Model save
Browse files- README.md +13 -0
- all_results.json +6 -6
- model-00001-of-00003.safetensors +1 -1
- model-00002-of-00003.safetensors +1 -1
- model-00003-of-00003.safetensors +1 -1
- train_results.json +6 -6
- trainer_state.json +476 -100
- training_args.bin +1 -1
README.md
CHANGED
|
@@ -14,6 +14,16 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 14 |
# Mistral-7B-v0.3-spin-10k
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on an unknown dataset.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
## Model description
|
| 19 |
|
|
@@ -48,6 +58,9 @@ The following hyperparameters were used during training:
|
|
| 48 |
|
| 49 |
### Training results
|
| 50 |
|
|
|
|
|
|
|
|
|
|
| 51 |
|
| 52 |
|
| 53 |
### Framework versions
|
|
|
|
| 14 |
# Mistral-7B-v0.3-spin-10k
|
| 15 |
|
| 16 |
This model is a fine-tuned version of [mistralai/Mistral-7B-v0.3](https://huggingface.co/mistralai/Mistral-7B-v0.3) on an unknown dataset.
|
| 17 |
+
It achieves the following results on the evaluation set:
|
| 18 |
+
- Loss: 0.1078
|
| 19 |
+
- Rewards/real: 3.1288
|
| 20 |
+
- Rewards/generated: -13.0818
|
| 21 |
+
- Rewards/accuracies: 1.0
|
| 22 |
+
- Rewards/margins: 16.2106
|
| 23 |
+
- Logps/generated: -384.0584
|
| 24 |
+
- Logps/real: -105.4004
|
| 25 |
+
- Logits/generated: -2.7094
|
| 26 |
+
- Logits/real: -2.4102
|
| 27 |
|
| 28 |
## Model description
|
| 29 |
|
|
|
|
| 58 |
|
| 59 |
### Training results
|
| 60 |
|
| 61 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/real | Rewards/generated | Rewards/accuracies | Rewards/margins | Logps/generated | Logps/real | Logits/generated | Logits/real |
|
| 62 |
+
|:-------------:|:-----:|:----:|:---------------:|:------------:|:-----------------:|:------------------:|:---------------:|:---------------:|:----------:|:----------------:|:-----------:|
|
| 63 |
+
| 0.0622 | 0.64 | 200 | 0.1078 | 3.1288 | -13.0818 | 1.0 | 16.2106 | -384.0584 | -105.4004 | -2.7094 | -2.4102 |
|
| 64 |
|
| 65 |
|
| 66 |
### Framework versions
|
all_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0.
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.086211400011029,
|
| 5 |
+
"train_runtime": 3213.2155,
|
| 6 |
+
"train_samples": 9995,
|
| 7 |
+
"train_samples_per_second": 3.111,
|
| 8 |
+
"train_steps_per_second": 0.097
|
| 9 |
}
|
model-00001-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4949453792
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4a987962444565affbba933895a7a0f58ccae500e52853d1582d4fbaa9b805cf
|
| 3 |
size 4949453792
|
model-00002-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4999819336
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:92073236b3262e6737491003158e803e064cdf08c1f9d5519bbe243e478c0a56
|
| 3 |
size 4999819336
|
model-00003-of-00003.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4546807800
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:dc16aacf156b4ebf222ecdc2a6e86343c65c93eacbfcee15f655a414f99eba25
|
| 3 |
size 4546807800
|
train_results.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch": 0.
|
| 3 |
"total_flos": 0.0,
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples":
|
| 7 |
-
"train_samples_per_second":
|
| 8 |
-
"train_steps_per_second": 0.
|
| 9 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 0.9984,
|
| 3 |
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.086211400011029,
|
| 5 |
+
"train_runtime": 3213.2155,
|
| 6 |
+
"train_samples": 9995,
|
| 7 |
+
"train_samples_per_second": 3.111,
|
| 8 |
+
"train_steps_per_second": 0.097
|
| 9 |
}
|
trainer_state.json
CHANGED
|
@@ -1,22 +1,22 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 200,
|
| 6 |
-
"global_step":
|
| 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.
|
| 13 |
-
"grad_norm":
|
| 14 |
-
"learning_rate":
|
| 15 |
-
"logits/generated": -
|
| 16 |
-
"logits/real": -2.
|
| 17 |
-
"logps/generated": -
|
| 18 |
-
"logps/real": -
|
| 19 |
-
"loss": 0.
