End of training
Browse files- README.md +14 -2
- all_results.json +16 -0
- eval_results.json +10 -0
- tokenizer.json +1 -6
- train_results.json +9 -0
- trainer_state.json +686 -0
README.md
CHANGED
@@ -3,11 +3,23 @@ license: other
|
|
3 |
base_model: Qwen/Qwen1.5-4B
|
4 |
tags:
|
5 |
- generated_from_trainer
|
|
|
|
|
6 |
metrics:
|
7 |
- accuracy
|
8 |
model-index:
|
9 |
- name: lmind_nq_train6000_eval6489_v1_docidx_v3_Qwen_Qwen1.5-4B_5e-5_lora2
|
10 |
-
results:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
library_name: peft
|
12 |
---
|
13 |
|
@@ -16,7 +28,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
16 |
|
17 |
# lmind_nq_train6000_eval6489_v1_docidx_v3_Qwen_Qwen1.5-4B_5e-5_lora2
|
18 |
|
19 |
-
This model is a fine-tuned version of [Qwen/Qwen1.5-4B](https://huggingface.co/Qwen/Qwen1.5-4B) on
|
20 |
It achieves the following results on the evaluation set:
|
21 |
- Loss: 5.0355
|
22 |
- Accuracy: 0.4273
|
|
|
3 |
base_model: Qwen/Qwen1.5-4B
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
+
datasets:
|
7 |
+
- tyzhu/lmind_nq_train6000_eval6489_v1_docidx_v3
|
8 |
metrics:
|
9 |
- accuracy
|
10 |
model-index:
|
11 |
- name: lmind_nq_train6000_eval6489_v1_docidx_v3_Qwen_Qwen1.5-4B_5e-5_lora2
|
12 |
+
results:
|
13 |
+
- task:
|
14 |
+
name: Causal Language Modeling
|
15 |
+
type: text-generation
|
16 |
+
dataset:
|
17 |
+
name: tyzhu/lmind_nq_train6000_eval6489_v1_docidx_v3
|
18 |
+
type: tyzhu/lmind_nq_train6000_eval6489_v1_docidx_v3
|
19 |
+
metrics:
|
20 |
+
- name: Accuracy
|
21 |
+
type: accuracy
|
22 |
+
value: 0.42728205128205127
|
23 |
library_name: peft
|
24 |
---
|
25 |
|
|
|
28 |
|
29 |
# lmind_nq_train6000_eval6489_v1_docidx_v3_Qwen_Qwen1.5-4B_5e-5_lora2
|
30 |
|
31 |
+
This model is a fine-tuned version of [Qwen/Qwen1.5-4B](https://huggingface.co/Qwen/Qwen1.5-4B) on the tyzhu/lmind_nq_train6000_eval6489_v1_docidx_v3 dataset.
|
32 |
It achieves the following results on the evaluation set:
|
33 |
- Loss: 5.0355
|
34 |
- Accuracy: 0.4273
|
all_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 19.970717423133237,
|
3 |
+
"eval_accuracy": 0.42728205128205127,
|
4 |
+
"eval_loss": 5.035463809967041,
|
5 |
+
"eval_runtime": 5.6207,
|
6 |
+
"eval_samples": 500,
|
7 |
+
"eval_samples_per_second": 88.957,
|
8 |
+
"eval_steps_per_second": 11.209,
|
9 |
+
"perplexity": 153.77089654928625,
|
10 |
+
"total_flos": 5.856471132500132e+17,
|
11 |
+
"train_loss": 1.188865839770812,
|
12 |
+
"train_runtime": 15402.4164,
|
13 |
+
"train_samples": 10925,
|
14 |
+
"train_samples_per_second": 14.186,
|
15 |
+
"train_steps_per_second": 0.443
|
16 |
+
}
|
eval_results.json
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 19.970717423133237,
|
3 |
+
"eval_accuracy": 0.42728205128205127,
|
4 |
+
"eval_loss": 5.035463809967041,
|
5 |
+
"eval_runtime": 5.6207,
|
6 |
+
"eval_samples": 500,
|
7 |
+
"eval_samples_per_second": 88.957,
|
8 |
+
