Upload folder using huggingface_hub
Browse files
slots/14/checkpoint-1421/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"TwinyForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_dropout": 0.0,
|
| 6 |
+
"dtype": "float32",
|
| 7 |
+
"hidden_dropout": 0.0,
|
| 8 |
+
"hidden_size": 768,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 3072,
|
| 11 |
+
"max_position_embeddings": 128,
|
| 12 |
+
"model_type": "twiny",
|
| 13 |
+
"neftune_alpha": 0.0,
|
| 14 |
+
"num_attention_heads": 12,
|
| 15 |
+
"num_hidden_layers": 3,
|
| 16 |
+
"num_key_value_heads": 3,
|
| 17 |
+
"qk_norm": true,
|
| 18 |
+
"rezero_init": 1.0,
|
| 19 |
+
"rms_norm_eps": 1e-06,
|
| 20 |
+
"rope_theta": 10000.0,
|
| 21 |
+
"transformers_version": "5.0.0",
|
| 22 |
+
"use_cache": false,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
slots/14/checkpoint-1421/trainer_state.json
ADDED
|
@@ -0,0 +1,538 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.22164092532356755,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 1421,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0001559753169060996,
|
| 14 |
+
"grad_norm": Infinity,
|
| 15 |
+
"learning_rate": 5e-05,
|
| 16 |
+
"loss": 143.68887329101562,
|
| 17 |
+
"step": 1
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0031195063381219924,
|
| 21 |
+
"grad_norm": 52.297000885009766,
|
| 22 |
+
"learning_rate": 4.999993729149804e-05,
|
| 23 |
+
"loss": 63.87562962582237,
|
| 24 |
+
"step": 20
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.006239012676243985,
|
| 28 |
+
"grad_norm": 18.09991455078125,
|
| 29 |
+
"learning_rate": 4.999968253874798e-05,
|
| 30 |
+
"loss": 10.082135009765626,
|
| 31 |
+
"step": 40
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.009358519014365976,
|
| 35 |
+
"grad_norm": 15.637947082519531,
|
| 36 |
+
"learning_rate": 4.99992318244638e-05,
|
| 37 |
+
"loss": 5.112002944946289,
|
| 38 |
+
"step": 60
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.01247802535248797,
|
| 42 |
+
"grad_norm": 12.031600952148438,
|
| 43 |
+
"learning_rate": 4.9998585152178466e-05,
|
| 44 |
+
"loss": 2.7785192489624024,
|
| 45 |
+
"step": 80
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.015597531690609961,
|
| 49 |
+
"grad_norm": 6.920973300933838,
|
| 50 |
+
"learning_rate": 4.999774252696095e-05,
|
| 51 |
+
"loss": 2.0463287353515627,
|
| 52 |
+
"step": 100
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.018717038028731953,
|
| 56 |
+
"grad_norm": 7.486307144165039,
|
| 57 |
+
"learning_rate": 4.999670395541621e-05,
|
| 58 |
+
"loss": 1.7835391998291015,
|
| 59 |
+
"step": 120
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.021836544366853946,
|
| 63 |
+
"grad_norm": 8.828413963317871,
|
| 64 |
+
"learning_rate": 4.999546944568517e-05,
|
| 65 |
+
"loss": 1.2352788925170899,
|
| 66 |
+
"step": 140
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.02495605070497594,
|
| 70 |
+
"grad_norm": 9.396143913269043,
|
| 71 |
+
"learning_rate": 4.9994039007444616e-05,
|
| 72 |
+
"loss": 1.0546810150146484,
|
| 73 |
+
"step": 160
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.02807555704309793,
|
| 77 |
+
"grad_norm": 7.957871913909912,
|
| 78 |
+
"learning_rate": 4.9992412651907096e-05,
|
| 79 |
+
"loss": 0.9882014274597168,
|
| 80 |
+
"step": 180
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.031195063381219922,
|
| 84 |
+
"grad_norm": 4.318814277648926,
|
| 85 |
+
"learning_rate": 4.999059039182092e-05,
|
| 86 |
+
"loss": 0.7930902957916259,
|
| 87 |
+
"step": 200
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.034314569719341916,
|
| 91 |
+
"grad_norm": 5.154116630554199,
|
| 92 |
+
"learning_rate": 4.998857224146998e-05,
|
| 93 |
+
