"num_train_epochs": 5,
|
| 105 |
"save_steps": 500,
|
| 106 |
"stateful_callbacks": {
|
| 107 |
"TrainerControl": {
|
|
|
|
| 119 |
"train_batch_size": 16,
|
| 120 |
"trial_name": null,
|
| 121 |
"trial_params": {
|
| 122 |
+
"alpha": 0.7896637522578164,
|
| 123 |
+
"num_train_epochs": 5,
|
| 124 |
+
"temperature": 4
|
| 125 |
}
|
| 126 |
}
|
run-2/checkpoint-2000/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-2500/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d1d0bbc4c63e7a063b5ef81cdfded47d4f3ed44ed2c1ab894a511efd6be31ead
|
| 3 |
size 598898116
|
run-2/checkpoint-2500/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:250d40838bad8850a5df6cc228a0857f752d202f5661c1dac483641e146782ac
|
| 3 |
size 1197886411
|
run-2/checkpoint-2500/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d7b695bc3dfc76643ff10ec841e274d6113ed0c2a13104892b36fd692fc3b1cc
|
| 3 |
size 1465
|
run-2/checkpoint-2500/trainer_state.json
CHANGED
|
@@ -11,111 +11,111 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
-
"grad_norm": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"learning_rate":
|
| 81 |
-
"loss": 0.
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
-
"eval_accuracy": 0.
|
| 87 |
-
"eval_loss": 0.
|
| 88 |
-
"eval_runtime":
|
| 89 |
-
"eval_samples_per_second":
|
| 90 |
-
"eval_steps_per_second":
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
-
"grad_norm": 0.
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
-
"grad_norm": 0.
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
-
"grad_norm":
|
| 110 |
-
"learning_rate": 1.
|
| 111 |
-
"loss": 0.
|
| 112 |
"step": 2400
|
| 113 |
}
|
| 114 |
],
|
| 115 |
"logging_steps": 200,
|
| 116 |
-
"max_steps":
|
| 117 |
"num_input_tokens_seen": 0,
|
| 118 |
-
"num_train_epochs":
|
| 119 |
"save_steps": 500,
|
| 120 |
"stateful_callbacks": {
|
| 121 |
"TrainerControl": {
|
|
@@ -133,8 +133,8 @@
|
|
| 133 |
"train_batch_size": 16,
|
| 134 |
"trial_name": null,
|
| 135 |
"trial_params": {
|
| 136 |
-
"alpha": 0.
|
| 137 |
-
"num_train_epochs":
|
| 138 |
-
"temperature":
|
| 139 |
}
|
| 140 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
+
"grad_norm": 0.17968730628490448,
|
| 73 |
+
"learning_rate": 1.9943396226415094e-05,
|
| 74 |
+
"loss": 0.0487,
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
+
"grad_norm": 17.465627670288086,
|
| 80 |
+
"learning_rate": 1.868553459119497e-05,
|
| 81 |
+
"loss": 0.0556,
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
+
"eval_accuracy": 0.9496774193548387,
|
| 87 |
+
"eval_loss": 0.23183350265026093,
|
| 88 |
+
"eval_runtime": 6.1991,
|
| 89 |
+
"eval_samples_per_second": 500.07,
|
| 90 |
+
"eval_steps_per_second": 31.295,
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
+
"grad_norm": 0.020424585789442062,
|
| 96 |
+
"learning_rate": 1.742767295597484e-05,
|
| 97 |
+
"loss": 0.0409,
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
+
"grad_norm": 0.036295052617788315,
|
| 103 |
+
"learning_rate": 1.6169811320754717e-05,
|
| 104 |
+
"loss": 0.0135,
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
+
"grad_norm": 6.396214008331299,
|
| 110 |
+
"learning_rate": 1.491194968553459e-05,
|
| 111 |
+
"loss": 0.0199,
|
| 112 |
"step": 2400
|
| 113 |
}
|
| 114 |
],
|
| 115 |
"logging_steps": 200,
|
| 116 |
+
"max_steps": 4770,
|
| 117 |
"num_input_tokens_seen": 0,
|
| 118 |
+
"num_train_epochs": 5,
|
| 119 |
"save_steps": 500,
|
| 120 |
"stateful_callbacks": {
|
| 121 |
"TrainerControl": {
|
|
|
|
| 133 |
"train_batch_size": 16,
|
| 134 |
"trial_name": null,
|
| 135 |
"trial_params": {
|
| 136 |
+
"alpha": 0.7896637522578164,
|
| 137 |
+
"num_train_epochs": 5,
|
| 138 |
+
"temperature": 4
|
| 139 |
}
|
| 140 |
}
|
run-2/checkpoint-2500/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-3000/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6144b67e0c970718b1d25507b5a35543beaa7bd44ad4eaa02f23769e108f4d40
|
| 3 |
size 598898116
|
run-2/checkpoint-3000/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:24339245e57fbea92595807384396c41fe482046841a3d51a43b5c8d6a0daf05
|
| 3 |
size 1197886411
|
run-2/checkpoint-3000/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4fae9e5d4cea97d8b9da2a50e7f1db2a81314b175343651f02628e7aadce229e
|
| 3 |
size 1465
|
run-2/checkpoint-3000/trainer_state.json
CHANGED
|
@@ -11,141 +11,141 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
-
"grad_norm": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"learning_rate":
|
| 81 |
-
"loss": 0.