"eval_steps_per_second": 11.209,
|
9 |
+
"perplexity": 153.77089654928625
|
10 |
+
}
|
tokenizer.json
CHANGED
@@ -1,11 +1,6 @@
|
|
1 |
{
|
2 |
"version": "1.0",
|
3 |
-
"truncation":
|
4 |
-
"direction": "Right",
|
5 |
-
"max_length": 1024,
|
6 |
-
"strategy": "LongestFirst",
|
7 |
-
"stride": 0
|
8 |
-
},
|
9 |
"padding": null,
|
10 |
"added_tokens": [
|
11 |
{
|
|
|
1 |
{
|
2 |
"version": "1.0",
|
3 |
+
"truncation": null,
|
|
|
|
|
|
|
|
|
|
|
4 |
"padding": null,
|
5 |
"added_tokens": [
|
6 |
{
|
train_results.json
ADDED
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 19.970717423133237,
|
3 |
+
"total_flos": 5.856471132500132e+17,
|
4 |
+
"train_loss": 1.188865839770812,
|
5 |
+
"train_runtime": 15402.4164,
|
6 |
+
"train_samples": 10925,
|
7 |
+
"train_samples_per_second": 14.186,
|
8 |
+
"train_steps_per_second": 0.443
|
9 |
+
}
|
trainer_state.json
ADDED
@@ -0,0 +1,686 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": null,
|
3 |
+
"best_model_checkpoint": null,
|
4 |
+
"epoch": 19.970717423133237,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 6820,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.29282576866764276,
|
13 |
+
"grad_norm": 0.25126561522483826,
|
14 |
+
"learning_rate": 5e-05,
|
15 |
+
"loss": 2.0212,
|
16 |
+
"step": 100
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.5856515373352855,
|
20 |
+
"grad_norm": 0.2594548463821411,
|
21 |
+
"learning_rate": 5e-05,
|
22 |
+
"loss": 1.9601,
|
23 |
+
"step": 200
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.8784773060029283,
|
27 |
+
"grad_norm": 0.267731249332428,
|
28 |
+
"learning_rate": 5e-05,
|
29 |
+
"loss": 1.9626,
|
30 |
+
"step": 300
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.9985358711566618,
|
34 |
+
"eval_accuracy": 0.4726666666666667,
|
35 |
+
"eval_loss": 2.9919469356536865,
|
36 |
+
"eval_runtime": 5.796,
|
37 |
+
"eval_samples_per_second": 86.266,
|
38 |
+
"eval_steps_per_second": 10.87,
|
39 |
+
"step": 341
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"epoch": 1.171303074670571,
|
43 |
+
"grad_norm": 0.3525784909725189,
|
44 |
+
"learning_rate": 5e-05,
|
45 |
+
"loss": 1.9477,
|
46 |
+
"step": 400
|
47 |
+
},
|
48 |
+
{
|
49 |
+
"epoch": 1.4641288433382138,
|
50 |
+
"grad_norm": 0.3649675250053406,
|
51 |
+
"learning_rate": 5e-05,
|
52 |
+
"loss": 1.9044,
|
53 |
+
"step": 500
|
54 |
+
},
|
55 |
+
{
|
56 |
+
"epoch": 1.7569546120058566,
|
57 |
+
"grad_norm": 0.4195611774921417,
|
58 |
+
"learning_rate": 5e-05,
|
59 |
+
"loss": 1.9158,
|
60 |
+
"step": 600
|
61 |
+
},
|
62 |
+
{
|
63 |
+
"epoch": 2.0,
|
64 |
+
"eval_accuracy": 0.47374358974358977,
|
65 |
+
"eval_loss": 2.9863882064819336,
|
66 |
+
"eval_runtime": 5.61,
|
67 |
+
"eval_samples_per_second": 89.126,
|
68 |
+
"eval_steps_per_second": 11.23,
|
69 |
+
"step": 683
|
70 |
+
},
|
71 |
+
{
|
72 |
+
"epoch": 2.049780380673499,
|
73 |
+
"grad_norm": 0.41343557834625244,
|
74 |
+