"loss": 0.7149968624114991,
|
| 94 |
+
"step": 220
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.037434076057463905,
|
| 98 |
+
"grad_norm": 11.299652099609375,
|
| 99 |
+
"learning_rate": 4.998635821667367e-05,
|
| 100 |
+
"loss": 0.5597743034362793,
|
| 101 |
+
"step": 240
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.040553582395585895,
|
| 105 |
+
"grad_norm": 4.086544036865234,
|
| 106 |
+
"learning_rate": 4.9983948334786756e-05,
|
| 107 |
+
"loss": 0.5503968715667724,
|
| 108 |
+
"step": 260
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.04367308873370789,
|
| 112 |
+
"grad_norm": 4.867542743682861,
|
| 113 |
+
"learning_rate": 4.998134261469924e-05,
|
| 114 |
+
"loss": 0.5038466930389405,
|
| 115 |
+
"step": 280
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.04679259507182988,
|
| 119 |
+
"grad_norm": 3.7527730464935303,
|
| 120 |
+
"learning_rate": 4.997854107683623e-05,
|
| 121 |
+
"loss": 0.42429962158203127,
|
| 122 |
+
"step": 300
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.04991210140995188,
|
| 126 |
+
"grad_norm": 3.1628715991973877,
|
| 127 |
+
"learning_rate": 4.997554374315774e-05,
|
| 128 |
+
"loss": 0.43418636322021487,
|
| 129 |
+
"step": 320
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.05303160774807387,
|
| 133 |
+
"grad_norm": 6.594268798828125,
|
| 134 |
+
"learning_rate": 4.997235063715855e-05,
|
| 135 |
+
"loss": 0.3593674898147583,
|
| 136 |
+
"step": 340
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.05615111408619586,
|
| 140 |
+
"grad_norm": 3.662322521209717,
|
| 141 |
+
"learning_rate": 4.996896178386803e-05,
|
| 142 |
+
"loss": 0.3210789680480957,
|
| 143 |
+
"step": 360
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.059270620424317855,
|
| 147 |
+
"grad_norm": 3.9782955646514893,
|
| 148 |
+
"learning_rate": 4.9965377209849904e-05,
|
| 149 |
+
"loss": 0.30758669376373293,
|
| 150 |
+
"step": 380
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.062390126762439845,
|
| 154 |
+
"grad_norm": 5.329382419586182,
|
| 155 |
+
"learning_rate": 4.996159694320207e-05,
|
| 156 |
+
"loss": 0.2638844966888428,
|
| 157 |
+
"step": 400
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.06550963310056183,
|
| 161 |
+
"grad_norm": 3.7222423553466797,
|
| 162 |
+
"learning_rate": 4.995762101355639e-05,
|
| 163 |
+
"loss": 0.21310510635375976,
|
| 164 |
+
"step": 420
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.06862913943868383,
|
| 168 |
+
"grad_norm": 3.9508121013641357,
|
| 169 |
+
"learning_rate": 4.9953449452078414e-05,
|
| 170 |
+
"loss": 0.21608681678771974,
|
| 171 |
+
"step": 440
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.07174864577680581,
|
| 175 |
+
"grad_norm": 2.9517951011657715,
|
| 176 |
+
"learning_rate": 4.99490822914672e-05,
|
| 177 |
+
"loss": 0.20024800300598145,
|
| 178 |
+
"step": 460
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.07486815211492781,
|
| 182 |
+
"grad_norm": 1.2055240869522095,
|
| 183 |
+
"learning_rate": 4.994451956595499e-05,
|
| 184 |
+
"loss": 0.1932399868965149,
|
| 185 |
+
"step": 480
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.07798765845304981,
|
| 189 |
+
"grad_norm": 1.6800686120986938,
|
| 190 |
+
"learning_rate": 4.993976131130699e-05,
|
| 191 |
+
"loss": 0.1796867847442627,
|
| 192 |
+
"step": 500
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.08110716479117179,
|
| 196 |
+
"grad_norm": 3.093266725540161,
|
| 197 |
+
"learning_rate": 4.9934807564821065e-05,
|
| 198 |
+
"loss": 0.14106053113937378,
|
| 199 |
+
"step": 520
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.08422667112929379,
|
| 203 |
+
"grad_norm": 1.3461378812789917,
|
| 204 |
+
"learning_rate": 4.992965836532746e-05,