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/generated": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
@@ -24,122 +24,498 @@
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
-
"epoch": 0.
|
| 28 |
-
"grad_norm":
|
| 29 |
-
"learning_rate":
|
| 30 |
-
"logits/generated": -2.
|
| 31 |
-
"logits/real": -2.
|
| 32 |
-
"logps/generated": -
|
| 33 |
-
"logps/real": -
|
| 34 |
-
"loss": 0.
|
| 35 |
"rewards/accuracies": 0.8888888955116272,
|
| 36 |
-
"rewards/generated": -
|
| 37 |
-
"rewards/margins":
|
| 38 |
-
"rewards/real":
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
-
"epoch": 0.
|
| 43 |
-
"grad_norm": 1.
|
| 44 |
-
"learning_rate":
|
| 45 |
-
"logits/generated": -3.
|
| 46 |
-
"logits/real": -2.
|
| 47 |
-
"logps/generated": -
|
| 48 |
-
"logps/real": -
|
| 49 |
-
"loss": 0.
|
| 50 |
-
"rewards/accuracies": 1.0,
|
| 51 |
-
"rewards/generated": -
|
| 52 |
-
"rewards/margins":
|
| 53 |
-
"rewards/real":
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
-
"epoch": 0.
|
| 58 |
-
"grad_norm": 1.
|
| 59 |
-
"learning_rate":
|
| 60 |
-
"logits/generated": -2.
|
| 61 |
-
"logits/real": -2.
|
| 62 |
-
"logps/generated": -
|
| 63 |
-
"logps/real": -
|
| 64 |
-
"loss": 0.
|
| 65 |
-
"rewards/accuracies": 1.0,
|
| 66 |
-
"rewards/generated": -
|
| 67 |
-
"rewards/margins": 12.
|
| 68 |
-
"rewards/real": 3.
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
-
"epoch": 0.
|
| 73 |
-
"grad_norm": 1.
|
| 74 |
-
"learning_rate":
|
| 75 |
-
"logits/generated": -2.
|
| 76 |
-
"logits/real": -2.
|
| 77 |
-
"logps/generated": -
|
| 78 |
-
"logps/real": -
|
| 79 |
-
"loss": 0.
|
| 80 |
-
"rewards/accuracies": 1.0,
|
| 81 |
-
"rewards/generated": -
|
| 82 |
-
"rewards/margins":
|
| 83 |
-
"rewards/real":
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
-
"epoch": 0.
|
| 88 |
-
"grad_norm": 1.
|
| 89 |
-
"learning_rate":
|
| 90 |
-
"logits/generated": -2.
|
| 91 |
-
"logits/real": -2.
|
| 92 |
-
"logps/generated": -
|
| 93 |
-
"logps/real": -
|
| 94 |
-
"loss": 0.
|
| 95 |
-
"rewards/accuracies": 1.0,
|
| 96 |
-
"rewards/generated": -10.
|
| 97 |
-
"rewards/margins": 14.
|
| 98 |
-
"rewards/real":
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
-
"epoch": 0.
|
| 103 |
-
"grad_norm": 1.
|
| 104 |
-
"learning_rate":
|
| 105 |
-
"logits/generated": -2.
|
| 106 |
-
"logits/real": -2.
|
| 107 |
-
"logps/generated": -
|
| 108 |
-
"logps/real": -
|
| 109 |
-
"loss": 0.
|
| 110 |
-
"rewards/accuracies": 1.0,
|
| 111 |
-
"rewards/generated": -
|
| 112 |
-
"rewards/margins":
|
| 113 |
-
"rewards/real":
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
-
"epoch": 0.
|
| 118 |
-
"grad_norm": 1.
|
| 119 |
-
"learning_rate":
|
| 120 |
-
"logits/generated": -2.
|
| 121 |
-
"logits/real": -2.
|
| 122 |
-
"logps/generated": -
|
| 123 |
-
"logps/real": -
|
| 124 |
-
"loss": 0.
|
| 125 |
-
"rewards/accuracies": 1.0,
|
| 126 |
-
"rewards/generated": -
|
| 127 |
-
"rewards/margins":
|
| 128 |
-
"rewards/real":
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"epoch": 0.
|
| 133 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 134 |
"total_flos": 0.0,
|
| 135 |
-
"train_loss": 0.
|
| 136 |
-
"train_runtime":
|
| 137 |
-
"train_samples_per_second":
|
| 138 |
-
"train_steps_per_second": 0.