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
-
"eval_accuracy": 0.
|
| 87 |
-
"eval_loss": 0.
|
| 88 |
-
"eval_runtime":
|
| 89 |
-
"eval_samples_per_second":
|
| 90 |
-
"eval_steps_per_second":
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
-
"grad_norm": 0.
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
-
"grad_norm": 0.
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
-
"grad_norm":
|
| 110 |
-
"learning_rate": 1.
|
| 111 |
-
"loss": 0.
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
-
"grad_norm": 0.
|
| 117 |
-
"learning_rate": 1.
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
-
"grad_norm": 0.
|
| 124 |
-
"learning_rate": 1.
|
| 125 |
-
"loss": 0.
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
-
"eval_accuracy": 0.
|
| 131 |
-
"eval_loss": 0.
|
| 132 |
-
"eval_runtime":
|
| 133 |
-
"eval_samples_per_second":
|
| 134 |
-
"eval_steps_per_second":
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
-
"grad_norm": 0.
|
| 140 |
-
"learning_rate": 1.
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 3000
|
| 143 |
}
|
| 144 |
],
|
| 145 |
"logging_steps": 200,
|
| 146 |
-
"max_steps":
|
| 147 |
"num_input_tokens_seen": 0,
|
| 148 |
-
"num_train_epochs":
|
| 149 |
"save_steps": 500,
|
| 150 |
"stateful_callbacks": {
|
| 151 |
"TrainerControl": {
|
|
@@ -163,8 +163,8 @@
|
|
| 163 |
"train_batch_size": 16,
|
| 164 |
"trial_name": null,
|
| 165 |
"trial_params": {
|
| 166 |
-
"alpha": 0.
|
| 167 |
-
"num_train_epochs":
|
| 168 |
-
"temperature":
|
| 169 |
}
|
| 170 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
+
"grad_norm": 0.17968730628490448,
|
| 73 |
+
"learning_rate": 1.9943396226415094e-05,
|
| 74 |
+
"loss": 0.0487,
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
+
"grad_norm": 17.465627670288086,
|
| 80 |
+
"learning_rate": 1.868553459119497e-05,
|
| 81 |
+
"loss": 0.0556,
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
+
"eval_accuracy": 0.9496774193548387,
|
| 87 |
+
"eval_loss": 0.23183350265026093,
|
| 88 |
+
"eval_runtime": 6.1991,
|
| 89 |
+
"eval_samples_per_second": 500.07,
|
| 90 |
+
"eval_steps_per_second": 31.295,
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
+
"grad_norm": 0.020424585789442062,
|
| 96 |
+
"learning_rate": 1.742767295597484e-05,
|
| 97 |
+
"loss": 0.0409,
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
+
"grad_norm": 0.036295052617788315,
|
| 103 |
+
"learning_rate": 1.6169811320754717e-05,
|
| 104 |
+
"loss": 0.0135,
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
+
"grad_norm": 6.396214008331299,
|
| 110 |
+
"learning_rate": 1.491194968553459e-05,
|
| 111 |
+
"loss": 0.0199,
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
+
"grad_norm": 0.07038699835538864,
|
| 117 |
+
"learning_rate": 1.3654088050314464e-05,
|
| 118 |
+
"loss": 0.0084,
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
+
"grad_norm": 0.09982823580503464,
|
| 124 |
+
"learning_rate": 1.239622641509434e-05,
|
| 125 |
+
"loss": 0.008,
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
+
"eval_accuracy": 0.9609677419354838,
|
| 131 |
+
"eval_loss": 0.19234703481197357,
|
| 132 |
+
"eval_runtime": 8.1342,
|
| 133 |
+
"eval_samples_per_second": 381.109,
|
| 134 |
+
"eval_steps_per_second": 23.85,
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
+
"grad_norm": 0.007687446661293507,
|
| 140 |
+
"learning_rate": 1.1138364779874214e-05,
|
| 141 |
+
"loss": 0.0041,
|
| 142 |
"step": 3000
|
| 143 |
}
|
| 144 |
],
|
| 145 |
"logging_steps": 200,
|
| 146 |
+
"max_steps": 4770,
|
| 147 |
"num_input_tokens_seen": 0,
|
| 148 |
+
"num_train_epochs": 5,
|
| 149 |
"save_steps": 500,
|
| 150 |
"stateful_callbacks": {
|
| 151 |
"TrainerControl": {
|
|
|
|
| 163 |
"train_batch_size": 16,
|
| 164 |
"trial_name": null,
|
| 165 |
"trial_params": {
|
| 166 |
+
"alpha": 0.7896637522578164,
|
| 167 |
+
"num_train_epochs": 5,
|
| 168 |
+
"temperature": 4
|
| 169 |
}
|
| 170 |
}
|
run-2/checkpoint-3000/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-3500/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2687be008f3100c5d4839c8c99c609f93d017b4216060a8dc8b46d2ab36c5011
|
| 3 |
size 598898116
|
run-2/checkpoint-3500/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:82f171274b6618bef8a7a4e12bfdc1d5f37e19a48112ee1d8bcf1f49ea8f0c74
|
| 3 |
size 1197886411
|
run-2/checkpoint-3500/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fe421286099f53ea00a29a2d1d78d57e96dcfee55d59a8d0b496851ddc6108f0
|
| 3 |
size 1465
|
run-2/checkpoint-3500/trainer_state.json
CHANGED
|
@@ -11,155 +11,155 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
-
"grad_norm": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"learning_rate":
|
| 81 |
-
"loss": 0.
