scottsuk0306
commited on
Commit
•
abd3264
1
Parent(s):
78f14a9
Model save
Browse files- README.md +8 -12
- all_results.json +6 -11
- model-00001-of-00005.safetensors +1 -1
- model-00002-of-00005.safetensors +1 -1
- model-00003-of-00005.safetensors +1 -1
- model-00004-of-00005.safetensors +1 -1
- model-00005-of-00005.safetensors +1 -1
- tokenizer.json +2 -2
- train_results.json +6 -6
- trainer_state.json +130 -25
- training_args.bin +1 -1
README.md
CHANGED
@@ -2,10 +2,6 @@
|
|
2 |
license: gemma
|
3 |
base_model: google/gemma-2-9b
|
4 |
tags:
|
5 |
-
- easylm
|
6 |
-
- trl
|
7 |
-
- sft
|
8 |
-
- generated_from_trainer
|
9 |
- trl
|
10 |
- sft
|
11 |
- generated_from_trainer
|
@@ -23,7 +19,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
23 |
|
24 |
This model is a fine-tuned version of [google/gemma-2-9b](https://huggingface.co/google/gemma-2-9b) on the alpaca_farm dataset.
|
25 |
It achieves the following results on the evaluation set:
|
26 |
-
- Loss: 1.
|
27 |
|
28 |
## Model description
|
29 |
|
@@ -43,13 +39,13 @@ More information needed
|
|
43 |
|
44 |
The following hyperparameters were used during training:
|
45 |
- learning_rate: 2e-05
|
46 |
-
- train_batch_size:
|
47 |
-
- eval_batch_size:
|
48 |
- seed: 42
|
49 |
- distributed_type: multi-GPU
|
50 |
- num_devices: 4
|
51 |
-
- total_train_batch_size:
|
52 |
-
- total_eval_batch_size:
|
53 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
54 |
- lr_scheduler_type: cosine
|
55 |
- num_epochs: 3
|
@@ -58,9 +54,9 @@ The following hyperparameters were used during training:
|
|
58 |
|
59 |
| Training Loss | Epoch | Step | Validation Loss |
|
60 |
|:-------------:|:-----:|:----:|:---------------:|
|
61 |
-
|
|
62 |
-
|
|
63 |
-
|
|
64 |
|
65 |
|
66 |
### Framework versions
|
|
|
2 |
license: gemma
|
3 |
base_model: google/gemma-2-9b
|
4 |
tags:
|
|
|
|
|
|
|
|
|
5 |
- trl
|
6 |
- sft
|
7 |
- generated_from_trainer
|
|
|
19 |
|
20 |
This model is a fine-tuned version of [google/gemma-2-9b](https://huggingface.co/google/gemma-2-9b) on the alpaca_farm dataset.
|
21 |
It achieves the following results on the evaluation set:
|
22 |
+
- Loss: 1.4481
|
23 |
|
24 |
## Model description
|
25 |
|
|
|
39 |
|
40 |
The following hyperparameters were used during training:
|
41 |
- learning_rate: 2e-05
|
42 |
+
- train_batch_size: 1
|
43 |
+
- eval_batch_size: 1
|
44 |
- seed: 42
|
45 |
- distributed_type: multi-GPU
|
46 |
- num_devices: 4
|
47 |
+
- total_train_batch_size: 4
|
48 |
+
- total_eval_batch_size: 4
|
49 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
50 |
- lr_scheduler_type: cosine
|
51 |
- num_epochs: 3
|
|
|
54 |
|
55 |
| Training Loss | Epoch | Step | Validation Loss |
|
56 |
|:-------------:|:-----:|:----:|:---------------:|
|
57 |
+
| 0.9381 | 1.0 | 2500 | 0.9376 |
|
58 |
+
| 0.4124 | 2.0 | 5000 | 1.0478 |
|
59 |
+
| 0.1515 | 3.0 | 7500 | 1.4481 |
|
60 |
|
61 |
|
62 |
### Framework versions
|
all_results.json
CHANGED
@@ -1,14 +1,9 @@
|
|
1 |
{
|
2 |
"epoch": 3.0,
|
3 |
-
"
|
4 |
-
"
|
5 |
-
"
|
6 |
-
"
|
7 |
-
"
|
8 |
-
"
|
9 |
-
"train_loss": 0.4380333729279347,
|
10 |
-
"train_runtime": 62.4476,
|
11 |
-
"train_samples": 100,
|
12 |
-
"train_samples_per_second": 4.804,
|
13 |
-
"train_steps_per_second": 0.625
|
14 |
}
|
|
|
1 |
{
|
2 |
"epoch": 3.0,
|
3 |
+
"total_flos": 2.3924465601190298e+17,
|
4 |
+
"train_loss": 0.5169153635660807,
|
5 |
+
"train_runtime": 5042.7608,
|
6 |
+
"train_samples": 10000,
|
7 |
+
"train_samples_per_second": 5.949,
|
8 |
+
"train_steps_per_second": 1.487
|
|
|
|
|
|
|
|
|
|
|
9 |
}
|
model-00001-of-00005.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4903351912
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d5961b1d69d275c39dc069772d844b00dc76548033be4945b533f969d48cf2d2