|
| 205 |
+
"loss": 0.15992594957351686,
|
| 206 |
+
"step": 540
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.08734617746741578,
|
| 210 |
+
"grad_norm": 1.4411957263946533,
|
| 211 |
+
"learning_rate": 4.992431375318849e-05,
|
| 212 |
+
"loss": 0.13545843362808227,
|
| 213 |
+
"step": 560
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 0.09046568380553778,
|
| 217 |
+
"grad_norm": 2.334609031677246,
|
| 218 |
+
"learning_rate": 4.991877377029823e-05,
|
| 219 |
+
"loss": 0.12385048866271972,
|
| 220 |
+
"step": 580
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.09358519014365976,
|
| 224 |
+
"grad_norm": 1.9647951126098633,
|
| 225 |
+
"learning_rate": 4.991303846008216e-05,
|
| 226 |
+
"loss": 0.15953426361083983,
|
| 227 |
+
"step": 600
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.09670469648178176,
|
| 231 |
+
"grad_norm": 0.9261314868927002,
|
| 232 |
+
"learning_rate": 4.990710786749686e-05,
|
| 233 |
+
"loss": 0.13601958751678467,
|
| 234 |
+
"step": 620
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.09982420281990376,
|
| 238 |
+
"grad_norm": 2.8501081466674805,
|
| 239 |
+
"learning_rate": 4.990098203902965e-05,
|
| 240 |
+
"loss": 0.12980480194091798,
|
| 241 |
+
"step": 640
|
| 242 |
+
},
|
| 243 |
+
{
|
| 244 |
+
"epoch": 0.10294370915802574,
|
| 245 |
+
"grad_norm": 2.030174493789673,
|
| 246 |
+
"learning_rate": 4.989466102269821e-05,
|
| 247 |
+
"loss": 0.11676372289657592,
|
| 248 |
+
"step": 660
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.10606321549614774,
|
| 252 |
+
"grad_norm": 2.160588502883911,
|
| 253 |
+
"learning_rate": 4.9888144868050194e-05,
|
| 254 |
+
"loss": 0.11160085201263428,
|
| 255 |
+
"step": 680
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 0.10918272183426973,
|
| 259 |
+
"grad_norm": 1.138016939163208,
|
| 260 |
+
"learning_rate": 4.9881433626162896e-05,
|
| 261 |
+
"loss": 0.09880762696266174,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.11230222817239172,
|
| 266 |
+
"grad_norm": 2.918583869934082,
|
| 267 |
+
"learning_rate": 4.98745273496428e-05,
|
| 268 |
+
"loss": 0.08724074363708496,
|
| 269 |
+
"step": 720
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.11542173451051371,
|
| 273 |
+
"grad_norm": 2.165208578109741,
|
| 274 |
+
"learning_rate": 4.986742609262515e-05,
|
| 275 |
+
"loss": 0.11841287612915039,
|
| 276 |
+
"step": 740
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.11854124084863571,
|
| 280 |
+
"grad_norm": 2.163217306137085,
|
| 281 |
+
"learning_rate": 4.986012991077359e-05,
|
| 282 |
+
"loss": 0.10830552577972412,
|
| 283 |
+
"step": 760
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.12166074718675769,
|
| 287 |
+
"grad_norm": 1.270262598991394,
|
| 288 |
+
"learning_rate": 4.98526388612797e-05,
|
| 289 |
+
"loss": 0.06507933735847474,
|
| 290 |
+
"step": 780
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.12478025352487969,
|
| 294 |
+
"grad_norm": 1.2272371053695679,
|
| 295 |
+
"learning_rate": 4.984495300286252e-05,
|
| 296 |
+
"loss": 0.07868309020996093,
|
| 297 |
+
"step": 800
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.12789975986300167,
|
| 301 |
+
"grad_norm": 1.2555146217346191,
|
| 302 |
+
"learning_rate": 4.98370723957681e-05,
|
| 303 |
+
"loss": 0.06831848621368408,
|
| 304 |
+
"step": 820
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 0.13101926620112367,
|
| 308 |
+
"grad_norm": 0.2989479601383209,
|
| 309 |
+
"learning_rate": 4.982899710176909e-05,
|
| 310 |
+
"loss": 0.10411404371261597,
|
| 311 |
+
"step": 840
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 0.13413877253924567,
|
| 315 |
+
"grad_norm": 1.6604547500610352,
|
| 316 |
+