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
-
"eval_accuracy": 0.
|
| 87 |
-
"eval_loss": 0.
|
| 88 |
-
"eval_runtime":
|
| 89 |
-
"eval_samples_per_second":
|
| 90 |
-
"eval_steps_per_second":
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
-
"grad_norm": 0.
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
-
"grad_norm": 0.
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
-
"grad_norm":
|
| 110 |
-
"learning_rate": 1.
|
| 111 |
-
"loss": 0.
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
-
"grad_norm": 0.
|
| 117 |
-
"learning_rate": 1.
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
-
"grad_norm": 0.
|
| 124 |
-
"learning_rate": 1.
|
| 125 |
-
"loss": 0.
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
-
"eval_accuracy": 0.
|
| 131 |
-
"eval_loss": 0.
|
| 132 |
-
"eval_runtime":
|
| 133 |
-
"eval_samples_per_second":
|
| 134 |
-
"eval_steps_per_second":
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
-
"grad_norm": 0.
|
| 140 |
-
"learning_rate": 1.
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 3000
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 3.3542976939203353,
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 3200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 3.5639412997903563,
|
| 153 |
-
"grad_norm": 0.
|
| 154 |
-
"learning_rate":
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 3400
|
| 157 |
}
|
| 158 |
],
|
| 159 |
"logging_steps": 200,
|
| 160 |
-
"max_steps":
|
| 161 |
"num_input_tokens_seen": 0,
|
| 162 |
-
"num_train_epochs":
|
| 163 |
"save_steps": 500,
|
| 164 |
"stateful_callbacks": {
|
| 165 |
"TrainerControl": {
|
|
@@ -177,8 +177,8 @@
|
|
| 177 |
"train_batch_size": 16,
|
| 178 |
"trial_name": null,
|
| 179 |
"trial_params": {
|
| 180 |
-
"alpha": 0.
|
| 181 |
-
"num_train_epochs":
|
| 182 |
-
"temperature":
|
| 183 |
}
|
| 184 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
+
"grad_norm": 0.17968730628490448,
|
| 73 |
+
"learning_rate": 1.9943396226415094e-05,
|
| 74 |
+
"loss": 0.0487,
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
+
"grad_norm": 17.465627670288086,
|
| 80 |
+
"learning_rate": 1.868553459119497e-05,
|
| 81 |
+
"loss": 0.0556,
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
+
"eval_accuracy": 0.9496774193548387,
|
| 87 |
+
"eval_loss": 0.23183350265026093,
|
| 88 |
+
"eval_runtime": 6.1991,
|
| 89 |
+
"eval_samples_per_second": 500.07,
|
| 90 |
+
"eval_steps_per_second": 31.295,
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
+
"grad_norm": 0.020424585789442062,
|
| 96 |
+
"learning_rate": 1.742767295597484e-05,
|
| 97 |
+
"loss": 0.0409,
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
+
"grad_norm": 0.036295052617788315,
|
| 103 |
+
"learning_rate": 1.6169811320754717e-05,
|
| 104 |
+
"loss": 0.0135,
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
+
"grad_norm": 6.396214008331299,
|
| 110 |
+
"learning_rate": 1.491194968553459e-05,
|
| 111 |
+
"loss": 0.0199,
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
+
"grad_norm": 0.07038699835538864,
|
| 117 |
+
"learning_rate": 1.3654088050314464e-05,
|
| 118 |
+
"loss": 0.0084,
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
+
"grad_norm": 0.09982823580503464,
|
| 124 |
+
"learning_rate": 1.239622641509434e-05,
|
| 125 |
+
"loss": 0.008,
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
+
"eval_accuracy": 0.9609677419354838,
|
| 131 |
+
"eval_loss": 0.19234703481197357,
|
| 132 |
+
"eval_runtime": 8.1342,
|
| 133 |
+
"eval_samples_per_second": 381.109,
|
| 134 |
+
"eval_steps_per_second": 23.85,
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
+
"grad_norm": 0.007687446661293507,
|
| 140 |
+
"learning_rate": 1.1138364779874214e-05,
|
| 141 |
+
"loss": 0.0041,
|
| 142 |
"step": 3000
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 3.3542976939203353,
|
| 146 |
+
"grad_norm": 2.6632392406463623,
|
| 147 |
+
"learning_rate": 9.880503144654089e-06,
|
| 148 |
+
"loss": 0.0055,
|
| 149 |
