Upload folder using huggingface_hub
Browse files- checkpoint-10945/config.json +228 -0
- checkpoint-10945/model.safetensors +3 -0
- checkpoint-10945/optimizer.pt +3 -0
- checkpoint-10945/preprocessor_config.json +36 -0
- checkpoint-10945/rng_state.pth +3 -0
- checkpoint-10945/scheduler.pt +3 -0
- checkpoint-10945/trainer_state.json +177 -0
- checkpoint-10945/training_args.bin +3 -0
- model.safetensors +1 -1
- training_args.bin +1 -1
checkpoint-10945/config.json
ADDED
@@ -0,0 +1,228 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "dima806/mushrooms_image_detection",
|
3 |
+
"architectures": [
|
4 |
+
"ViTForImageClassification"
|
5 |
+
],
|
6 |
+
"attention_probs_dropout_prob": 0.0,
|
7 |
+
"encoder_stride": 16,
|
8 |
+
"hidden_act": "gelu",
|
9 |
+
"hidden_dropout_prob": 0.0,
|
10 |
+
"hidden_size": 768,
|
11 |
+
"id2label": {
|
12 |
+
"0": "Urnula craterium",
|
13 |
+
"1": "Leccinum albostipitatum",
|
14 |
+
"2": "Lactarius deliciosus",
|
15 |
+
"3": "Clitocybe nebularis",
|
16 |
+
"4": "Hypholoma fasciculare",
|
17 |
+
"5": "Lactarius torminosus",
|
18 |
+
"6": "Lycoperdon perlatum",
|
19 |
+
"7": "Verpa bohemica",
|
20 |
+
"8": "Schizophyllum commune",
|
21 |
+
"9": "Leccinum aurantiacum",
|
22 |
+
"10": "Phellinus igniarius",
|
23 |
+
"11": "Suillus luteus",
|
24 |
+
"12": "Coltricia perennis",
|
25 |
+
"13": "Cetraria islandica",
|
26 |
+
"14": "Amanita muscaria",
|
27 |
+
"15": "Pholiota aurivella",
|
28 |
+
"16": "Trichaptum biforme",
|
29 |
+
"17": "Artomyces pyxidatus",
|
30 |
+
"18": "Calocera viscosa",
|
31 |
+
"19": "Sarcosoma globosum",
|
32 |
+
"20": "Evernia prunastri",
|
33 |
+
"21": "Laetiporus sulphureus",
|
34 |
+
"22": "Lobaria pulmonaria",
|
35 |
+
"23": "Bjerkandera adusta",
|
36 |
+
"24": "Vulpicida pinastri",
|
37 |
+
"25": "Imleria badia",
|
38 |
+
"26": "Evernia mesomorpha",
|
39 |
+
"27": "Physcia adscendens",
|
40 |
+
"28": "Coprinellus micaceus",
|
41 |
+
"29": "Armillaria borealis",
|
42 |
+
"30": "Trametes ochracea",
|
43 |
+
"31": "Cantharellus cibarius",
|
44 |
+
"32": "Pseudevernia furfuracea",
|
45 |
+
"33": "Tremella mesenterica",
|
46 |
+
"34": "Gyromitra infula",
|
47 |
+
"35": "Leccinum versipelle",
|
48 |
+
"36": "Mutinus ravenelii",
|
49 |
+
"37": "Pholiota squarrosa",
|
50 |
+
"38": "Amanita rubescens",
|
51 |
+
"39": "Amanita pantherina",
|
52 |
+
"40": "Sarcoscypha austriaca",
|
53 |
+
"41": "Boletus edulis",
|
54 |
+
"42": "Coprinus comatus",
|
55 |
+
"43": "Merulius tremellosus",
|
56 |
+
"44": "Stropharia aeruginosa",
|
57 |
+
"45": "Cladonia fimbriata",
|
58 |
+
"46": "Suillus grevillei",
|
59 |
+
"47": "Apioperdon pyriforme",
|
60 |
+
"48": "Cerioporus squamosus",