"learning_rate": 5e-05,
|
75 |
+
"loss": 1.9046,
|
76 |
+
"step": 700
|
77 |
+
},
|
78 |
+
{
|
79 |
+
"epoch": 2.342606149341142,
|
80 |
+
"grad_norm": 0.49672290682792664,
|
81 |
+
"learning_rate": 5e-05,
|
82 |
+
"loss": 1.8367,
|
83 |
+
"step": 800
|
84 |
+
},
|
85 |
+
{
|
86 |
+
"epoch": 2.6354319180087846,
|
87 |
+
"grad_norm": 0.5728709697723389,
|
88 |
+
"learning_rate": 5e-05,
|
89 |
+
"loss": 1.8573,
|
90 |
+
"step": 900
|
91 |
+
},
|
92 |
+
{
|
93 |
+
"epoch": 2.9282576866764276,
|
94 |
+
"grad_norm": 0.5571523904800415,
|
95 |
+
"learning_rate": 5e-05,
|
96 |
+
"loss": 1.8622,
|
97 |
+
"step": 1000
|
98 |
+
},
|
99 |
+
{
|
100 |
+
"epoch": 2.998535871156662,
|
101 |
+
"eval_accuracy": 0.47102564102564104,
|
102 |
+
"eval_loss": 3.0419609546661377,
|
103 |
+
"eval_runtime": 5.7173,
|
104 |
+
"eval_samples_per_second": 87.454,
|
105 |
+
"eval_steps_per_second": 11.019,
|
106 |
+
"step": 1024
|
107 |
+
},
|
108 |
+
{
|
109 |
+
"epoch": 3.22108345534407,
|
110 |
+
"grad_norm": 0.5878937840461731,
|
111 |
+
"learning_rate": 5e-05,
|
112 |
+
"loss": 1.7839,
|
113 |
+
"step": 1100
|
114 |
+
},
|
115 |
+
{
|
116 |
+
"epoch": 3.513909224011713,
|
117 |
+
"grad_norm": 0.6718343496322632,
|
118 |
+
"learning_rate": 5e-05,
|
119 |
+
"loss": 1.7601,
|
120 |
+
"step": 1200
|
121 |
+
},
|
122 |
+
{
|
123 |
+
"epoch": 3.8067349926793557,
|
124 |
+
"grad_norm": 0.7238633632659912,
|
125 |
+
"learning_rate": 5e-05,
|
126 |
+
"loss": 1.786,
|
127 |
+
"step": 1300
|
128 |
+
},
|
129 |
+
{
|
130 |
+
"epoch": 4.0,
|
131 |
+
"eval_accuracy": 0.46615384615384614,
|
132 |
+
"eval_loss": 3.1526739597320557,
|
133 |
+
"eval_runtime": 5.5502,
|
134 |
+
"eval_samples_per_second": 90.087,
|
135 |
+
"eval_steps_per_second": 11.351,
|
136 |
+
"step": 1366
|
137 |
+
},
|
138 |
+
{
|
139 |
+
"epoch": 4.099560761346998,
|
140 |
+
"grad_norm": 0.8255831599235535,
|
141 |
+
"learning_rate": 5e-05,
|
142 |
+
"loss": 1.7459,
|
143 |
+
"step": 1400
|
144 |
+
},
|
145 |
+
{
|
146 |
+
"epoch": 4.392386530014641,
|
147 |
+
"grad_norm": 0.7857334613800049,
|
148 |
+
"learning_rate": 5e-05,
|
149 |
+
"loss": 1.6799,
|
150 |
+
"step": 1500
|
151 |
+
},
|
152 |
+
{
|
153 |
+
"epoch": 4.685212298682284,
|
154 |
+
"grad_norm": 0.8550590872764587,
|
155 |
+
"learning_rate": 5e-05,
|
156 |
+
"loss": 1.68,
|
157 |
+
"step": 1600
|
158 |
+
},
|
159 |
+
{
|
160 |
+
"epoch": 4.978038067349927,
|
161 |
+
"grad_norm": 0.9140918254852295,
|
162 |
+
"learning_rate": 5e-05,
|
163 |
+
"loss": 1.7019,
|
164 |
+
"step": 1700
|
165 |
+
},
|
166 |
+
{
|
167 |
+
"epoch": 4.998535871156662,
|
168 |
+
"eval_accuracy": 0.4634358974358974,
|
169 |
+
"eval_loss": 3.381866693496704,
|
170 |
+
"eval_runtime": 5.9878,
|
171 |
+
"eval_samples_per_second": 83.503,
|
172 |
+
"eval_steps_per_second": 10.521,
|
173 |
+
"step": 1707
|
174 |
+
},
|
175 |
+
{
|
176 |
+
"epoch": 5.270863836017569,
|
177 |
+
"grad_norm": 0.958633542060852,