"step": 3200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 3.5639412997903563,
|
| 153 |
+
"grad_norm": 0.014160693623125553,
|
| 154 |
+
"learning_rate": 8.622641509433963e-06,
|
| 155 |
+
"loss": 0.002,
|
| 156 |
"step": 3400
|
| 157 |
}
|
| 158 |
],
|
| 159 |
"logging_steps": 200,
|
| 160 |
+
"max_steps": 4770,
|
| 161 |
"num_input_tokens_seen": 0,
|
| 162 |
+
"num_train_epochs": 5,
|
| 163 |
"save_steps": 500,
|
| 164 |
"stateful_callbacks": {
|
| 165 |
"TrainerControl": {
|
|
|
|
| 177 |
"train_batch_size": 16,
|
| 178 |
"trial_name": null,
|
| 179 |
"trial_params": {
|
| 180 |
+
"alpha": 0.7896637522578164,
|
| 181 |
+
"num_train_epochs": 5,
|
| 182 |
+
"temperature": 4
|
| 183 |
}
|
| 184 |
}
|
run-2/checkpoint-3500/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-4000/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e85cb2426b5b984e4089fb5675451860f62cc1aaeef32ac93b49caa68908ba4
|
| 3 |
size 598898116
|
run-2/checkpoint-4000/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cffde57c8ec2e5043b3899eb91a9379813dc4ad6ea80c57fb3555a093f7453a
|
| 3 |
size 1197886411
|
run-2/checkpoint-4000/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b8ebfbc57a5ee18fb08554d9eccaa7709c1407d3c3ebd69b471935f08a8b07fe
|
| 3 |
size 1465
|
run-2/checkpoint-4000/trainer_state.json
CHANGED
|
@@ -11,185 +11,185 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
-
"grad_norm": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"learning_rate":
|
| 81 |
-
"loss": 0.
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
-
"eval_accuracy": 0.
|
| 87 |
-
"eval_loss": 0.
|
| 88 |
-
"eval_runtime":
|
| 89 |
-
"eval_samples_per_second":
|
| 90 |
-
"eval_steps_per_second":
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
-
"grad_norm": 0.
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
-
"grad_norm": 0.
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
-
"grad_norm":
|
| 110 |
-
"learning_rate": 1.
|
| 111 |
-
"loss": 0.
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
-
"grad_norm": 0.
|
| 117 |
-
"learning_rate": 1.
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
-
"grad_norm": 0.
|
| 124 |
-
"learning_rate": 1.
|
| 125 |
-
"loss": 0.
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
-
"eval_accuracy": 0.
|
| 131 |
-
"eval_loss": 0.
|
| 132 |
-
"eval_runtime":
|
| 133 |
-
"eval_samples_per_second":
|
| 134 |
-
"eval_steps_per_second":
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
-
"grad_norm": 0.
|
| 140 |
-
"learning_rate": 1.
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 3000
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 3.3542976939203353,
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 3200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 3.5639412997903563,
|
| 153 |
-
"grad_norm": 0.
|
| 154 |
-
"learning_rate":
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 3400
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 3.7735849056603774,
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate":
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 3600
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 3.9832285115303985,
|
| 167 |
-
"grad_norm": 0.
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 3800
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 4.0,
|
| 174 |
-
"eval_accuracy": 0.
|
| 175 |
-
"eval_loss": 0.
|
| 176 |
-
"eval_runtime": 6.
|
| 177 |
-
"eval_samples_per_second":
|
| 178 |
-
"eval_steps_per_second":
|
| 179 |
"step": 3816
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 4.1928721174004195,
|
| 183 |
-
"grad_norm": 0.
|
| 184 |
-
"learning_rate":
|
| 185 |
-
"loss": 0.
|
| 186 |
"step": 4000
|
| 187 |
}
|
| 188 |
],
|
| 189 |
"logging_steps": 200,
|
| 190 |
-
"max_steps":
|
| 191 |
"num_input_tokens_seen": 0,
|
| 192 |
-
"num_train_epochs":
|
| 193 |
"save_steps": 500,
|
| 194 |
"stateful_callbacks": {
|
| 195 |
"TrainerControl": {
|
|
@@ -207,8 +207,8 @@
|
|
| 207 |
"train_batch_size": 16,
|
| 208 |
"trial_name": null,
|
| 209 |
"trial_params": {
|
| 210 |
-
"alpha": 0.
