Add files using upload-large-folder tool
Browse files- model.safetensors +1 -1
- trainer_state.json +711 -3
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2740113872
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:251592d6646f19c540dca28cb759c20bbddf9b68e86c390c0300386f84e83a8c
|
| 3 |
size 2740113872
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 1000,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2124,6 +2124,714 @@
|
|
| 2124 |
"learning_rate": 4.557779879023483e-05,
|
| 2125 |
"loss": 6.5202,
|
| 2126 |
"step": 2990
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2127 |
}
|
| 2128 |
],
|
| 2129 |
"logging_steps": 10,
|
|
@@ -2143,7 +2851,7 @@
|
|
| 2143 |
"attributes": {}
|
| 2144 |
}
|
| 2145 |
},
|
| 2146 |
-
"total_flos":
|
| 2147 |
"train_batch_size": 30,
|
| 2148 |
"trial_name": null,
|
| 2149 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.2893485554295093,
|
| 5 |
"eval_steps": 1000,
|
| 6 |
+
"global_step": 3996,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2124 |
"learning_rate": 4.557779879023483e-05,
|
| 2125 |
"loss": 6.5202,
|
| 2126 |
"step": 2990
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.21722864521734933,
|
| 2130 |
+
"grad_norm": 1.328125,
|
| 2131 |
+
"learning_rate": 4.554444538063113e-05,
|
| 2132 |
+
"loss": 6.5246,
|
| 2133 |
+
"step": 3000
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.21722864521734933,
|
| 2137 |
+
"eval_loss": 6.682950973510742,
|
| 2138 |
+
"eval_runtime": 1.4758,
|
| 2139 |
+
"eval_samples_per_second": 8.131,
|
| 2140 |
+
"eval_steps_per_second": 2.033,
|
| 2141 |
+
"step": 3000
|
| 2142 |
+
},
|
| 2143 |
+
{
|
| 2144 |
+
"epoch": 0.21795274070140716,
|
| 2145 |
+
"grad_norm": 1.3125,
|
| 2146 |
+
"learning_rate": 4.551097896319838e-05,
|
| 2147 |
+
"loss": 6.5144,
|
| 2148 |
+
"step": 3010
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 0.218676836185465,
|
| 2152 |
+
"grad_norm": 1.3671875,
|
| 2153 |
+
"learning_rate": 4.5477399722023674e-05,
|
| 2154 |
+
"loss": 6.5295,
|
| 2155 |
+
"step": 3020
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 0.21940093166952282,
|
| 2159 |
+
"grad_norm": 1.3125,
|
| 2160 |
+
"learning_rate": 4.5443707841814695e-05,
|
| 2161 |
+
"loss": 6.5278,
|
| 2162 |
+
"step": 3030
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.22012502715358065,
|
| 2166 |
+
"grad_norm": 1.2734375,
|
| 2167 |
+
"learning_rate": 4.540990350789875e-05,
|
| 2168 |
+
"loss": 6.5584,
|
| 2169 |
+
"step": 3040
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 0.22084912263763848,
|
| 2173 |
+
"grad_norm": 1.3046875,
|
| 2174 |
+
"learning_rate": 4.537598690622167e-05,
|
| 2175 |
+
"loss": 6.5528,
|
| 2176 |
+
"step": 3050
|
| 2177 |
+
},
|
| 2178 |
+
{
|
| 2179 |
+
"epoch": 0.2215732181216963,
|
| 2180 |
+
"grad_norm": 1.203125,
|
| 2181 |
+
"learning_rate": 4.534195822334686e-05,
|
| 2182 |
+