|
178 |
+
"learning_rate": 5e-05,
|
179 |
+
"loss": 1.6064,
|
180 |
+
"step": 1800
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"epoch": 5.563689604685212,
|
184 |
+
"grad_norm": 0.9240352511405945,
|
185 |
+
"learning_rate": 5e-05,
|
186 |
+
"loss": 1.5911,
|
187 |
+
"step": 1900
|
188 |
+
},
|
189 |
+
{
|
190 |
+
"epoch": 5.856515373352855,
|
191 |
+
"grad_norm": 0.971266508102417,
|
192 |
+
"learning_rate": 5e-05,
|
193 |
+
"loss": 1.6036,
|
194 |
+
"step": 2000
|
195 |
+
},
|
196 |
+
{
|
197 |
+
"epoch": 6.0,
|
198 |
+
"eval_accuracy": 0.45887179487179486,
|
199 |
+
"eval_loss": 3.4968838691711426,
|
200 |
+
"eval_runtime": 5.9643,
|
201 |
+
"eval_samples_per_second": 83.832,
|
202 |
+
"eval_steps_per_second": 10.563,
|
203 |
+
"step": 2049
|
204 |
+
},
|
205 |
+
{
|
206 |
+
"epoch": 6.149341142020498,
|
207 |
+
"grad_norm": 1.107942819595337,
|
208 |
+
"learning_rate": 5e-05,
|
209 |
+
"loss": 1.5305,
|
210 |
+
"step": 2100
|
211 |
+
},
|
212 |
+
{
|
213 |
+
"epoch": 6.44216691068814,
|
214 |
+
"grad_norm": 1.133771300315857,
|
215 |
+
"learning_rate": 5e-05,
|
216 |
+
"loss": 1.4897,
|
217 |
+
"step": 2200
|
218 |
+
},
|
219 |
+
{
|
220 |
+
"epoch": 6.734992679355783,
|
221 |
+
"grad_norm": 1.037837266921997,
|
222 |
+
"learning_rate": 5e-05,
|
223 |
+
"loss": 1.5175,
|
224 |
+
"step": 2300
|
225 |
+
},
|
226 |
+
{
|
227 |
+
"epoch": 6.998535871156662,
|
228 |
+
"eval_accuracy": 0.45774358974358975,
|
229 |
+
"eval_loss": 3.64119553565979,
|
230 |
+
"eval_runtime": 5.7326,
|
231 |
+
"eval_samples_per_second": 87.22,
|
232 |
+
"eval_steps_per_second": 10.99,
|
233 |
+
"step": 2390
|
234 |
+
},
|
235 |
+
{
|
236 |
+
"epoch": 7.027818448023426,
|
237 |
+
"grad_norm": 1.2089684009552002,
|
238 |
+
"learning_rate": 5e-05,
|
239 |
+
"loss": 1.4929,
|
240 |
+
"step": 2400
|
241 |
+
},
|
242 |
+
{
|
243 |
+
"epoch": 7.320644216691068,
|
244 |
+
"grad_norm": 1.2794983386993408,
|
245 |
+
"learning_rate": 5e-05,
|
246 |
+
"loss": 1.3851,
|
247 |
+
"step": 2500
|
248 |
+
},
|
249 |
+
{
|
250 |
+
"epoch": 7.613469985358711,
|
251 |
+
"grad_norm": 1.2079399824142456,
|
252 |
+
"learning_rate": 5e-05,
|
253 |
+
"loss": 1.4075,
|
254 |
+
"step": 2600
|
255 |
+
},
|
256 |
+
{
|
257 |
+
"epoch": 7.906295754026354,
|
258 |
+
"grad_norm": 1.1294775009155273,
|
259 |
+
"learning_rate": 5e-05,
|
260 |
+
"loss": 1.4007,
|
261 |
+
"step": 2700
|
262 |
+
},
|
263 |
+
{
|
264 |
+
"epoch": 8.0,
|
265 |
+
"eval_accuracy": 0.45374358974358975,
|
266 |
+
"eval_loss": 3.831012487411499,
|
267 |
+
"eval_runtime": 5.871,
|
268 |
+
"eval_samples_per_second": 85.164,
|
269 |
+
"eval_steps_per_second": 10.731,
|
270 |
+
"step": 2732
|
271 |
+
},
|
272 |
+
{
|
273 |
+
"epoch": 8.199121522693996,
|
274 |
+
"grad_norm": 1.3208889961242676,
|
275 |
+
"learning_rate": 5e-05,
|
276 |
+
"loss": 1.3076,
|
277 |
+
"step": 2800
|
278 |
+
},
|
279 |
+
{
|
280 |
+
"epoch": 8.49194729136164,