|
| 139 |
}
|
| 140 |
],
|
| 141 |
"logging_steps": 10,
|
| 142 |
-
"max_steps":
|
| 143 |
"num_input_tokens_seen": 0,
|
| 144 |
"num_train_epochs": 1,
|
| 145 |
"save_steps": 200,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984,
|
| 5 |
"eval_steps": 200,
|
| 6 |
+
"global_step": 312,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
+
"epoch": 0.0032,
|
| 13 |
+
"grad_norm": 309.07493134536526,
|
| 14 |
+
"learning_rate": 1.5625e-08,
|
| 15 |
+
"logits/generated": -3.1874351501464844,
|
| 16 |
+
"logits/real": -2.811344623565674,
|
| 17 |
+
"logps/generated": -277.39678955078125,
|
| 18 |
+
"logps/real": -164.29153442382812,
|
| 19 |
+
"loss": 0.8248,
|
| 20 |
"rewards/accuracies": 0.0,
|
| 21 |
"rewards/generated": 0.0,
|
| 22 |
"rewards/margins": 0.0,
|
|
|
|
| 24 |
"step": 1
|
| 25 |
},
|
| 26 |
{
|
| 27 |
+
"epoch": 0.032,
|
| 28 |
+
"grad_norm": 13.846248517181778,
|
| 29 |
+
"learning_rate": 1.5624999999999999e-07,
|
| 30 |
+
"logits/generated": -2.9546539783477783,
|
| 31 |
+
"logits/real": -2.5018956661224365,
|
| 32 |
+
"logps/generated": -249.66824340820312,
|
| 33 |
+
"logps/real": -122.42621612548828,
|
| 34 |
+
"loss": 0.5265,
|
| 35 |
"rewards/accuracies": 0.8888888955116272,
|
| 36 |
+
"rewards/generated": -0.8262240886688232,
|
| 37 |
+
"rewards/margins": 1.2902787923812866,
|
| 38 |
+
"rewards/real": 0.46405479311943054,
|
| 39 |
"step": 10
|
| 40 |
},
|
| 41 |
{
|
| 42 |
+
"epoch": 0.064,
|
| 43 |
+
"grad_norm": 1.537832573373159,
|
| 44 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 45 |
+
"logits/generated": -3.0161662101745605,
|
| 46 |
+
"logits/real": -2.225759267807007,
|
| 47 |
+
"logps/generated": -312.00897216796875,
|
| 48 |
+
"logps/real": -104.78621673583984,
|
| 49 |
+
"loss": 0.1131,
|
| 50 |
+
"rewards/accuracies": 1.0,
|
| 51 |
+
"rewards/generated": -5.391150951385498,
|
| 52 |
+
"rewards/margins": 8.365974426269531,
|
| 53 |
+
"rewards/real": 2.9748237133026123,
|
| 54 |
"step": 20
|
| 55 |
},
|
| 56 |
{
|
| 57 |
+
"epoch": 0.096,
|
| 58 |
+
"grad_norm": 1.5948197394392936,
|
| 59 |
+
"learning_rate": 4.6874999999999996e-07,
|
| 60 |
+
"logits/generated": -2.8325116634368896,
|
| 61 |
+
"logits/real": -2.309696674346924,
|
| 62 |
+
"logps/generated": -343.28424072265625,
|
| 63 |
+
"logps/real": -100.86705780029297,
|
| 64 |
+
"loss": 0.1004,
|
| 65 |
+
"rewards/accuracies": 1.0,
|
| 66 |
+
"rewards/generated": -8.481008529663086,
|
| 67 |
+
"rewards/margins": 12.336121559143066,
|
| 68 |
+
"rewards/real": 3.855113983154297,
|
| 69 |
"step": 30
|
| 70 |
},
|
| 71 |
{
|
| 72 |
+
"epoch": 0.128,
|
| 73 |
+
"grad_norm": 1.360990457967047,
|
| 74 |
+
"learning_rate": 4.857142857142857e-07,
|
| 75 |
+
"logits/generated": -2.8159852027893066,
|
| 76 |
+
"logits/real": -2.518409013748169,
|
| 77 |
+
"logps/generated": -355.71441650390625,
|
| 78 |
+
"logps/real": -102.14837646484375,
|
| 79 |
+
"loss": 0.1018,
|
| 80 |
+
"rewards/accuracies": 1.0,
|
| 81 |
+
"rewards/generated": -10.088699340820312,
|
| 82 |
+
"rewards/margins": 14.381967544555664,
|
| 83 |
+
"rewards/real": 4.293266773223877,
|
| 84 |
"step": 40
|
| 85 |
},
|
| 86 |
{
|
| 87 |
+
"epoch": 0.16,
|
| 88 |
+
"grad_norm": 1.1637353343229784,
|
| 89 |
+
"learning_rate": 4.6785714285714283e-07,
|
| 90 |
+
"logits/generated": -2.810716152191162,
|
| 91 |
+
"logits/real": -2.397970199584961,