"loss": 6.5392,
|
| 2183 |
+
"step": 3060
|
| 2184 |
+
},
|
| 2185 |
+
{
|
| 2186 |
+
"epoch": 0.22229731360575414,
|
| 2187 |
+
"grad_norm": 1.3984375,
|
| 2188 |
+
"learning_rate": 4.530781764645422e-05,
|
| 2189 |
+
"loss": 6.5403,
|
| 2190 |
+
"step": 3070
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 0.22302140908981197,
|
| 2194 |
+
"grad_norm": 1.3125,
|
| 2195 |
+
"learning_rate": 4.5273565363339185e-05,
|
| 2196 |
+
"loss": 6.5396,
|
| 2197 |
+
"step": 3080
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 0.2237455045738698,
|
| 2201 |
+
"grad_norm": 1.3125,
|
| 2202 |
+
"learning_rate": 4.52392015624116e-05,
|
| 2203 |
+
"loss": 6.5249,
|
| 2204 |
+
"step": 3090
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.22446960005792763,
|
| 2208 |
+
"grad_norm": 1.25,
|
| 2209 |
+
"learning_rate": 4.520472643269477e-05,
|
| 2210 |
+
"loss": 6.5199,
|
| 2211 |
+
"step": 3100
|
| 2212 |
+
},
|
| 2213 |
+
{
|
| 2214 |
+
"epoch": 0.22519369554198548,
|
| 2215 |
+
"grad_norm": 1.375,
|
| 2216 |
+
"learning_rate": 4.517014016382432e-05,
|
| 2217 |
+
"loss": 6.5302,
|
| 2218 |
+
"step": 3110
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 0.2259177910260433,
|
| 2222 |
+
"grad_norm": 1.28125,
|
| 2223 |
+
"learning_rate": 4.5135442946047294e-05,
|
| 2224 |
+
"loss": 6.5274,
|
| 2225 |
+
"step": 3120
|
| 2226 |
+
},
|
| 2227 |
+
{
|
| 2228 |
+
"epoch": 0.22664188651010114,
|
| 2229 |
+
"grad_norm": 1.1953125,
|
| 2230 |
+
"learning_rate": 4.5100634970220967e-05,
|
| 2231 |
+
"loss": 6.5431,
|
| 2232 |
+
"step": 3130
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 0.22736598199415897,
|
| 2236 |
+
"grad_norm": 1.2421875,
|
| 2237 |
+
"learning_rate": 4.5065716427811874e-05,
|
| 2238 |
+
"loss": 6.5362,
|
| 2239 |
+
"step": 3140
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 0.2280900774782168,
|
| 2243 |
+
"grad_norm": 1.3046875,
|
| 2244 |
+
"learning_rate": 4.503068751089474e-05,
|
| 2245 |
+
"loss": 6.5307,
|
| 2246 |
+
"step": 3150
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 0.22881417296227463,
|
| 2250 |
+
"grad_norm": 1.296875,
|
| 2251 |
+
"learning_rate": 4.499554841215143e-05,
|
| 2252 |
+
"loss": 6.5172,
|
| 2253 |
+
"step": 3160
|
| 2254 |
+
},
|
| 2255 |
+
{
|
| 2256 |
+
"epoch": 0.22953826844633246,
|
| 2257 |
+
"grad_norm": 1.203125,
|
| 2258 |
+
"learning_rate": 4.496029932486986e-05,
|
| 2259 |
+
"loss": 6.521,
|
| 2260 |
+
"step": 3170
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 0.2302623639303903,
|
| 2264 |
+
"grad_norm": 1.1796875,
|
| 2265 |
+
"learning_rate": 4.492494044294297e-05,
|
| 2266 |
+
"loss": 6.5346,
|
| 2267 |
+
"step": 3180
|
| 2268 |
+
},
|
| 2269 |
+
{
|
| 2270 |
+
"epoch": 0.23098645941444812,
|
| 2271 |
+
"grad_norm": 1.2578125,
|
| 2272 |
+
"learning_rate": 4.4889471960867635e-05,
|
| 2273 |
+
"loss": 6.5422,
|
| 2274 |
+
"step": 3190