|
281 |
+
"grad_norm": 1.345077395439148,
|
282 |
+
"learning_rate": 5e-05,
|
283 |
+
"loss": 1.2867,
|
284 |
+
"step": 2900
|
285 |
+
},
|
286 |
+
{
|
287 |
+
"epoch": 8.784773060029282,
|
288 |
+
"grad_norm": 1.4141809940338135,
|
289 |
+
"learning_rate": 5e-05,
|
290 |
+
"loss": 1.326,
|
291 |
+
"step": 3000
|
292 |
+
},
|
293 |
+
{
|
294 |
+
"epoch": 8.998535871156662,
|
295 |
+
"eval_accuracy": 0.44866666666666666,
|
296 |
+
"eval_loss": 3.9176931381225586,
|
297 |
+
"eval_runtime": 5.7802,
|
298 |
+
"eval_samples_per_second": 86.502,
|
299 |
+
"eval_steps_per_second": 10.899,
|
300 |
+
"step": 3073
|
301 |
+
},
|
302 |
+
{
|
303 |
+
"epoch": 9.077598828696924,
|
304 |
+
"grad_norm": 1.3622578382492065,
|
305 |
+
"learning_rate": 5e-05,
|
306 |
+
"loss": 1.2725,
|
307 |
+
"step": 3100
|
308 |
+
},
|
309 |
+
{
|
310 |
+
"epoch": 9.370424597364568,
|
311 |
+
"grad_norm": 1.632681965827942,
|
312 |
+
"learning_rate": 5e-05,
|
313 |
+
"loss": 1.1894,
|
314 |
+
"step": 3200
|
315 |
+
},
|
316 |
+
{
|
317 |
+
"epoch": 9.66325036603221,
|
318 |
+
"grad_norm": 1.5909069776535034,
|
319 |
+
"learning_rate": 5e-05,
|
320 |
+
"loss": 1.1959,
|
321 |
+
"step": 3300
|
322 |
+
},
|
323 |
+
{
|
324 |
+
"epoch": 9.956076134699854,
|
325 |
+
"grad_norm": 1.5544390678405762,
|
326 |
+
"learning_rate": 5e-05,
|
327 |
+
"loss": 1.231,
|
328 |
+
"step": 3400
|
329 |
+
},
|
330 |
+
{
|
331 |
+
"epoch": 10.0,
|
332 |
+
"eval_accuracy": 0.4450769230769231,
|
333 |
+
"eval_loss": 4.066500186920166,
|
334 |
+
"eval_runtime": 5.7535,
|
335 |
+
"eval_samples_per_second": 86.904,
|
336 |
+
"eval_steps_per_second": 10.95,
|
337 |
+
"step": 3415
|
338 |
+
},
|
339 |
+
{
|
340 |
+
"epoch": 10.248901903367496,
|
341 |
+
"grad_norm": 1.678530216217041,
|
342 |
+
"learning_rate": 5e-05,
|
343 |
+
"loss": 1.1034,
|
344 |
+
"step": 3500
|
345 |
+
},
|
346 |
+
{
|
347 |
+
"epoch": 10.541727672035138,
|
348 |
+
"grad_norm": 1.7260808944702148,
|
349 |
+
"learning_rate": 5e-05,
|
350 |
+
"loss": 1.0997,
|
351 |
+
"step": 3600
|
352 |
+
},
|
353 |
+
{
|
354 |
+
"epoch": 10.834553440702782,
|
355 |
+
"grad_norm": 1.8344779014587402,
|
356 |
+
"learning_rate": 5e-05,
|
357 |
+
"loss": 1.1298,
|
358 |
+
"step": 3700
|
359 |
+
},
|
360 |
+
{
|
361 |
+
"epoch": 10.998535871156662,
|
362 |
+
"eval_accuracy": 0.44,
|
363 |
+
"eval_loss": 4.177348613739014,
|
364 |
+
"eval_runtime": 5.7468,
|
365 |
+
"eval_samples_per_second": 87.005,
|
366 |
+
"eval_steps_per_second": 10.963,
|
367 |
+
"step": 3756
|
368 |
+
},
|
369 |
+
{
|
370 |
+
"epoch": 11.127379209370424,
|
371 |
+
"grad_norm": 1.9434887170791626,
|
372 |
+
"learning_rate": 5e-05,
|
373 |
+
"loss": 1.0734,
|
374 |
+
"step": 3800
|
375 |
+
},
|
376 |
+
{
|
377 |
+
"epoch": 11.420204978038067,
|
378 |
+
"grad_norm": 1.9285024404525757,
|
379 |
+
"learning_rate": 5e-05,
|
380 |
+
"loss": 1.0046,
|
381 |
+
"step": 3900
|
382 |
+
},
|
383 |
+
{
|
384 |