|
61 |
+
"49": "Leccinum scabrum",
|
62 |
+
"50": "Rhytisma acerinum",
|
63 |
+
"51": "Hypholoma lateritium",
|
64 |
+
"52": "Flammulina velutipes",
|
65 |
+
"53": "Tricholomopsis rutilans",
|
66 |
+
"54": "Coprinopsis atramentaria",
|
67 |
+
"55": "Trametes versicolor",
|
68 |
+
"56": "Graphis scripta",
|
69 |
+
"57": "Ganoderma applanatum",
|
70 |
+
"58": "Phellinus tremulae",
|
71 |
+
"59": "Peltigera aphthosa",
|
72 |
+
"60": "Parmelia sulcata",
|
73 |
+
"61": "Fomitopsis betulina",
|
74 |
+
"62": "Pleurotus pulmonarius",
|
75 |
+
"63": "Fomitopsis pinicola",
|
76 |
+
"64": "Daedaleopsis confragosa",
|
77 |
+
"65": "Hericium coralloides",
|
78 |
+
"66": "Trametes hirsuta",
|
79 |
+
"67": "Coprinellus disseminatus",
|
80 |
+
"68": "Kuehneromyces mutabilis",
|
81 |
+
"69": "Pleurotus ostreatus",
|
82 |
+
"70": "Phlebia radiata",
|
83 |
+
"71": "Boletus reticulatus",
|
84 |
+
"72": "Phallus impudicus",
|
85 |
+
"73": "Macrolepiota procera",
|
86 |
+
"74": "Fomes fomentarius",
|
87 |
+
"75": "Suillus granulatus",
|
88 |
+
"76": "Gyromitra esculenta",
|
89 |
+
"77": "Xanthoria parietina",
|
90 |
+
"78": "Nectria cinnabarina",
|
91 |
+
"79": "Sarcomyxa serotina",
|
92 |
+
"80": "Inonotus obliquus",
|
93 |
+
"81": "Panellus stipticus",
|
94 |
+
"82": "Hypogymnia physodes",
|
95 |
+
"83": "Hygrophoropsis aurantiaca",
|
96 |
+
"84": "Cladonia rangiferina",
|
97 |
+
"85": "Platismatia glauca",
|
98 |
+
"86": "Calycina citrina",
|
99 |
+
"87": "Cladonia stellaris",
|
100 |
+
"88": "Amanita citrina",
|
101 |
+
"89": "Lepista nuda",
|
102 |
+
"90": "Gyromitra gigas",
|
103 |
+
"91": "Crucibulum laeve",
|
104 |
+
"92": "Daedaleopsis tricolor",
|
105 |
+
"93": "Stereum hirsutum",
|
106 |
+
"94": "Paxillus involutus",
|
107 |
+
"95": "Lactarius turpis",
|
108 |
+
"96": "Chlorociboria aeruginascens",
|
109 |
+
"97": "Chondrostereum purpureum",
|
110 |
+
"98": "Phaeophyscia orbicularis",
|
111 |
+
"99": "Peltigera praetextata"
|
112 |
+
},
|
113 |
+
"image_size": 224,
|
114 |
+
"initializer_range": 0.02,
|
115 |
+
"intermediate_size": 3072,
|
116 |
+
"label2id": {
|
117 |
+
"Amanita citrina": 88,
|
118 |
+
"Amanita muscaria": 14,
|
119 |
+
"Amanita pantherina": 39,
|
120 |
+
"Amanita rubescens": 38,
|
121 |
+
"Apioperdon pyriforme": 47,
|
122 |
+
"Armillaria borealis": 29,
|
123 |
+
"Artomyces pyxidatus": 17,
|
124 |
+
"Bjerkandera adusta": 23,
|
125 |
+
"Boletus edulis": 41,
|
126 |
+
"Boletus reticulatus": 71,
|
127 |
+
"Calocera viscosa": 18,
|
128 |
+
"Calycina citrina": 86,
|
129 |
+
"Cantharellus cibarius": 31,
|
130 |
+
"Cerioporus squamosus": 48,
|
131 |
+
"Cetraria islandica": 13,
|
132 |
+
"Chlorociboria aeruginascens": 96,