|
| 92 |
+
"logps/generated": -361.4827575683594,
|
| 93 |
+
"logps/real": -93.07586669921875,
|
| 94 |
+
"loss": 0.0866,
|
| 95 |
+
"rewards/accuracies": 1.0,
|
| 96 |
+
"rewards/generated": -10.693376541137695,
|
| 97 |
+
"rewards/margins": 14.997743606567383,
|
| 98 |
+
"rewards/real": 4.304368019104004,
|
| 99 |
"step": 50
|
| 100 |
},
|
| 101 |
{
|
| 102 |
+
"epoch": 0.192,
|
| 103 |
+
"grad_norm": 1.5949083844151817,
|
| 104 |
+
"learning_rate": 4.5e-07,
|
| 105 |
+
"logits/generated": -2.5569677352905273,
|
| 106 |
+
"logits/real": -2.209700584411621,
|
| 107 |
+
"logps/generated": -356.9215393066406,
|
| 108 |
+
"logps/real": -90.63230895996094,
|
| 109 |
+
"loss": 0.0871,
|
| 110 |
+
"rewards/accuracies": 1.0,
|
| 111 |
+
"rewards/generated": -11.640242576599121,
|
| 112 |
+
"rewards/margins": 16.65045928955078,
|
| 113 |
+
"rewards/real": 5.0102152824401855,
|
| 114 |
"step": 60
|
| 115 |
},
|
| 116 |
{
|
| 117 |
+
"epoch": 0.224,
|
| 118 |
+
"grad_norm": 1.1997502879904265,
|
| 119 |
+
"learning_rate": 4.3214285714285713e-07,
|
| 120 |
+
"logits/generated": -2.778857946395874,
|
| 121 |
+
"logits/real": -2.4418845176696777,
|
| 122 |
+
"logps/generated": -363.03106689453125,
|
| 123 |
+
"logps/real": -83.89906311035156,
|
| 124 |
+
"loss": 0.0889,
|
| 125 |
+
"rewards/accuracies": 1.0,
|
| 126 |
+
"rewards/generated": -11.029804229736328,
|
| 127 |
+
"rewards/margins": 16.131534576416016,
|
| 128 |
+
"rewards/real": 5.1017303466796875,
|
| 129 |
"step": 70
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.256,
|
| 133 |
+
"grad_norm": 1.2541389286949767,
|
| 134 |
+
"learning_rate": 4.142857142857143e-07,
|
| 135 |
+
"logits/generated": -2.748169422149658,
|
| 136 |
+
"logits/real": -2.537468910217285,
|
| 137 |
+
"logps/generated": -352.0079040527344,
|
| 138 |
+
"logps/real": -90.74658966064453,
|
| 139 |
+
"loss": 0.0847,
|
| 140 |
+
"rewards/accuracies": 1.0,
|
| 141 |
+
"rewards/generated": -10.95399284362793,
|
| 142 |
+
"rewards/margins": 16.11164093017578,
|
| 143 |
+
"rewards/real": 5.157646179199219,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.288,
|
| 148 |
+
"grad_norm": 1.3716409313987092,
|
| 149 |
+
"learning_rate": 3.9642857142857137e-07,
|
| 150 |
+
"logits/generated": -2.7057158946990967,
|
| 151 |
+
"logits/real": -2.4590487480163574,
|
| 152 |
+
"logps/generated": -369.472412109375,
|
| 153 |
+
"logps/real": -93.29542541503906,
|
| 154 |
+
"loss": 0.0867,
|
| 155 |
+
"rewards/accuracies": 1.0,
|
| 156 |
+
"rewards/generated": -11.155221939086914,
|
| 157 |
+
"rewards/margins": 17.421768188476562,
|
| 158 |
+
"rewards/real": 6.266547203063965,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.32,
|
| 163 |
+
"grad_norm": 1.2776829690193363,
|
| 164 |
+
"learning_rate": 3.785714285714285e-07,
|
| 165 |
+
"logits/generated": -2.7385950088500977,
|
| 166 |
+
"logits/real": -2.558948040008545,
|
| 167 |
+
"logps/generated": -360.800048828125,
|
| 168 |
+
"logps/real": -94.03065490722656,
|
| 169 |
+
"loss": 0.0812,
|
| 170 |
+
"rewards/accuracies": 1.0,
|
| 171 |
+
"rewards/generated": -11.058977127075195,
|
| 172 |
+
"rewards/margins": 17.12522315979004,
|
| 173 |
+
"rewards/real": 6.066246032714844,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.352,
|
| 178 |
+
"grad_norm": 1.1502347571032556,
|
| 179 |
+
"learning_rate": 3.607142857142857e-07,