+
"epoch": 11.71303074670571,
|
385 |
+
"grad_norm": 1.9000359773635864,
|
386 |
+
"learning_rate": 5e-05,
|
387 |
+
"loss": 1.0276,
|
388 |
+
"step": 4000
|
389 |
+
},
|
390 |
+
{
|
391 |
+
"epoch": 12.0,
|
392 |
+
"eval_accuracy": 0.43784615384615383,
|
393 |
+
"eval_loss": 4.287516117095947,
|
394 |
+
"eval_runtime": 5.6257,
|
395 |
+
"eval_samples_per_second": 88.878,
|
396 |
+
"eval_steps_per_second": 11.199,
|
397 |
+
"step": 4098
|
398 |
+
},
|
399 |
+
{
|
400 |
+
"epoch": 12.005856515373353,
|
401 |
+
"grad_norm": 1.707350492477417,
|
402 |
+
"learning_rate": 5e-05,
|
403 |
+
"loss": 1.0392,
|
404 |
+
"step": 4100
|
405 |
+
},
|
406 |
+
{
|
407 |
+
"epoch": 12.298682284040996,
|
408 |
+
"grad_norm": 1.9523829221725464,
|
409 |
+
"learning_rate": 5e-05,
|
410 |
+
"loss": 0.9029,
|
411 |
+
"step": 4200
|
412 |
+
},
|
413 |
+
{
|
414 |
+
"epoch": 12.591508052708638,
|
415 |
+
"grad_norm": 1.9846230745315552,
|
416 |
+
"learning_rate": 5e-05,
|
417 |
+
"loss": 0.9324,
|
418 |
+
"step": 4300
|
419 |
+
},
|
420 |
+
{
|
421 |
+
"epoch": 12.88433382137628,
|
422 |
+
"grad_norm": 2.4833240509033203,
|
423 |
+
"learning_rate": 5e-05,
|
424 |
+
"loss": 0.9525,
|
425 |
+
"step": 4400
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"epoch": 12.998535871156662,
|
429 |
+
"eval_accuracy": 0.4351794871794872,
|
430 |
+
"eval_loss": 4.42730188369751,
|
431 |
+
"eval_runtime": 5.835,
|
432 |
+
"eval_samples_per_second": 85.69,
|
433 |
+
"eval_steps_per_second": 10.797,
|
434 |
+
"step": 4439
|
435 |
+
},
|
436 |
+
{
|
437 |
+
"epoch": 13.177159590043924,
|
438 |
+
"grad_norm": 1.8055790662765503,
|
439 |
+
"learning_rate": 5e-05,
|
440 |
+
"loss": 0.8664,
|
441 |
+
"step": 4500
|
442 |
+
},
|
443 |
+
{
|
444 |
+
"epoch": 13.469985358711567,
|
445 |
+
"grad_norm": 2.5267982482910156,
|
446 |
+
"learning_rate": 5e-05,
|
447 |
+
"loss": 0.8457,
|
448 |
+
"step": 4600
|
449 |
+
},
|
450 |
+
{
|
451 |
+
"epoch": 13.762811127379209,
|
452 |
+
"grad_norm": 2.2086236476898193,
|
453 |
+
"learning_rate": 5e-05,
|
454 |
+
"loss": 0.8616,
|
455 |
+
"step": 4700
|
456 |
+
},
|
457 |
+
{
|
458 |
+
"epoch": 14.0,
|
459 |
+
"eval_accuracy": 0.43241025641025643,
|
460 |
+
"eval_loss": 4.44843053817749,
|
461 |
+
"eval_runtime": 5.7935,
|
462 |
+
"eval_samples_per_second": 86.303,
|
463 |
+
"eval_steps_per_second": 10.874,
|
464 |
+
"step": 4781
|
465 |
+
},
|
466 |
+
{
|
467 |
+
"epoch": 14.055636896046853,
|
468 |
+
"grad_norm": 2.44354510307312,
|
469 |
+
"learning_rate": 5e-05,
|
470 |
+
"loss": 0.8402,
|
471 |
+
"step": 4800
|
472 |
+
},
|
473 |
+
{
|
474 |
+
"epoch": 14.348462664714495,
|
475 |
+
"grad_norm": 2.1125502586364746,
|
476 |
+
"learning_rate": 5e-05,
|
477 |
+
"loss": 0.7608,
|
478 |
+
"step": 4900
|
479 |
+
},
|
480 |
+
{
|
481 |
+
"epoch": 14.641288433382137,
|
482 |
+
"grad_norm": 2.1573476791381836,
|
483 |
+
"learning_rate": 5e-05,
|
484 |
+
"loss": 0.7736,