"learning_rate": 4.982072718416416e-05,
|
| 317 |
+
"loss": 0.07505239248275757,
|
| 318 |
+
"step": 860
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.13725827887736766,
|
| 322 |
+
"grad_norm": 6.027090549468994,
|
| 323 |
+
"learning_rate": 4.981226270777756e-05,
|
| 324 |
+
"loss": 0.08949573040008545,
|
| 325 |
+
"step": 880
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.14037778521548966,
|
| 329 |
+
"grad_norm": 5.2599077224731445,
|
| 330 |
+
"learning_rate": 4.9803603738958604e-05,
|
| 331 |
+
"loss": 0.04615107178688049,
|
| 332 |
+
"step": 900
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.14349729155361163,
|
| 336 |
+
"grad_norm": 2.0441677570343018,
|
| 337 |
+
"learning_rate": 4.979475034558115e-05,
|
| 338 |
+
"loss": 0.08460248112678528,
|
| 339 |
+
"step": 920
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.14661679789173362,
|
| 343 |
+
"grad_norm": 1.892174482345581,
|
| 344 |
+
"learning_rate": 4.9785702597043064e-05,
|
| 345 |
+
"loss": 0.06970399022102355,
|
| 346 |
+
"step": 940
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 0.14973630422985562,
|
| 350 |
+
"grad_norm": 1.8726390600204468,
|
| 351 |
+
"learning_rate": 4.977646056426567e-05,
|
| 352 |
+
"loss": 0.07341498732566834,
|
| 353 |
+
"step": 960
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 0.15285581056797762,
|
| 357 |
+
"grad_norm": 3.0138962268829346,
|
| 358 |
+
"learning_rate": 4.97670243196932e-05,
|
| 359 |
+
"loss": 0.06328916549682617,
|
| 360 |
+
"step": 980
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 0.15597531690609961,
|
| 364 |
+
"grad_norm": 1.26374089717865,
|
| 365 |
+
"learning_rate": 4.975739393729224e-05,
|
| 366 |
+
"loss": 0.06133637428283691,
|
| 367 |
+
"step": 1000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 0.1590948232442216,
|
| 371 |
+
"grad_norm": 2.468918561935425,
|
| 372 |
+
"learning_rate": 4.974756949255112e-05,
|
| 373 |
+
"loss": 0.040934398770332336,
|
| 374 |
+
"step": 1020
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.16221432958234358,
|
| 378 |
+
"grad_norm": 1.0652263164520264,
|
| 379 |
+
"learning_rate": 4.973755106247936e-05,
|
| 380 |
+
"loss": 0.053039383888244626,
|
| 381 |
+
"step": 1040
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.16533383592046558,
|
| 385 |
+
"grad_norm": 8.056346893310547,
|
| 386 |
+
"learning_rate": 4.972733872560702e-05,
|
| 387 |
+
"loss": 0.05064538717269897,
|
| 388 |
+
"step": 1060
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.16845334225858757,
|
| 392 |
+
"grad_norm": 0.27906128764152527,
|
| 393 |
+
"learning_rate": 4.971693256198414e-05,
|
| 394 |
+
"loss": 0.038862642645835874,
|
| 395 |
+
"step": 1080
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.17157284859670957,
|
| 399 |
+
"grad_norm": 0.9866332411766052,
|
| 400 |
+
"learning_rate": 4.9706332653180055e-05,
|
| 401 |
+
"loss": 0.05685908198356628,
|
| 402 |
+
"step": 1100
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.17469235493483157,
|
| 406 |
+
"grad_norm": 0.1504131406545639,
|
| 407 |
+
"learning_rate": 4.9695539082282805e-05,
|
| 408 |
+
"loss": 0.054278552532196045,
|
| 409 |
+
"step": 1120
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 0.17781186127295356,
|
| 413 |
+
"grad_norm": 0.6855164170265198,
|
| 414 |
+
"learning_rate": 4.968455193389845e-05,
|
| 415 |
+
"loss": 0.038184747099876404,
|
| 416 |
+
"step": 1140
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.18093136761107556,
|
| 420 |
+
"grad_norm": 1.5174620151519775,
|
| 421 |
+
"learning_rate": 4.967337129415041e-05,
|
| 422 |
+
"loss": 0.032207447290420535,
|
| 423 |
+
"step": 1160
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.18405087394919753,
|
| 427 |