|
| 211 |
-
"num_train_epochs":
|
| 212 |
-
"temperature":
|
| 213 |
}
|
| 214 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
+
"grad_norm": 0.17968730628490448,
|
| 73 |
+
"learning_rate": 1.9943396226415094e-05,
|
| 74 |
+
"loss": 0.0487,
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
+
"grad_norm": 17.465627670288086,
|
| 80 |
+
"learning_rate": 1.868553459119497e-05,
|
| 81 |
+
"loss": 0.0556,
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
+
"eval_accuracy": 0.9496774193548387,
|
| 87 |
+
"eval_loss": 0.23183350265026093,
|
| 88 |
+
"eval_runtime": 6.1991,
|
| 89 |
+
"eval_samples_per_second": 500.07,
|
| 90 |
+
"eval_steps_per_second": 31.295,
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
+
"grad_norm": 0.020424585789442062,
|
| 96 |
+
"learning_rate": 1.742767295597484e-05,
|
| 97 |
+
"loss": 0.0409,
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
+
"grad_norm": 0.036295052617788315,
|
| 103 |
+
"learning_rate": 1.6169811320754717e-05,
|
| 104 |
+
"loss": 0.0135,
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
+
"grad_norm": 6.396214008331299,
|
| 110 |
+
"learning_rate": 1.491194968553459e-05,
|
| 111 |
+
"loss": 0.0199,
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
+
"grad_norm": 0.07038699835538864,
|
| 117 |
+
"learning_rate": 1.3654088050314464e-05,
|
| 118 |
+
"loss": 0.0084,
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
+
"grad_norm": 0.09982823580503464,
|
| 124 |
+
"learning_rate": 1.239622641509434e-05,
|
| 125 |
+
"loss": 0.008,
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
+
"eval_accuracy": 0.9609677419354838,
|
| 131 |
+
"eval_loss": 0.19234703481197357,
|
| 132 |
+
"eval_runtime": 8.1342,
|
| 133 |
+
"eval_samples_per_second": 381.109,
|
| 134 |
+
"eval_steps_per_second": 23.85,
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
+
"grad_norm": 0.007687446661293507,
|
| 140 |
+
"learning_rate": 1.1138364779874214e-05,
|
| 141 |
+
"loss": 0.0041,
|
| 142 |
"step": 3000
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 3.3542976939203353,
|
| 146 |
+
"grad_norm": 2.6632392406463623,
|
| 147 |
+
"learning_rate": 9.880503144654089e-06,
|
| 148 |
+
"loss": 0.0055,
|
| 149 |
"step": 3200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 3.5639412997903563,
|
| 153 |
+
"grad_norm": 0.014160693623125553,
|
| 154 |
+
"learning_rate": 8.622641509433963e-06,
|
| 155 |
+
"loss": 0.002,
|
| 156 |
"step": 3400
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 3.7735849056603774,
|
| 160 |
+
"grad_norm": 0.059565939009189606,
|
| 161 |
+
"learning_rate": 7.364779874213837e-06,
|
| 162 |
+
"loss": 0.0017,
|
| 163 |
"step": 3600
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 3.9832285115303985,
|
| 167 |
+
"grad_norm": 0.0034850463271141052,
|
| 168 |
+
"learning_rate": 6.1069182389937105e-06,
|
| 169 |
+
"loss": 0.0075,
|
| 170 |
"step": 3800
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 4.0,
|
| 174 |
+
"eval_accuracy": 0.96,
|
| 175 |
+
"eval_loss": 0.190092071890831,
|
| 176 |
+
"eval_runtime": 6.4445,
|
| 177 |
+
"eval_samples_per_second": 481.034,
|
| 178 |
+
"eval_steps_per_second": 30.103,
|
| 179 |
"step": 3816
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 4.1928721174004195,
|
| 183 |
+
"grad_norm": 0.012302987277507782,
|
| 184 |
+
"learning_rate": 4.849056603773585e-06,
|
| 185 |
+
"loss": 0.0031,
|
| 186 |
"step": 4000
|
| 187 |
}
|
| 188 |
],
|
| 189 |
"logging_steps": 200,
|
| 190 |
+
"max_steps": 4770,
|
| 191 |
"num_input_tokens_seen": 0,
|
| 192 |
+
"num_train_epochs": 5,
|
| 193 |
"save_steps": 500,
|
| 194 |
"stateful_callbacks": {
|
| 195 |
"TrainerControl": {
|
|
|
|
| 207 |
"train_batch_size": 16,
|
| 208 |
"trial_name": null,
|
| 209 |
"trial_params": {
|
| 210 |
+
"alpha": 0.7896637522578164,
|
| 211 |
+
"num_train_epochs": 5,
|
| 212 |
+
"temperature": 4
|
| 213 |
}
|
| 214 |
}
|
run-2/checkpoint-4000/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
run-2/checkpoint-4500/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 598898116
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bd48eacc76940b767b8081cd817fd1c37797da71518ea157e83924dd4e29138
|
| 3 |
size 598898116
|
run-2/checkpoint-4500/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1197886411
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dae27d802154d955bcbee1b027965678c6ef789f5b7390766c5c16ea9999afdd
|
| 3 |
size 1197886411
|
run-2/checkpoint-4500/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2d29df8255baf50ad67160d39d36cfb8aa791577efd7c6a3108d4d01a125b92e
|
| 3 |
size 1465
|
run-2/checkpoint-4500/trainer_state.json
CHANGED
|
@@ -11,199 +11,199 @@
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
-
"grad_norm": 19.
|
| 15 |
-
"learning_rate": 2.
|
| 16 |
-
"loss": 3.