|
3 |
size 4903351912
|
model-00002-of-00005.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4947570872
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d9412bf9310b91d24ab66be912aaaca7e1b4a948ae143bf3a3cc544989ac5acf
|
3 |
size 4947570872
|
model-00003-of-00005.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4962221464
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:76ad58c2416f8d3e641c4e0a048d7e79be909b3504d20083fe875ccc83e67442
|
3 |
size 4962221464
|
model-00004-of-00005.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3670322200
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f37d1089517faab0a6bd850d428abecdbe471a6fdd34d39e24e0d6092c17ea3d
|
3 |
size 3670322200
|
model-00005-of-00005.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1835008128
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:005757341b2e5052d7c5f78e1807a373a49bdd0938d82a2328f33cad9a6579a5
|
3 |
size 1835008128
|
tokenizer.json
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 sha256:7da53ca29fb16f6b2489482fc0bc6a394162cdab14d12764a1755ebc583fea79
|
3 |
+
size 17518525
|
train_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"epoch": 3.0,
|
3 |
-
"total_flos":
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
-
"train_samples":
|
7 |
-
"train_samples_per_second":
|
8 |
-
"train_steps_per_second":
|
9 |
}
|
|
|
1 |
{
|
2 |
"epoch": 3.0,
|
3 |
+
"total_flos": 2.3924465601190298e+17,
|
4 |
+
"train_loss": 0.5169153635660807,
|
5 |
+
"train_runtime": 5042.7608,
|
6 |
+
"train_samples": 10000,
|
7 |
+
"train_samples_per_second": 5.949,
|
8 |
+
"train_steps_per_second": 1.487
|
9 |
}
|
trainer_state.json
CHANGED
@@ -3,47 +3,152 @@
|
|
3 |
"best_model_checkpoint": null,
|
4 |
"epoch": 3.0,
|
5 |
"eval_steps": 500,
|
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": 1.0,
|
13 |
-
"
|
14 |
-
"
|
15 |
-
"
|
16 |
-
"
|
17 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
},
|
19 |
{
|
20 |
"epoch": 2.0,
|
21 |
-
"eval_loss":
|
22 |
-
"eval_runtime":
|
23 |
-
"eval_samples_per_second":
|
24 |
-
"eval_steps_per_second":
|
25 |
-
"step":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
},
|
27 |
{
|
28 |
"epoch": 3.0,
|
29 |
-
"eval_loss": 1.
|
30 |
-
"eval_runtime":
|
31 |
-
"eval_samples_per_second":
|
32 |
-
"eval_steps_per_second":
|
33 |
-
"step":
|
34 |
},
|
35 |
{
|
36 |
"epoch": 3.0,
|
37 |
-
"step":
|
38 |
-
"total_flos":
|
39 |
-
"train_loss": 0.
|
40 |
-
"train_runtime":
|
41 |
-
"train_samples_per_second":
|
42 |
-
"train_steps_per_second":
|
43 |
}
|
44 |
],
|
45 |
"logging_steps": 500,
|
46 |
-
"max_steps":
|
47 |
"num_input_tokens_seen": 0,
|
48 |
"num_train_epochs": 3,
|
49 |
"save_steps": 500,
|
@@ -59,8 +164,8 @@
|
|
59 |
"attributes": {}
|
60 |
}
|
61 |
},
|
62 |
-
"total_flos":
|
63 |
-
"train_batch_size":
|
64 |
"trial_name": null,
|
65 |
"trial_params": null
|
66 |
}
|
|
|
3 |
"best_model_checkpoint": null,
|
4 |
"epoch": 3.0,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 7500,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
10 |
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.2,
|
13 |
+
"grad_norm": 5.575812339782715,
|
14 |
+
"learning_rate": 1.9781476007338058e-05,
|
15 |
+
"loss": 0.9324,
|
16 |
+
"step": 500
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.4,
|
20 |
+
"grad_norm": 4.583652973175049,
|
21 |
+
"learning_rate": 1.913545457642601e-05,
|
22 |
+
"loss": 0.9619,
|
23 |
+
"step": 1000
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.6,
|
27 |
+
"grad_norm": 3.9456448554992676,
|
28 |
+
"learning_rate": 1.8090169943749477e-05,
|
29 |
+
"loss": 0.979,
|
30 |
+
"step": 1500
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.8,
|
34 |
+
"grad_norm": 5.700883865356445,