|
485 |
+
"step": 5000
|
486 |
+
},
|
487 |
+
{
|
488 |
+
"epoch": 14.93411420204978,
|
489 |
+
"grad_norm": 2.4622254371643066,
|
490 |
+
"learning_rate": 5e-05,
|
491 |
+
"loss": 0.7799,
|
492 |
+
"step": 5100
|
493 |
+
},
|
494 |
+
{
|
495 |
+
"epoch": 14.998535871156662,
|
496 |
+
"eval_accuracy": 0.43133333333333335,
|
497 |
+
"eval_loss": 4.622844219207764,
|
498 |
+
"eval_runtime": 5.87,
|
499 |
+
"eval_samples_per_second": 85.178,
|
500 |
+
"eval_steps_per_second": 10.732,
|
501 |
+
"step": 5122
|
502 |
+
},
|
503 |
+
{
|
504 |
+
"epoch": 15.226939970717423,
|
505 |
+
"grad_norm": 2.4932661056518555,
|
506 |
+
"learning_rate": 5e-05,
|
507 |
+
"loss": 0.7057,
|
508 |
+
"step": 5200
|
509 |
+
},
|
510 |
+
{
|
511 |
+
"epoch": 15.519765739385067,
|
512 |
+
"grad_norm": 2.3488550186157227,
|
513 |
+
"learning_rate": 5e-05,
|
514 |
+
"loss": 0.686,
|
515 |
+
"step": 5300
|
516 |
+
},
|
517 |
+
{
|
518 |
+
"epoch": 15.812591508052709,
|
519 |
+
"grad_norm": 2.2296109199523926,
|
520 |
+
"learning_rate": 5e-05,
|
521 |
+
"loss": 0.7084,
|
522 |
+
"step": 5400
|
523 |
+
},
|
524 |
+
{
|
525 |
+
"epoch": 16.0,
|
526 |
+
"eval_accuracy": 0.4303076923076923,
|
527 |
+
"eval_loss": 4.723867416381836,
|
528 |
+
"eval_runtime": 5.6093,
|
529 |
+
"eval_samples_per_second": 89.137,
|
530 |
+
"eval_steps_per_second": 11.231,
|
531 |
+
"step": 5464
|
532 |
+
},
|
533 |
+
{
|
534 |
+
"epoch": 16.105417276720353,
|
535 |
+
"grad_norm": 2.141963243484497,
|
536 |
+
"learning_rate": 5e-05,
|
537 |
+
"loss": 0.6755,
|
538 |
+
"step": 5500
|
539 |
+
},
|
540 |
+
{
|
541 |
+
"epoch": 16.398243045387993,
|
542 |
+
"grad_norm": 2.679950475692749,
|
543 |
+
"learning_rate": 5e-05,
|
544 |
+
"loss": 0.6229,
|
545 |
+
"step": 5600
|
546 |
+
},
|
547 |
+
{
|
548 |
+
"epoch": 16.691068814055637,
|
549 |
+
"grad_norm": 2.506779193878174,
|
550 |
+
"learning_rate": 5e-05,
|
551 |
+
"loss": 0.6377,
|
552 |
+
"step": 5700
|
553 |
+
},
|
554 |
+
{
|
555 |
+
"epoch": 16.98389458272328,
|
556 |
+
"grad_norm": 3.019353151321411,
|
557 |
+
"learning_rate": 5e-05,
|
558 |
+
"loss": 0.6478,
|
559 |
+
"step": 5800
|
560 |
+
},
|
561 |
+
{
|
562 |
+
"epoch": 16.998535871156662,
|
563 |
+
"eval_accuracy": 0.430974358974359,
|
564 |
+
"eval_loss": 4.816666126251221,
|
565 |
+
"eval_runtime": 5.9538,
|
566 |
+
"eval_samples_per_second": 83.98,
|
567 |
+
"eval_steps_per_second": 10.581,
|
568 |
+
"step": 5805
|
569 |
+
},
|
570 |
+
{
|
571 |
+
"epoch": 17.27672035139092,
|
572 |
+
"grad_norm": 2.6447300910949707,
|
573 |
+
"learning_rate": 5e-05,
|
574 |
+
"loss": 0.5511,
|
575 |
+
"step": 5900
|
576 |
+
},
|
577 |
+
{
|
578 |
+
"epoch": 17.569546120058565,
|
579 |
+
"grad_norm": 2.720766067504883,
|
580 |
+
"learning_rate": 5e-05,
|
581 |
+
"loss": 0.5638,
|
582 |
+
"step": 6000
|
583 |
+
},
|
584 |
+
{
|
585 |
+
"epoch": 17.86237188872621,
|
586 |
+
"grad_norm": 2.4070911407470703,
|
587 |
+