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
-
"grad_norm":
|
| 22 |
-
"learning_rate": 2.
|
| 23 |
-
"loss": 0.
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
-
"grad_norm": 7.
|
| 29 |
-
"learning_rate": 2.
|
| 30 |
-
"loss": 0.
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
-
"grad_norm": 11.
|
| 36 |
-
"learning_rate": 2.
|
| 37 |
-
"loss": 0.
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
-
"eval_accuracy": 0.
|
| 43 |
-
"eval_loss": 0.
|
| 44 |
-
"eval_runtime": 6.
|
| 45 |
-
"eval_samples_per_second":
|
| 46 |
-
"eval_steps_per_second":
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
-
"grad_norm":
|
| 52 |
-
"learning_rate": 2.
|
| 53 |
-
"loss": 0.
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
-
"grad_norm":
|
| 59 |
-
"learning_rate": 2.
|
| 60 |
-
"loss": 0.
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
-
"grad_norm": 0.
|
| 66 |
-
"learning_rate": 2.
|
| 67 |
-
"loss": 0.
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
-
"grad_norm": 0.
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss": 0.
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
-
"grad_norm":
|
| 80 |
-
"learning_rate":
|
| 81 |
-
"loss": 0.
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
-
"eval_accuracy": 0.
|
| 87 |
-
"eval_loss": 0.
|
| 88 |
-
"eval_runtime":
|
| 89 |
-
"eval_samples_per_second":
|
| 90 |
-
"eval_steps_per_second":
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
-
"grad_norm": 0.
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
-
"grad_norm": 0.
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
-
"grad_norm":
|
| 110 |
-
"learning_rate": 1.
|
| 111 |
-
"loss": 0.
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
-
"grad_norm": 0.
|
| 117 |
-
"learning_rate": 1.
|
| 118 |
-
"loss": 0.
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
-
"grad_norm": 0.
|
| 124 |
-
"learning_rate": 1.
|
| 125 |
-
"loss": 0.
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
-
"eval_accuracy": 0.
|
| 131 |
-
"eval_loss": 0.
|
| 132 |
-
"eval_runtime":
|
| 133 |
-
"eval_samples_per_second":
|
| 134 |
-
"eval_steps_per_second":
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
-
"grad_norm": 0.
|
| 140 |
-
"learning_rate": 1.
|
| 141 |
-
"loss": 0.
|
| 142 |
"step": 3000
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 3.3542976939203353,
|
| 146 |
-
"grad_norm":
|
| 147 |
-
"learning_rate":
|
| 148 |
-
"loss": 0.
|
| 149 |
"step": 3200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 3.5639412997903563,
|
| 153 |
-
"grad_norm": 0.
|
| 154 |
-
"learning_rate":
|
| 155 |
-
"loss": 0.
|
| 156 |
"step": 3400
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 3.7735849056603774,
|
| 160 |
-
"grad_norm":
|
| 161 |
-
"learning_rate":
|
| 162 |
-
"loss": 0.
|
| 163 |
"step": 3600
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 3.9832285115303985,
|
| 167 |
-
"grad_norm": 0.
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 3800
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 4.0,
|
| 174 |
-
"eval_accuracy": 0.
|
| 175 |
-
"eval_loss": 0.
|
| 176 |
-
"eval_runtime": 6.
|
| 177 |
-
"eval_samples_per_second":
|
| 178 |
-
"eval_steps_per_second":
|
| 179 |
"step": 3816
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 4.1928721174004195,
|
| 183 |
-
"grad_norm": 0.
|
| 184 |
-
"learning_rate":
|
| 185 |
-
"loss": 0.
|
| 186 |
"step": 4000
|
| 187 |
},
|
| 188 |
{
|
| 189 |
"epoch": 4.40251572327044,
|
| 190 |
-
"grad_norm": 0.
|
| 191 |
-
"learning_rate":
|
| 192 |
-
"loss": 0.
|
| 193 |
"step": 4200
|
| 194 |
},
|
| 195 |
{
|
| 196 |
"epoch": 4.612159329140461,
|
| 197 |
-
"grad_norm": 0.
|
| 198 |
-
"learning_rate":
|
| 199 |
-
"loss": 0.
|
| 200 |
"step": 4400
|
| 201 |
}
|
| 202 |
],
|
| 203 |
"logging_steps": 200,
|
| 204 |
-
"max_steps":
|
| 205 |
"num_input_tokens_seen": 0,
|
| 206 |
-
"num_train_epochs":
|
| 207 |
"save_steps": 500,
|
| 208 |
"stateful_callbacks": {
|
| 209 |
"TrainerControl": {
|
|
@@ -221,8 +221,8 @@
|
|
| 221 |
"train_batch_size": 16,
|
| 222 |
"trial_name": null,
|
| 223 |
"trial_params": {
|
| 224 |
-
"alpha": 0.
