Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +36 -6
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 369133600
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:21a26dda1a848821358083c7732e4535140a06456808e219ed564f84b137c404
|
| 3 |
size 369133600
|
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 738413771
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9debcfec4e33c7eca37f273a805c541fda79f76eacb2c8990bede1c43a1e2e19
|
| 3 |
size 738413771
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c99e9eb1649a644c8be6e8e889139d6797bf8dd316223ef128559cf7c1e450b6
|
| 3 |
size 14645
|
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 sha256:87ce6f951be02f274d9d382b245e4b161a33bf9a5e8133937fef0aa6cd190d69
|
| 3 |
size 1465
|
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 0.
|
| 4 |
-
"best_model_checkpoint": "/content/drive/MyDrive/lora_model/outputs/task15_microsoft/Phi-4-mini-instruct/checkpoint-
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 10,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -83,6 +83,36 @@
|
|
| 83 |
"eval_samples_per_second": 9.025,
|
| 84 |
"eval_steps_per_second": 1.203,
|
| 85 |
"step": 50
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 86 |
}
|
| 87 |
],
|
| 88 |
"logging_steps": 10,
|
|
@@ -102,7 +132,7 @@
|
|
| 102 |
"attributes": {}
|
| 103 |
}
|
| 104 |
},
|
| 105 |
-
"total_flos":
|
| 106 |
"train_batch_size": 1,
|
| 107 |
"trial_name": null,
|
| 108 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 70,
|
| 3 |
+
"best_metric": 0.4610312283039093,
|
| 4 |
+
"best_model_checkpoint": "/content/drive/MyDrive/lora_model/outputs/task15_microsoft/Phi-4-mini-instruct/checkpoint-70",
|
| 5 |
+
"epoch": 4.666666666666667,
|
| 6 |
"eval_steps": 10,
|
| 7 |
+
"global_step": 70,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 83 |
"eval_samples_per_second": 9.025,
|
| 84 |
"eval_steps_per_second": 1.203,
|
| 85 |
"step": 50
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 4.0,
|
| 89 |
+
"grad_norm": 0.7584931254386902,
|
| 90 |
+
"learning_rate": 0.00023614713127100752,
|
| 91 |
+
"loss": 0.6581,
|
| 92 |
+
"step": 60
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 4.0,
|
| 96 |
+
"eval_loss": 0.5533124208450317,
|
| 97 |
+
"eval_runtime": 3.3307,
|
| 98 |
+
"eval_samples_per_second": 9.007,
|
| 99 |
+
"eval_steps_per_second": 1.201,
|
| 100 |
+
"step": 60
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 4.666666666666667,
|
| 104 |
+
"grad_norm": 0.6449056267738342,
|
| 105 |
+
"learning_rate": 0.00014691092063152418,
|
| 106 |
+
"loss": 0.5216,
|
| 107 |
+
"step": 70
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 4.666666666666667,
|
| 111 |
+
"eval_loss": 0.4610312283039093,
|
| 112 |
+
"eval_runtime": 3.3266,
|
| 113 |
+
"eval_samples_per_second": 9.018,
|
| 114 |
+
"eval_steps_per_second": 1.202,
|
| 115 |
+
"step": 70
|
| 116 |
}
|
| 117 |
],
|
| 118 |
"logging_steps": 10,
|
|
|
|
| 132 |
"attributes": {}
|
| 133 |
}
|
| 134 |
},
|
| 135 |
+
"total_flos": 2887531530977280.0,
|
| 136 |
"train_batch_size": 1,
|
| 137 |
"trial_name": null,
|
| 138 |
"trial_params": null
|