|
| 180 |
+
"logits/generated": -2.7538702487945557,
|
| 181 |
+
"logits/real": -2.577524185180664,
|
| 182 |
+
"logps/generated": -369.98388671875,
|
| 183 |
+
"logps/real": -81.54554748535156,
|
| 184 |
+
"loss": 0.0768,
|
| 185 |
+
"rewards/accuracies": 1.0,
|
| 186 |
+
"rewards/generated": -12.037736892700195,
|
| 187 |
+
"rewards/margins": 18.702112197875977,
|
| 188 |
+
"rewards/real": 6.664376735687256,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.384,
|
| 193 |
+
"grad_norm": 1.3343308865576284,
|
| 194 |
+
"learning_rate": 3.4285714285714286e-07,
|
| 195 |
+
"logits/generated": -2.6860907077789307,
|
| 196 |
+
"logits/real": -2.498666763305664,
|
| 197 |
+
"logps/generated": -367.09027099609375,
|
| 198 |
+
"logps/real": -77.16963195800781,
|
| 199 |
+
"loss": 0.0813,
|
| 200 |
+
"rewards/accuracies": 1.0,
|
| 201 |
+
"rewards/generated": -12.206511497497559,
|
| 202 |
+
"rewards/margins": 18.975597381591797,
|
| 203 |
+
"rewards/real": 6.7690839767456055,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.416,
|
| 208 |
+
"grad_norm": 1.3770646126862427,
|
| 209 |
+
"learning_rate": 3.25e-07,
|
| 210 |
+
"logits/generated": -2.624868869781494,
|
| 211 |
+
"logits/real": -2.4215009212493896,
|
| 212 |
+
"logps/generated": -371.2015380859375,
|
| 213 |
+
"logps/real": -70.61559295654297,
|
| 214 |
+
"loss": 0.0775,
|
| 215 |
+
"rewards/accuracies": 1.0,
|
| 216 |
+
"rewards/generated": -12.158758163452148,
|
| 217 |
+
"rewards/margins": 18.808048248291016,
|
| 218 |
+
"rewards/real": 6.649291038513184,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.448,
|
| 223 |
+
"grad_norm": 1.4072742854376892,
|
| 224 |
+
"learning_rate": 3.0714285714285716e-07,
|
| 225 |
+
"logits/generated": -2.632169246673584,
|
| 226 |
+
"logits/real": -2.5124592781066895,
|
| 227 |
+
"logps/generated": -378.91949462890625,
|
| 228 |
+
"logps/real": -85.79364013671875,
|
| 229 |
+
"loss": 0.0783,
|
| 230 |
+
"rewards/accuracies": 1.0,
|
| 231 |
+
"rewards/generated": -12.807360649108887,
|
| 232 |
+
"rewards/margins": 20.574337005615234,
|
| 233 |
+
"rewards/real": 7.7669782638549805,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.48,
|
| 238 |
+
"grad_norm": 1.6731854026570883,
|
| 239 |
+
"learning_rate": 2.892857142857143e-07,
|
| 240 |
+
"logits/generated": -2.7312309741973877,
|
| 241 |
+
"logits/real": -2.513092517852783,
|
| 242 |
+
"logps/generated": -376.5494689941406,
|
| 243 |
+
"logps/real": -86.8825912475586,
|
| 244 |
+
"loss": 0.0719,
|
| 245 |
+
"rewards/accuracies": 1.0,
|
| 246 |
+
"rewards/generated": -12.506660461425781,
|
| 247 |
+
"rewards/margins": 19.993375778198242,
|
| 248 |
+
"rewards/real": 7.486716270446777,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.512,
|
| 253 |
+
"grad_norm": 1.7208597186189485,
|
| 254 |
+
"learning_rate": 2.714285714285714e-07,
|
| 255 |
+
"logits/generated": -2.6538147926330566,
|
| 256 |
+
"logits/real": -2.4443583488464355,
|
| 257 |
+
"logps/generated": -379.8941955566406,
|
| 258 |
+
"logps/real": -63.8643798828125,
|
| 259 |
+
"loss": 0.0664,
|
| 260 |
+
"rewards/accuracies": 1.0,
|
| 261 |
+
"rewards/generated": -12.761404991149902,
|
| 262 |
+
"rewards/margins": 21.195444107055664,
|
| 263 |
+
"rewards/real": 8.434039115905762,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.544,
|
| 268 |
+