|
133 |
+
"Chondrostereum purpureum": 97,
|
134 |
+
"Cladonia fimbriata": 45,
|
135 |
+
"Cladonia rangiferina": 84,
|
136 |
+
"Cladonia stellaris": 87,
|
137 |
+
"Clitocybe nebularis": 3,
|
138 |
+
"Coltricia perennis": 12,
|
139 |
+
"Coprinellus disseminatus": 67,
|
140 |
+
"Coprinellus micaceus": 28,
|
141 |
+
"Coprinopsis atramentaria": 54,
|
142 |
+
"Coprinus comatus": 42,
|
143 |
+
"Crucibulum laeve": 91,
|
144 |
+
"Daedaleopsis confragosa": 64,
|
145 |
+
"Daedaleopsis tricolor": 92,
|
146 |
+
"Evernia mesomorpha": 26,
|
147 |
+
"Evernia prunastri": 20,
|
148 |
+
"Flammulina velutipes": 52,
|
149 |
+
"Fomes fomentarius": 74,
|
150 |
+
"Fomitopsis betulina": 61,
|
151 |
+
"Fomitopsis pinicola": 63,
|
152 |
+
"Ganoderma applanatum": 57,
|
153 |
+
"Graphis scripta": 56,
|
154 |
+
"Gyromitra esculenta": 76,
|
155 |
+
"Gyromitra gigas": 90,
|
156 |
+
"Gyromitra infula": 34,
|
157 |
+
"Hericium coralloides": 65,
|
158 |
+
"Hygrophoropsis aurantiaca": 83,
|
159 |
+
"Hypholoma fasciculare": 4,
|
160 |
+
"Hypholoma lateritium": 51,
|
161 |
+
"Hypogymnia physodes": 82,
|
162 |
+
"Imleria badia": 25,
|
163 |
+
"Inonotus obliquus": 80,
|
164 |
+
"Kuehneromyces mutabilis": 68,
|
165 |
+
"Lactarius deliciosus": 2,
|
166 |
+
"Lactarius torminosus": 5,
|
167 |
+
"Lactarius turpis": 95,
|
168 |
+
"Laetiporus sulphureus": 21,
|
169 |
+
"Leccinum albostipitatum": 1,
|
170 |
+
"Leccinum aurantiacum": 9,
|
171 |
+
"Leccinum scabrum": 49,
|
172 |
+
"Leccinum versipelle": 35,
|
173 |
+
"Lepista nuda": 89,
|
174 |
+
"Lobaria pulmonaria": 22,
|
175 |
+
"Lycoperdon perlatum": 6,
|
176 |
+
"Macrolepiota procera": 73,
|
177 |
+
"Merulius tremellosus": 43,
|
178 |
+
"Mutinus ravenelii": 36,
|
179 |
+
"Nectria cinnabarina": 78,
|
180 |
+
"Panellus stipticus": 81,
|
181 |
+
"Parmelia sulcata": 60,
|
182 |
+
"Paxillus involutus": 94,
|
183 |
+
"Peltigera aphthosa": 59,
|
184 |
+
"Peltigera praetextata": 99,
|
185 |
+
"Phaeophyscia orbicularis": 98,
|
186 |
+
"Phallus impudicus": 72,
|
187 |
+
"Phellinus igniarius": 10,
|
188 |
+
"Phellinus tremulae": 58,
|
189 |
+
"Phlebia radiata": 70,
|
190 |
+
"Pholiota aurivella": 15,
|
191 |
+
"Pholiota squarrosa": 37,
|
192 |
+
"Physcia adscendens": 27,
|
193 |
+
"Platismatia glauca": 85,
|
194 |
+
"Pleurotus ostreatus": 69,
|
195 |
+
"Pleurotus pulmonarius": 62,
|
196 |
+
"Pseudevernia furfuracea": 32,
|
197 |
+
"Rhytisma acerinum": 50,
|
198 |
+
"Sarcomyxa serotina": 79,
|
199 |
+
"Sarcoscypha austriaca": 40,
|
200 |
+
"Sarcosoma globosum": 19,
|
201 |
+
"Schizophyllum commune": 8,
|
202 |
+
"Stereum hirsutum": 93,
|
203 |
+
"Stropharia aeruginosa": 44,
|
204 |
+
"Suillus granulatus": 75,