|
| 225 |
-
"num_train_epochs":
|
| 226 |
-
"temperature":
|
| 227 |
}
|
| 228 |
}
|
|
|
|
| 11 |
"log_history": [
|
| 12 |
{
|
| 13 |
"epoch": 0.20964360587002095,
|
| 14 |
+
"grad_norm": 19.058637619018555,
|
| 15 |
+
"learning_rate": 2.8748427672955975e-05,
|
| 16 |
+
"loss": 3.1779,
|
| 17 |
"step": 200
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"epoch": 0.4192872117400419,
|
| 21 |
+
"grad_norm": 16.188446044921875,
|
| 22 |
+
"learning_rate": 2.749056603773585e-05,
|
| 23 |
+
"loss": 0.6362,
|
| 24 |
"step": 400
|
| 25 |
},
|
| 26 |
{
|
| 27 |
"epoch": 0.6289308176100629,
|
| 28 |
+
"grad_norm": 7.167396545410156,
|
| 29 |
+
"learning_rate": 2.6232704402515723e-05,
|
| 30 |
+
"loss": 0.3306,
|
| 31 |
"step": 600
|
| 32 |
},
|
| 33 |
{
|
| 34 |
"epoch": 0.8385744234800838,
|
| 35 |
+
"grad_norm": 11.347878456115723,
|
| 36 |
+
"learning_rate": 2.49748427672956e-05,
|
| 37 |
+
"loss": 0.2273,
|
| 38 |
"step": 800
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 1.0,
|
| 42 |
+
"eval_accuracy": 0.9325806451612904,
|
| 43 |
+
"eval_loss": 0.2720068097114563,
|
| 44 |
+
"eval_runtime": 6.3485,
|
| 45 |
+
"eval_samples_per_second": 488.302,
|
| 46 |
+
"eval_steps_per_second": 30.558,
|
| 47 |
"step": 954
|
| 48 |
},
|
| 49 |
{
|
| 50 |
"epoch": 1.0482180293501049,
|
| 51 |
+
"grad_norm": 15.08842658996582,
|
| 52 |
+
"learning_rate": 2.3716981132075474e-05,
|
| 53 |
+
"loss": 0.1632,
|
| 54 |
"step": 1000
|
| 55 |
},
|
| 56 |
{
|
| 57 |
"epoch": 1.2578616352201257,
|
| 58 |
+
"grad_norm": 1.7787141799926758,
|
| 59 |
+
"learning_rate": 2.2459119496855346e-05,
|
| 60 |
+
"loss": 0.0698,
|
| 61 |
"step": 1200
|
| 62 |
},
|
| 63 |
{
|
| 64 |
"epoch": 1.4675052410901468,
|
| 65 |
+
"grad_norm": 0.3887692987918854,
|
| 66 |
+
"learning_rate": 2.120125786163522e-05,
|
| 67 |
+
"loss": 0.0551,
|
| 68 |
"step": 1400
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"epoch": 1.6771488469601676,
|
| 72 |
+
"grad_norm": 0.17968730628490448,
|
| 73 |
+
"learning_rate": 1.9943396226415094e-05,
|
| 74 |
+
"loss": 0.0487,
|
| 75 |
"step": 1600
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 1.8867924528301887,
|
| 79 |
+
"grad_norm": 17.465627670288086,
|
| 80 |
+
"learning_rate": 1.868553459119497e-05,
|
| 81 |
+
"loss": 0.0556,
|
| 82 |
"step": 1800
|
| 83 |
},
|
| 84 |
{
|
| 85 |
"epoch": 2.0,
|
| 86 |
+
"eval_accuracy": 0.9496774193548387,
|
| 87 |
+
"eval_loss": 0.23183350265026093,
|
| 88 |
+
"eval_runtime": 6.1991,
|
| 89 |
+
"eval_samples_per_second": 500.07,
|
| 90 |
+
"eval_steps_per_second": 31.295,
|
| 91 |
"step": 1908
|
| 92 |
},
|
| 93 |
{
|
| 94 |
"epoch": 2.0964360587002098,
|
| 95 |
+
"grad_norm": 0.020424585789442062,
|
| 96 |
+
"learning_rate": 1.742767295597484e-05,
|
| 97 |
+
"loss": 0.0409,
|
| 98 |
"step": 2000
|
| 99 |
},
|
| 100 |
{
|
| 101 |
"epoch": 2.3060796645702304,
|
| 102 |
+
"grad_norm": 0.036295052617788315,
|
| 103 |
+
"learning_rate": 1.6169811320754717e-05,
|
| 104 |
+
"loss": 0.0135,
|
| 105 |
"step": 2200
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 2.5157232704402515,
|
| 109 |
+
"grad_norm": 6.396214008331299,
|
| 110 |
+
"learning_rate": 1.491194968553459e-05,
|
| 111 |
+
"loss": 0.0199,
|
| 112 |
"step": 2400
|
| 113 |
},
|
| 114 |
{
|
| 115 |
"epoch": 2.7253668763102725,
|
| 116 |
+
"grad_norm": 0.07038699835538864,
|
| 117 |
+
"learning_rate": 1.3654088050314464e-05,
|
| 118 |
+
"loss": 0.0084,
|
| 119 |
"step": 2600
|
| 120 |