"grad_norm": 3.296728367897804,
|
| 269 |
+
"learning_rate": 2.5357142857142855e-07,
|
| 270 |
+
"logits/generated": -2.679011821746826,
|
| 271 |
+
"logits/real": -2.4271175861358643,
|
| 272 |
+
"logps/generated": -368.987060546875,
|
| 273 |
+
"logps/real": -65.5553207397461,
|
| 274 |
+
"loss": 0.0683,
|
| 275 |
+
"rewards/accuracies": 1.0,
|
| 276 |
+
"rewards/generated": -12.584733963012695,
|
| 277 |
+
"rewards/margins": 19.998342514038086,
|
| 278 |
+
"rewards/real": 7.413609504699707,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.576,
|
| 283 |
+
"grad_norm": 1.509303126217559,
|
| 284 |
+
"learning_rate": 2.357142857142857e-07,
|
| 285 |
+
"logits/generated": -2.706902503967285,
|
| 286 |
+
"logits/real": -2.339202404022217,
|
| 287 |
+
"logps/generated": -393.6845703125,
|
| 288 |
+
"logps/real": -57.10161590576172,
|
| 289 |
+
"loss": 0.0613,
|
| 290 |
+
"rewards/accuracies": 1.0,
|
| 291 |
+
"rewards/generated": -13.54321575164795,
|
| 292 |
+
"rewards/margins": 20.973468780517578,
|
| 293 |
+
"rewards/real": 7.430250644683838,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.608,
|
| 298 |
+
"grad_norm": 1.6226015420004847,
|
| 299 |
+
"learning_rate": 2.1785714285714284e-07,
|
| 300 |
+
"logits/generated": -2.6822266578674316,
|
| 301 |
+
"logits/real": -2.390279769897461,
|
| 302 |
+
"logps/generated": -391.8895263671875,
|
| 303 |
+
"logps/real": -64.2778549194336,
|
| 304 |
+
"loss": 0.0593,
|
| 305 |
+
"rewards/accuracies": 1.0,
|
| 306 |
+
"rewards/generated": -13.559152603149414,
|
| 307 |
+
"rewards/margins": 21.35408592224121,
|
| 308 |
+
"rewards/real": 7.794934272766113,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.64,
|
| 313 |
+
"grad_norm": 1.4621480589976257,
|
| 314 |
+
"learning_rate": 2e-07,
|
| 315 |
+
"logits/generated": -2.64497447013855,
|
| 316 |
+
"logits/real": -2.3843231201171875,
|
| 317 |
+
"logps/generated": -378.09307861328125,
|
| 318 |
+
"logps/real": -53.13884353637695,
|
| 319 |
+
"loss": 0.0622,
|
| 320 |
+
"rewards/accuracies": 1.0,
|
| 321 |
+
"rewards/generated": -12.936541557312012,
|
| 322 |
+
"rewards/margins": 21.156665802001953,
|
| 323 |
+
"rewards/real": 8.220122337341309,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.64,
|
| 328 |
+
"eval_logits/generated": -2.709409236907959,
|
| 329 |
+
"eval_logits/real": -2.410205602645874,
|
| 330 |
+
"eval_logps/generated": -384.0584411621094,
|
| 331 |
+
"eval_logps/real": -105.40039825439453,
|
| 332 |
+
"eval_loss": 0.10775644332170486,
|
| 333 |
+
"eval_rewards/accuracies": 1.0,
|
| 334 |
+
"eval_rewards/generated": -13.081828117370605,
|
| 335 |
+
"eval_rewards/margins": 16.210613250732422,
|
| 336 |
+
"eval_rewards/real": 3.1287851333618164,
|
| 337 |
+
"eval_runtime": 31.4401,
|
| 338 |
+
"eval_samples_per_second": 6.361,
|
| 339 |
+
"eval_steps_per_second": 0.413,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.672,
|
| 344 |
+
"grad_norm": 1.6581201639125833,
|
| 345 |
+
"learning_rate": 1.8214285714285714e-07,
|
| 346 |
+
"logits/generated": -2.6259570121765137,
|
| 347 |
+
"logits/real": -2.3245978355407715,
|
| 348 |
+
"logps/generated": -380.79241943359375,
|
| 349 |
+
"logps/real": -50.675270080566406,
|
| 350 |
+
"loss": 0.0563,
|
| 351 |
+
"rewards/accuracies": 1.0,
|
| 352 |
+
"rewards/generated": -13.35838508605957,
|
| 353 |
+
"rewards/margins": 21.13888168334961,