|
205 |
+
"Suillus grevillei": 46,
|
206 |
+
"Suillus luteus": 11,
|
207 |
+
"Trametes hirsuta": 66,
|
208 |
+
"Trametes ochracea": 30,
|
209 |
+
"Trametes versicolor": 55,
|
210 |
+
"Tremella mesenterica": 33,
|
211 |
+
"Trichaptum biforme": 16,
|
212 |
+
"Tricholomopsis rutilans": 53,
|
213 |
+
"Urnula craterium": 0,
|
214 |
+
"Verpa bohemica": 7,
|
215 |
+
"Vulpicida pinastri": 24,
|
216 |
+
"Xanthoria parietina": 77
|
217 |
+
},
|
218 |
+
"layer_norm_eps": 1e-12,
|
219 |
+
"model_type": "vit",
|
220 |
+
"num_attention_heads": 12,
|
221 |
+
"num_channels": 3,
|
222 |
+
"num_hidden_layers": 12,
|
223 |
+
"patch_size": 16,
|
224 |
+
"problem_type": "single_label_classification",
|
225 |
+
"qkv_bias": true,
|
226 |
+
"torch_dtype": "float32",
|
227 |
+
"transformers_version": "4.39.2"
|
228 |
+
}
|
checkpoint-10945/model.safetensors
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ccf47aecf2a5fe3fd2c3f2724fbbfb6295e0d057bb189850fbdf7f05b8d977d3
|
3 |
+
size 343525432
|
checkpoint-10945/optimizer.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f34f9061525e3d68314e07eba8fc9c7c851ebcc9f5ec6a69ea45435c3e3c6c7a
|
3 |
+
size 687171333
|
checkpoint-10945/preprocessor_config.json
ADDED
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_valid_processor_keys": [
|
3 |
+
"images",
|
4 |
+
"do_resize",
|
5 |
+
"size",
|
6 |
+
"resample",
|
7 |
+
"do_rescale",
|
8 |
+
"rescale_factor",
|
9 |
+
"do_normalize",
|
10 |
+
"image_mean",
|
11 |
+
"image_std",
|
12 |
+
"return_tensors",
|
13 |
+
"data_format",
|
14 |
+
"input_data_format"
|
15 |
+
],
|
16 |
+
"do_normalize": true,
|
17 |
+
"do_rescale": true,
|
18 |
+
"do_resize": true,
|
19 |
+
"image_mean": [
|
20 |
+
0.5,
|
21 |
+
0.5,
|
22 |
+
0.5
|
23 |
+
],
|
24 |
+
"image_processor_type": "ViTImageProcessor",
|
25 |
+
"image_std": [
|
26 |
+
0.5,
|
27 |
+
0.5,
|
28 |
+
0.5
|
29 |
+
],
|
30 |
+
"resample": 2,
|
31 |
+
"rescale_factor": 0.00392156862745098,
|
32 |
+
"size": {
|
33 |
+
"height": 224,
|
34 |
+
"width": 224
|
35 |
+
}
|
36 |
+
}
|
checkpoint-10945/rng_state.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6d1a23f83b056a08e277bbc4e490710bcc686f95d8686e5101ea17434157c2e7
|
3 |
+
size 14575
|
checkpoint-10945/scheduler.pt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:87ec7618851df92f5c0f27b5f44bb6d99267753deaac58543a1cbc20a80becb3
|
3 |
+
size 627
|
checkpoint-10945/trainer_state.json
ADDED
@@ -0,0 +1,177 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"best_metric": 0.6230026483535767,
|
3 |
+
"best_model_checkpoint": "mushrooms_image_detection/checkpoint-10945",
|
4 |
+
"epoch": 1.0,
|
5 |
+
"eval_steps": 500,
|
6 |
+