|
35 |
+
"learning_rate": 1.6691306063588583e-05,
|
36 |
+
"loss": 0.9591,
|
37 |
+
"step": 2000
|
38 |
+
},
|
39 |
{
|
40 |
"epoch": 1.0,
|
41 |
+
"grad_norm": 3.7766757011413574,
|
42 |
+
"learning_rate": 1.5000000000000002e-05,
|
43 |
+
"loss": 0.9381,
|
44 |
+
"step": 2500
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 1.0,
|
48 |
+
"eval_loss": 0.937586784362793,
|
49 |
+
"eval_runtime": 43.9207,
|
50 |
+
"eval_samples_per_second": 45.537,
|
51 |
+
"eval_steps_per_second": 11.384,
|
52 |
+
"step": 2500
|
53 |
+
},
|
54 |
+
{
|
55 |
+
"epoch": 1.2,
|
56 |
+
"grad_norm": 4.477228164672852,
|
57 |
+
"learning_rate": 1.3090169943749475e-05,
|
58 |
+
"loss": 0.4559,
|
59 |
+
"step": 3000
|
60 |
+
},
|
61 |
+
{
|
62 |
+
"epoch": 1.4,
|
63 |
+
"grad_norm": 4.878340721130371,
|
64 |
+
"learning_rate": 1.1045284632676535e-05,
|
65 |
+
"loss": 0.4521,
|
66 |
+
"step": 3500
|
67 |
+
},
|
68 |
+
{
|
69 |
+
"epoch": 1.6,
|
70 |
+
"grad_norm": 3.9909627437591553,
|
71 |
+
"learning_rate": 8.954715367323468e-06,
|
72 |
+
"loss": 0.4468,
|
73 |
+
"step": 4000
|
74 |
+
},
|
75 |
+
{
|
76 |
+
"epoch": 1.8,
|
77 |
+
"grad_norm": 2.667670726776123,
|
78 |
+
"learning_rate": 6.909830056250527e-06,
|
79 |
+
"loss": 0.4321,
|
80 |
+
"step": 4500
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"epoch": 2.0,
|
84 |
+
"grad_norm": 4.369731903076172,
|
85 |
+
"learning_rate": 5.000000000000003e-06,
|
86 |
+
"loss": 0.4124,
|
87 |
+
"step": 5000
|
88 |
},
|
89 |
{
|
90 |
"epoch": 2.0,
|
91 |
+
"eval_loss": 1.0478131771087646,
|
92 |
+
"eval_runtime": 41.0932,
|
93 |
+
"eval_samples_per_second": 48.67,
|
94 |
+
"eval_steps_per_second": 12.167,
|
95 |
+
"step": 5000
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"epoch": 2.2,
|
99 |
+
"grad_norm": 2.2593352794647217,
|
100 |
+
"learning_rate": 3.308693936411421e-06,
|
101 |
+
"loss": 0.1657,
|
102 |
+
"step": 5500
|
103 |
+
},
|
104 |
+
{
|
105 |
+
"epoch": 2.4,
|
106 |
+
"grad_norm": 2.3517982959747314,
|
107 |
+
"learning_rate": 1.9098300562505266e-06,
|
108 |
+
"loss": 0.1616,
|
109 |
+
"step": 6000
|
110 |
+
},
|
111 |
+
{
|
112 |
+
"epoch": 2.6,
|
113 |
+
"grad_norm": 2.2167694568634033,
|
114 |
+
"learning_rate": 8.645454235739903e-07,
|
115 |
+
"loss": 0.153,
|
116 |
+
"step": 6500
|
117 |
+
},
|
118 |
+
{
|
119 |
+
"epoch": 2.8,
|
120 |
+
"grad_norm": 2.2378008365631104,
|
121 |
+
"learning_rate": 2.1852399266194312e-07,
|
122 |
+
"loss": 0.1521,
|
123 |
+
"step": 7000
|
124 |
+
},
|
125 |
+
{
|
126 |
+
"epoch": 3.0,
|
127 |
+
"grad_norm": 1.863773226737976,
|
128 |
+
"learning_rate": 0.0,
|
129 |
+
"loss": 0.1515,
|
130 |
+
"step": 7500
|
131 |
},
|
132 |
{
|
133 |
"epoch": 3.0,
|
134 |
+
"eval_loss": 1.4480849504470825,
|
135 |
+
"eval_runtime": 41.4492,
|
136 |
+
"eval_samples_per_second": 48.252,
|
137 |
+
"eval_steps_per_second": 12.063,
|
138 |
+
"step": 7500
|
139 |
},
|
140 |
{
|
141 |
"epoch": 3.0,
|
142 |
+
"step": 7500,
|
143 |
+
"total_flos": 2.3924465601190298e+17,
|
144 |
+
"train_loss": 0.5169153635660807,
|
145 |
+
"train_runtime": 5042.7608,
|
146 |
+
"train_samples_per_second": 5.949,
|
147 |
+
"train_steps_per_second": 1.487
|
148 |
}
|
149 |
],
|
150 |
"logging_steps": 500,
|
151 |
+
"max_steps": 7500,
|
152 |
"num_input_tokens_seen": 0,
|
153 |
"num_train_epochs": 3,
|
154 |
"save_steps": 500,
|
|
|
164 |
"attributes": {}
|
165 |
}
|
166 |
},
|
167 |
+
"total_flos": 2.3924465601190298e+17,
|
168 |
+
"train_batch_size": 1,
|
169 |
"trial_name": null,
|
170 |
"trial_params": null
|
171 |
}
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 6520
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1e0430645b73c3ab0a8f9201865b6bc0ad3d630c39c2224f820bdb6c9006a4db
|
3 |
size 6520
|