|
| 354 |
+
"rewards/real": 7.780497074127197,
|
| 355 |
+
"step": 210
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.704,
|
| 359 |
+
"grad_norm": 2.0377232989636846,
|
| 360 |
+
"learning_rate": 1.6428571428571429e-07,
|
| 361 |
+
"logits/generated": -2.6223349571228027,
|
| 362 |
+
"logits/real": -2.4116692543029785,
|
| 363 |
+
"logps/generated": -388.216064453125,
|
| 364 |
+
"logps/real": -75.10206604003906,
|
| 365 |
+
"loss": 0.0601,
|
| 366 |
+
"rewards/accuracies": 1.0,
|
| 367 |
+
"rewards/generated": -13.288862228393555,
|
| 368 |
+
"rewards/margins": 22.80645751953125,
|
| 369 |
+
"rewards/real": 9.51759147644043,
|
| 370 |
+
"step": 220
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.736,
|
| 374 |
+
"grad_norm": 1.4117008274451168,
|
| 375 |
+
"learning_rate": 1.4642857142857143e-07,
|
| 376 |
+
"logits/generated": -2.6506168842315674,
|
| 377 |
+
"logits/real": -2.356983184814453,
|
| 378 |
+
"logps/generated": -402.24298095703125,
|
| 379 |
+
"logps/real": -52.900665283203125,
|
| 380 |
+
"loss": 0.051,
|
| 381 |
+
"rewards/accuracies": 1.0,
|
| 382 |
+
"rewards/generated": -14.113009452819824,
|
| 383 |
+
"rewards/margins": 22.862171173095703,
|
| 384 |
+
"rewards/real": 8.749161720275879,
|
| 385 |
+
"step": 230
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.768,
|
| 389 |
+
"grad_norm": 1.5408677183120254,
|
| 390 |
+
"learning_rate": 1.2857142857142855e-07,
|
| 391 |
+
"logits/generated": -2.663072109222412,
|
| 392 |
+
"logits/real": -2.295074939727783,
|
| 393 |
+
"logps/generated": -392.62335205078125,
|
| 394 |
+
"logps/real": -48.768280029296875,
|
| 395 |
+
"loss": 0.0565,
|
| 396 |
+
"rewards/accuracies": 1.0,
|
| 397 |
+
"rewards/generated": -13.465108871459961,
|
| 398 |
+
"rewards/margins": 21.69220542907715,
|
| 399 |
+
"rewards/real": 8.227096557617188,
|
| 400 |
+
"step": 240
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.8,
|
| 404 |
+
"grad_norm": 1.6617128645150359,
|
| 405 |
+
"learning_rate": 1.107142857142857e-07,
|
| 406 |
+
"logits/generated": -2.659288167953491,
|
| 407 |
+
"logits/real": -2.284973621368408,
|
| 408 |
+
"logps/generated": -391.0361022949219,
|
| 409 |
+
"logps/real": -49.54872131347656,
|
| 410 |
+
"loss": 0.0525,
|
| 411 |
+
"rewards/accuracies": 1.0,
|
| 412 |
+
"rewards/generated": -13.513246536254883,
|
| 413 |
+
"rewards/margins": 21.818363189697266,
|
| 414 |
+
"rewards/real": 8.30511474609375,
|
| 415 |
+
"step": 250
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.832,
|
| 419 |
+
"grad_norm": 1.9545108954634698,
|
| 420 |
+
"learning_rate": 9.285714285714286e-08,
|
| 421 |
+
"logits/generated": -2.5850462913513184,
|
| 422 |
+
"logits/real": -2.2890560626983643,
|
| 423 |
+
"logps/generated": -377.42095947265625,
|
| 424 |
+
"logps/real": -50.45993423461914,
|
| 425 |
+
"loss": 0.0541,
|
| 426 |
+
"rewards/accuracies": 1.0,
|
| 427 |
+
"rewards/generated": -13.420709609985352,
|
| 428 |
+
"rewards/margins": 21.94915008544922,
|
| 429 |
+
"rewards/real": 8.528444290161133,
|
| 430 |
+
"step": 260
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.864,
|
| 434 |
+
"grad_norm": 1.663160142834365,
|
| 435 |
+
"learning_rate": 7.5e-08,
|
| 436 |
+
"logits/generated": -2.639012098312378,
|
| 437 |
+
"logits/real": -2.241528272628784,
|
| 438 |
+
"logps/generated": -385.27972412109375,