"global_step": 10945,
|
7 |
+
"is_hyper_param_search": false,
|
8 |
+
"is_local_process_zero": true,
|
9 |
+
"is_world_process_zero": true,
|
10 |
+
"log_history": [
|
11 |
+
{
|
12 |
+
"epoch": 0.05,
|
13 |
+
"grad_norm": 5.758909702301025,
|
14 |
+
"learning_rate": 1.9173932996787516e-07,
|
15 |
+
"loss": 0.7238,
|
16 |
+
"step": 500
|
17 |
+
},
|
18 |
+
{
|
19 |
+
"epoch": 0.09,
|
20 |
+
"grad_norm": 7.06810188293457,
|
21 |
+
"learning_rate": 1.825608077099587e-07,
|
22 |
+
"loss": 0.7267,
|
23 |
+
"step": 1000
|
24 |
+
},
|
25 |
+
{
|
26 |
+
"epoch": 0.14,
|
27 |
+
"grad_norm": 6.429927349090576,
|
28 |
+
"learning_rate": 1.733822854520422e-07,
|
29 |
+
"loss": 0.711,
|
30 |
+
"step": 1500
|
31 |
+
},
|
32 |
+
{
|
33 |
+
"epoch": 0.18,
|
34 |
+
"grad_norm": 7.55098819732666,
|
35 |
+
"learning_rate": 1.6420376319412576e-07,
|
36 |
+
"loss": 0.7174,
|
37 |
+
"step": 2000
|
38 |
+
},
|
39 |
+
{
|
40 |
+
"epoch": 0.23,
|
41 |
+
"grad_norm": 7.272024154663086,
|
42 |
+
"learning_rate": 1.5502524093620926e-07,
|
43 |
+
"loss": 0.713,
|
44 |
+
"step": 2500
|
45 |
+
},
|
46 |
+
{
|
47 |
+
"epoch": 0.27,
|
48 |
+
"grad_norm": 8.493428230285645,
|
49 |
+
"learning_rate": 1.458467186782928e-07,
|
50 |
+
"loss": 0.6996,
|
51 |
+
"step": 3000
|
52 |
+
},
|
53 |
+
{
|
54 |
+
"epoch": 0.32,
|
55 |
+
"grad_norm": 4.434696197509766,
|
56 |
+
"learning_rate": 1.366681964203763e-07,
|
57 |
+
"loss": 0.7165,
|
58 |
+
"step": 3500
|
59 |
+
},
|
60 |
+
{
|
61 |
+
"epoch": 0.37,
|
62 |
+
"grad_norm": 6.052495956420898,
|
63 |
+
"learning_rate": 1.2748967416245983e-07,
|
64 |
+
"loss": 0.7169,
|
65 |
+
"step": 4000
|
66 |
+
},
|
67 |
+
{
|
68 |
+
"epoch": 0.41,
|
69 |
+
"grad_norm": 10.02051067352295,
|
70 |
+
"learning_rate": 1.1831115190454337e-07,
|
71 |
+
"loss": 0.6994,
|
72 |
+
"step": 4500
|
73 |
+
},
|
74 |
+
{
|
75 |
+
"epoch": 0.46,
|
76 |
+
"grad_norm": 4.502712726593018,
|
77 |
+
"learning_rate": 1.0913262964662688e-07,
|
78 |
+
"loss": 0.7075,
|
79 |
+
"step": 5000
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"epoch": 0.5,
|
83 |
+
"grad_norm": 4.787944793701172,
|
84 |
+
"learning_rate": 9.995410738871042e-08,
|
85 |
+
"loss": 0.7136,
|
86 |
+
"step": 5500
|
87 |
+
},
|
88 |
+
{
|
89 |
+
"epoch": 0.55,
|
90 |
+
"grad_norm": 8.738495826721191,
|
91 |
+
"learning_rate": 9.077558513079394e-08,
|
92 |
+
"loss": 0.6976,
|
93 |
+
"step": 6000
|
94 |
+
},
|
95 |
+
{
|
96 |
+
"epoch": 0.59,
|
97 |
+
"grad_norm": 6.709692001342773,
|
98 |
+
"learning_rate": 8.159706287287747e-08,
|
99 |
+
"loss": 0.6994,
|
100 |
+
"step": 6500
|
101 |
+
},
|
102 |
+
{
|
103 |
+
"epoch": 0.64,
|
104 |
+
"grad_norm": 2.5347442626953125,
|