},
|
| 121 |
{
|
| 122 |
"epoch": 2.9350104821802936,
|
| 123 |
+
"grad_norm": 0.09982823580503464,
|
| 124 |
+
"learning_rate": 1.239622641509434e-05,
|
| 125 |
+
"loss": 0.008,
|
| 126 |
"step": 2800
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"epoch": 3.0,
|
| 130 |
+
"eval_accuracy": 0.9609677419354838,
|
| 131 |
+
"eval_loss": 0.19234703481197357,
|
| 132 |
+
"eval_runtime": 8.1342,
|
| 133 |
+
"eval_samples_per_second": 381.109,
|
| 134 |
+
"eval_steps_per_second": 23.85,
|
| 135 |
"step": 2862
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 3.1446540880503147,
|
| 139 |
+
"grad_norm": 0.007687446661293507,
|
| 140 |
+
"learning_rate": 1.1138364779874214e-05,
|
| 141 |
+
"loss": 0.0041,
|
| 142 |
"step": 3000
|
| 143 |
},
|
| 144 |
{
|
| 145 |
"epoch": 3.3542976939203353,
|
| 146 |
+
"grad_norm": 2.6632392406463623,
|
| 147 |
+
"learning_rate": 9.880503144654089e-06,
|
| 148 |
+
"loss": 0.0055,
|
| 149 |
"step": 3200
|
| 150 |
},
|
| 151 |
{
|
| 152 |
"epoch": 3.5639412997903563,
|
| 153 |
+
"grad_norm": 0.014160693623125553,
|
| 154 |
+
"learning_rate": 8.622641509433963e-06,
|
| 155 |
+
"loss": 0.002,
|
| 156 |
"step": 3400
|
| 157 |
},
|
| 158 |
{
|
| 159 |
"epoch": 3.7735849056603774,
|
| 160 |
+
"grad_norm": 0.059565939009189606,
|
| 161 |
+
"learning_rate": 7.364779874213837e-06,
|
| 162 |
+
"loss": 0.0017,
|
| 163 |
"step": 3600
|
| 164 |
},
|
| 165 |
{
|
| 166 |
"epoch": 3.9832285115303985,
|
| 167 |
+
"grad_norm": 0.0034850463271141052,
|
| 168 |
+
"learning_rate": 6.1069182389937105e-06,
|
| 169 |
+
"loss": 0.0075,
|
| 170 |
"step": 3800
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 4.0,
|
| 174 |
+
"eval_accuracy": 0.96,
|
| 175 |
+
"eval_loss": 0.190092071890831,
|
| 176 |
+
"eval_runtime": 6.4445,
|
| 177 |
+
"eval_samples_per_second": 481.034,
|
| 178 |
+
"eval_steps_per_second": 30.103,
|
| 179 |
"step": 3816
|
| 180 |
},
|
| 181 |
{
|
| 182 |
"epoch": 4.1928721174004195,
|
| 183 |
+
"grad_norm": 0.012302987277507782,
|
| 184 |
+
"learning_rate": 4.849056603773585e-06,
|
| 185 |
+
"loss": 0.0031,
|
| 186 |
"step": 4000
|
| 187 |
},
|
| 188 |
{
|
| 189 |
"epoch": 4.40251572327044,
|
| 190 |
+
"grad_norm": 0.00425491388887167,
|
| 191 |
+
"learning_rate": 3.591194968553459e-06,
|
| 192 |
+
"loss": 0.0019,
|
| 193 |
"step": 4200
|
| 194 |
},
|
| 195 |
{
|
| 196 |
"epoch": 4.612159329140461,
|
| 197 |
+
"grad_norm": 0.012302517890930176,
|
| 198 |
+
"learning_rate": 2.3333333333333336e-06,
|
| 199 |
+
"loss": 0.0032,
|
| 200 |
"step": 4400
|
| 201 |
}
|
| 202 |
],
|
| 203 |
"logging_steps": 200,
|
| 204 |
+
"max_steps": 4770,
|
| 205 |
"num_input_tokens_seen": 0,
|
| 206 |
+
"num_train_epochs": 5,
|
| 207 |
"save_steps": 500,
|
| 208 |
"stateful_callbacks": {
|
| 209 |
"TrainerControl": {
|
|
|
|
| 221 |
"train_batch_size": 16,
|
| 222 |
"trial_name": null,
|
| 223 |
"trial_params": {
|
| 224 |
+
"alpha": 0.7896637522578164,
|
| 225 |
+
"num_train_epochs": 5,
|
| 226 |
+
"temperature": 4
|
| 227 |
}
|
| 228 |
}
|
run-2/checkpoint-4500/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5905
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:210679227dd2b3037b48d4414e3a7d05adcbca6f5fde35131afb41f4714ec72e
|
| 3 |
size 5905
|
runs/Oct23_11-35-49_ec65ff5e83b6/events.out.tfevents.1761223404.ec65ff5e83b6.4546.3
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:103cd3a099da6cb61ba71307470855a8f7c8ad777f367a272dd9e3bc7c1d9c44
|
| 3 |
+
size 18997
|