|
| 439 |
+
"logps/real": -44.987178802490234,
|
| 440 |
+
"loss": 0.0519,
|
| 441 |
+
"rewards/accuracies": 1.0,
|
| 442 |
+
"rewards/generated": -13.790578842163086,
|
| 443 |
+
"rewards/margins": 21.756423950195312,
|
| 444 |
+
"rewards/real": 7.965843200683594,
|
| 445 |
+
"step": 270
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.896,
|
| 449 |
+
"grad_norm": 1.7475888656947995,
|
| 450 |
+
"learning_rate": 5.714285714285714e-08,
|
| 451 |
+
"logits/generated": -2.634594440460205,
|
| 452 |
+
"logits/real": -2.309357166290283,
|
| 453 |
+
"logps/generated": -383.92266845703125,
|
| 454 |
+
"logps/real": -47.04390335083008,
|
| 455 |
+
"loss": 0.0523,
|
| 456 |
+
"rewards/accuracies": 1.0,
|
| 457 |
+
"rewards/generated": -13.899473190307617,
|
| 458 |
+
"rewards/margins": 22.34453582763672,
|
| 459 |
+
"rewards/real": 8.445062637329102,
|
| 460 |
+
"step": 280
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.928,
|
| 464 |
+
"grad_norm": 1.8843842875992995,
|
| 465 |
+
"learning_rate": 3.9285714285714285e-08,
|
| 466 |
+
"logits/generated": -2.6036441326141357,
|
| 467 |
+
"logits/real": -2.2659499645233154,
|
| 468 |
+
"logps/generated": -395.3347473144531,
|
| 469 |
+
"logps/real": -55.319923400878906,
|
| 470 |
+
"loss": 0.0537,
|
| 471 |
+
"rewards/accuracies": 1.0,
|
| 472 |
+
"rewards/generated": -14.550788879394531,
|
| 473 |
+
"rewards/margins": 24.063692092895508,
|
| 474 |
+
"rewards/real": 9.512903213500977,
|
| 475 |
+
"step": 290
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.96,
|
| 479 |
+
"grad_norm": 1.7336216562094997,
|
| 480 |
+
"learning_rate": 2.142857142857143e-08,
|
| 481 |
+
"logits/generated": -2.5489935874938965,
|
| 482 |
+
"logits/real": -2.3403241634368896,
|
| 483 |
+
"logps/generated": -381.0558166503906,
|
| 484 |
+
"logps/real": -60.32090377807617,
|
| 485 |
+
"loss": 0.0515,
|
| 486 |
+
"rewards/accuracies": 1.0,
|
| 487 |
+
"rewards/generated": -13.906036376953125,
|
| 488 |
+
"rewards/margins": 23.397653579711914,
|
| 489 |
+
"rewards/real": 9.491620063781738,
|
| 490 |
+
"step": 300
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.992,
|
| 494 |
+
"grad_norm": 1.7332047882804558,
|
| 495 |
+
"learning_rate": 3.571428571428571e-09,
|
| 496 |
+
"logits/generated": -2.6001510620117188,
|
| 497 |
+
"logits/real": -2.3211655616760254,
|
| 498 |
+
"logps/generated": -384.77154541015625,
|
| 499 |
+
"logps/real": -53.601829528808594,
|
| 500 |
+
"loss": 0.0525,
|
| 501 |
+
"rewards/accuracies": 1.0,
|
| 502 |
+
"rewards/generated": -13.838152885437012,
|
| 503 |
+
"rewards/margins": 23.025869369506836,
|
| 504 |
+
"rewards/real": 9.187714576721191,
|
| 505 |
+
"step": 310
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.9984,
|
| 509 |
+
"step": 312,
|
| 510 |
"total_flos": 0.0,
|
| 511 |
+
"train_loss": 0.086211400011029,
|
| 512 |
+
"train_runtime": 3213.2155,
|
| 513 |
+
"train_samples_per_second": 3.111,
|
| 514 |
+
"train_steps_per_second": 0.097
|
| 515 |
}
|
| 516 |
],
|
| 517 |
"logging_steps": 10,
|
| 518 |
+
"max_steps": 312,
|
| 519 |
"num_input_tokens_seen": 0,
|
| 520 |
"num_train_epochs": 1,
|
| 521 |
"save_steps": 200,
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 6392
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a29f4504162220fe9c1bf29b0b09c7afcda7d8ab0595ecbc6faea0331ff7b85
|
| 3 |
size 6392
|