105 |
+
"learning_rate": 7.241854061496099e-08,
|
106 |
+
"loss": 0.6996,
|
107 |
+
"step": 7000
|
108 |
+
},
|
109 |
+
{
|
110 |
+
"epoch": 0.69,
|
111 |
+
"grad_norm": 5.286170482635498,
|
112 |
+
"learning_rate": 6.32400183570445e-08,
|
113 |
+
"loss": 0.6961,
|
114 |
+
"step": 7500
|
115 |
+
},
|
116 |
+
{
|
117 |
+
"epoch": 0.73,
|
118 |
+
"grad_norm": 5.30819034576416,
|
119 |
+
"learning_rate": 5.406149609912804e-08,
|
120 |
+
"loss": 0.6897,
|
121 |
+
"step": 8000
|
122 |
+
},
|
123 |
+
{
|
124 |
+
"epoch": 0.78,
|
125 |
+
"grad_norm": 7.743193626403809,
|
126 |
+
"learning_rate": 4.488297384121156e-08,
|
127 |
+
"loss": 0.6855,
|
128 |
+
"step": 8500
|
129 |
+
},
|
130 |
+
{
|
131 |
+
"epoch": 0.82,
|
132 |
+
"grad_norm": 8.52515697479248,
|
133 |
+
"learning_rate": 3.5704451583295086e-08,
|
134 |
+
"loss": 0.6966,
|
135 |
+
"step": 9000
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"epoch": 0.87,
|
139 |
+
"grad_norm": 5.206400394439697,
|
140 |
+
"learning_rate": 2.6525929325378617e-08,
|
141 |
+
"loss": 0.7043,
|
142 |
+
"step": 9500
|
143 |
+
},
|
144 |
+
{
|
145 |
+
"epoch": 0.91,
|
146 |
+
"grad_norm": 6.084430694580078,
|
147 |
+
"learning_rate": 1.7347407067462138e-08,
|
148 |
+
"loss": 0.6943,
|
149 |
+
"step": 10000
|
150 |
+
},
|
151 |
+
{
|
152 |
+
"epoch": 0.96,
|
153 |
+
"grad_norm": 8.045140266418457,
|
154 |
+
"learning_rate": 8.168884809545663e-09,
|
155 |
+
"loss": 0.6869,
|
156 |
+
"step": 10500
|
157 |
+
},
|
158 |
+
{
|
159 |
+
"epoch": 1.0,
|
160 |
+
"eval_accuracy": 0.8854163097481583,
|
161 |
+
"eval_loss": 0.6230026483535767,
|
162 |
+
"eval_runtime": 2891.1403,
|
163 |
+
"eval_samples_per_second": 80.757,
|
164 |
+
"eval_steps_per_second": 10.095,
|
165 |
+
"step": 10945
|
166 |
+
}
|
167 |
+
],
|
168 |
+
"logging_steps": 500,
|
169 |
+
"max_steps": 10945,
|
170 |
+
"num_input_tokens_seen": 0,
|
171 |
+
"num_train_epochs": 1,
|
172 |
+
"save_steps": 500,
|
173 |
+
"total_flos": 2.716308216840831e+19,
|
174 |
+
"train_batch_size": 32,
|
175 |
+
"trial_name": null,
|
176 |
+
"trial_params": null
|
177 |
+
}
|
checkpoint-10945/training_args.bin
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dbe1b5c1ae744cd46bfc83fac7b8e6a1a1b6ccaf467352cf9058fbdc53c776f6
|
3 |
+
size 4411
|
model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 343525432
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ccf47aecf2a5fe3fd2c3f2724fbbfb6295e0d057bb189850fbdf7f05b8d977d3
|
3 |
size 343525432
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4411
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dbe1b5c1ae744cd46bfc83fac7b8e6a1a1b6ccaf467352cf9058fbdc53c776f6
|
3 |
size 4411
|