+
"grad_norm": 2.1656675338745117,
|
| 428 |
+
"learning_rate": 4.966199725067883e-05,
|
| 429 |
+
"loss": 0.04032683968544006,
|
| 430 |
+
"step": 1180
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.18717038028731953,
|
| 434 |
+
"grad_norm": 0.16235563158988953,
|
| 435 |
+
"learning_rate": 4.9650429892639824e-05,
|
| 436 |
+
"loss": 0.03260311484336853,
|
| 437 |
+
"step": 1200
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.19028988662544152,
|
| 441 |
+
"grad_norm": 0.7270789742469788,
|
| 442 |
+
"learning_rate": 4.9638669310704856e-05,
|
| 443 |
+
"loss": 0.039445298910140994,
|
| 444 |
+
"step": 1220
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.19340939296356352,
|
| 448 |
+
"grad_norm": 1.4053725004196167,
|
| 449 |
+
"learning_rate": 4.9626715597059936e-05,
|
| 450 |
+
"loss": 0.05901872515678406,
|
| 451 |
+
"step": 1240
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 0.19652889930168552,
|
| 455 |
+
"grad_norm": 0.149334117770195,
|
| 456 |
+
"learning_rate": 4.9614568845404994e-05,
|
| 457 |
+
"loss": 0.05250124335289001,
|
| 458 |
+
"step": 1260
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.1996484056398075,
|
| 462 |
+
"grad_norm": 1.7250016927719116,
|
| 463 |
+
"learning_rate": 4.9602229150953086e-05,
|
| 464 |
+
"loss": 0.03500291109085083,
|
| 465 |
+
"step": 1280
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 0.20276791197792948,
|
| 469 |
+
"grad_norm": 0.3273993730545044,
|
| 470 |
+
"learning_rate": 4.9589696610429635e-05,
|
| 471 |
+
"loss": 0.04656350910663605,
|
| 472 |
+
"step": 1300
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.20588741831605148,
|
| 476 |
+
"grad_norm": 2.355738878250122,
|
| 477 |
+
"learning_rate": 4.957697132207175e-05,
|
| 478 |
+
"loss": 0.0341707855463028,
|
| 479 |
+
"step": 1320
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.20900692465417348,
|
| 483 |
+
"grad_norm": 0.39558863639831543,
|
| 484 |
+
"learning_rate": 4.9564053385627375e-05,
|
| 485 |
+
"loss": 0.04657787382602692,
|
| 486 |
+
"step": 1340
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.21212643099229547,
|
| 490 |
+
"grad_norm": 0.7477391362190247,
|
| 491 |
+
"learning_rate": 4.955094290235452e-05,
|
| 492 |
+
"loss": 0.032107046246528624,
|
| 493 |
+
"step": 1360
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 0.21524593733041747,
|
| 497 |
+
"grad_norm": 8.392775535583496,
|
| 498 |
+
"learning_rate": 4.9537639975020525e-05,
|
| 499 |
+
"loss": 0.034580078721046445,
|
| 500 |
+
"step": 1380
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.21836544366853947,
|
| 504 |
+
"grad_norm": 1.3500334024429321,
|
| 505 |
+
"learning_rate": 4.95241447079012e-05,
|
| 506 |
+
"loss": 0.031853339076042174,
|
| 507 |
+
"step": 1400
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.22148495000666144,
|
| 511 |
+
"grad_norm": 0.7828538417816162,
|
| 512 |
+
"learning_rate": 4.9510457206779996e-05,
|
| 513 |
+
"loss": 0.041383099555969236,
|
| 514 |
+
"step": 1420
|
| 515 |
+
}
|
| 516 |
+
],
|
| 517 |
+
"logging_steps": 20,
|
| 518 |
+
"max_steps": 22442,
|
| 519 |
+
"num_input_tokens_seen": 0,
|
| 520 |
+
"num_train_epochs": 4,
|
| 521 |
+
"save_steps": 1000000000,
|
| 522 |
+
"stateful_callbacks": {
|
| 523 |
+
"TrainerControl": {
|
| 524 |
+
"args": {
|
| 525 |
+
"should_epoch_stop": false,
|
| 526 |
+
"should_evaluate": false,
|
| 527 |
+
"should_log": false,
|
| 528 |
+
"should_save": true,
|
| 529 |
+
"should_training_stop": false
|
| 530 |
+
},
|
| 531 |
+
"attributes": {}
|
| 532 |
+
}
|
| 533 |
+
},
|
| 534 |
+
"total_flos": 5449975495630848.0,
|
| 535 |
+
"train_batch_size": 1,
|
| 536 |
+
"trial_name": null,
|
| 537 |
+
"trial_params": null
|
| 538 |
+
}
|