"learning_rate": 5e-05,
|
588 |
+
"loss": 0.5862,
|
589 |
+
"step": 6100
|
590 |
+
},
|
591 |
+
{
|
592 |
+
"epoch": 18.0,
|
593 |
+
"eval_accuracy": 0.43025641025641026,
|
594 |
+
"eval_loss": 4.85101842880249,
|
595 |
+
"eval_runtime": 5.6529,
|
596 |
+
"eval_samples_per_second": 88.45,
|
597 |
+
"eval_steps_per_second": 11.145,
|
598 |
+
"step": 6147
|
599 |
+
},
|
600 |
+
{
|
601 |
+
"epoch": 18.15519765739385,
|
602 |
+
"grad_norm": 3.0029456615448,
|
603 |
+
"learning_rate": 5e-05,
|
604 |
+
"loss": 0.5335,
|
605 |
+
"step": 6200
|
606 |
+
},
|
607 |
+
{
|
608 |
+
"epoch": 18.448023426061493,
|
609 |
+
"grad_norm": 2.6631345748901367,
|
610 |
+
"learning_rate": 5e-05,
|
611 |
+
"loss": 0.5018,
|
612 |
+
"step": 6300
|
613 |
+
},
|
614 |
+
{
|
615 |
+
"epoch": 18.740849194729137,
|
616 |
+
"grad_norm": 2.603055477142334,
|
617 |
+
"learning_rate": 5e-05,
|
618 |
+
"loss": 0.5189,
|
619 |
+
"step": 6400
|
620 |
+
},
|
621 |
+
{
|
622 |
+
"epoch": 18.998535871156662,
|
623 |
+
"eval_accuracy": 0.42425641025641025,
|
624 |
+
"eval_loss": 4.926539897918701,
|
625 |
+
"eval_runtime": 5.8773,
|
626 |
+
"eval_samples_per_second": 85.073,
|
627 |
+
"eval_steps_per_second": 10.719,
|
628 |
+
"step": 6488
|
629 |
+
},
|
630 |
+
{
|
631 |
+
"epoch": 19.03367496339678,
|
632 |
+
"grad_norm": 2.3171300888061523,
|
633 |
+
"learning_rate": 5e-05,
|
634 |
+
"loss": 0.5225,
|
635 |
+
"step": 6500
|
636 |
+
},
|
637 |
+
{
|
638 |
+
"epoch": 19.32650073206442,
|
639 |
+
"grad_norm": 3.008113384246826,
|
640 |
+
"learning_rate": 5e-05,
|
641 |
+
"loss": 0.4485,
|
642 |
+
"step": 6600
|
643 |
+
},
|
644 |
+
{
|
645 |
+
"epoch": 19.619326500732065,
|
646 |
+
"grad_norm": 2.5204029083251953,
|
647 |
+
"learning_rate": 5e-05,
|
648 |
+
"loss": 0.4623,
|
649 |
+
"step": 6700
|
650 |
+
},
|
651 |
+
{
|
652 |
+
"epoch": 19.91215226939971,
|
653 |
+
"grad_norm": 2.9562718868255615,
|
654 |
+
"learning_rate": 5e-05,
|
655 |
+
"loss": 0.4767,
|
656 |
+
"step": 6800
|
657 |
+
},
|
658 |
+
{
|
659 |
+
"epoch": 19.970717423133237,
|
660 |
+
"eval_accuracy": 0.42728205128205127,
|
661 |
+
"eval_loss": 5.035463809967041,
|
662 |
+
"eval_runtime": 5.7429,
|
663 |
+
"eval_samples_per_second": 87.064,
|
664 |
+
"eval_steps_per_second": 10.97,
|
665 |
+
"step": 6820
|
666 |
+
},
|
667 |
+
{
|
668 |
+
"epoch": 19.970717423133237,
|
669 |
+
"step": 6820,
|
670 |
+
"total_flos": 5.856471132500132e+17,
|
671 |
+
"train_loss": 1.188865839770812,
|
672 |
+
"train_runtime": 15402.4164,
|
673 |
+
"train_samples_per_second": 14.186,
|
674 |
+
"train_steps_per_second": 0.443
|
675 |
+
}
|
676 |
+
],
|
677 |
+
"logging_steps": 100,
|
678 |
+
"max_steps": 6820,
|
679 |
+
"num_input_tokens_seen": 0,
|
680 |
+
"num_train_epochs": 20,
|
681 |
+
"save_steps": 500,
|
682 |
+
"total_flos": 5.856471132500132e+17,
|
683 |
+
"train_batch_size": 1,
|
684 |
+
"trial_name": null,
|
685 |
+
"trial_params": null
|
686 |
+
}
|