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 0.23171055489850595,
|
| 2278 |
+
"grad_norm": 1.34375,
|
| 2279 |
+
"learning_rate": 4.485389407374361e-05,
|
| 2280 |
+
"loss": 6.5498,
|
| 2281 |
+
"step": 3200
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 0.23243465038256378,
|
| 2285 |
+
"grad_norm": 1.265625,
|
| 2286 |
+
"learning_rate": 4.481820697727244e-05,
|
| 2287 |
+
"loss": 6.523,
|
| 2288 |
+
"step": 3210
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 0.2331587458666216,
|
| 2292 |
+
"grad_norm": 1.1875,
|
| 2293 |
+
"learning_rate": 4.47824108677564e-05,
|
| 2294 |
+
"loss": 6.5477,
|
| 2295 |
+
"step": 3220
|
| 2296 |
+
},
|
| 2297 |
+
{
|
| 2298 |
+
"epoch": 0.23388284135067944,
|
| 2299 |
+
"grad_norm": 1.3359375,
|
| 2300 |
+
"learning_rate": 4.47465059420974e-05,
|
| 2301 |
+
"loss": 6.5379,
|
| 2302 |
+
"step": 3230
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 0.23460693683473727,
|
| 2306 |
+
"grad_norm": 1.25,
|
| 2307 |
+
"learning_rate": 4.471049239779592e-05,
|
| 2308 |
+
"loss": 6.5389,
|
| 2309 |
+
"step": 3240
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 0.2353310323187951,
|
| 2313 |
+
"grad_norm": 1.3125,
|
| 2314 |
+
"learning_rate": 4.4674370432949905e-05,
|
| 2315 |
+
"loss": 6.5552,
|
| 2316 |
+
"step": 3250
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 0.23605512780285293,
|
| 2320 |
+
"grad_norm": 1.3046875,
|
| 2321 |
+
"learning_rate": 4.463814024625368e-05,
|
| 2322 |
+
"loss": 6.5114,
|
| 2323 |
+
"step": 3260
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 0.23677922328691076,
|
| 2327 |
+
"grad_norm": 1.28125,
|
| 2328 |
+
"learning_rate": 4.460180203699688e-05,
|
| 2329 |
+
"loss": 6.5101,
|
| 2330 |
+
"step": 3270
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 0.23750331877096859,
|
| 2334 |
+
"grad_norm": 1.1953125,
|
| 2335 |
+
"learning_rate": 4.4565356005063304e-05,
|
| 2336 |
+
"loss": 6.5051,
|
| 2337 |
+
"step": 3280
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 0.23822741425502644,
|
| 2341 |
+
"grad_norm": 1.3046875,
|
| 2342 |
+
"learning_rate": 4.452880235092987e-05,
|
| 2343 |
+
"loss": 6.5213,
|
| 2344 |
+
"step": 3290
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 0.23895150973908427,
|
| 2348 |
+
"grad_norm": 1.25,
|
| 2349 |
+
"learning_rate": 4.449214127566549e-05,
|
| 2350 |
+
"loss": 6.5246,
|
| 2351 |
+
"step": 3300
|
| 2352 |
+
},
|
| 2353 |
+
{
|
| 2354 |
+
"epoch": 0.2396756052231421,
|
| 2355 |
+
"grad_norm": 1.2578125,
|
| 2356 |
+
"learning_rate": 4.4455372980929935e-05,
|
| 2357 |
+
"loss": 6.5309,
|
| 2358 |
+
"step": 3310
|
| 2359 |
+
},
|
| 2360 |
+
{
|
| 2361 |
+
"epoch": 0.24039970070719993,
|
| 2362 |
+
"grad_norm": 1.359375,
|
| 2363 |
+
"learning_rate": 4.4418497668972785e-05,
|
| 2364 |
+
"loss": 6.5349,
|
| 2365 |
+
"step": 3320
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 0.24112379619125776,
|
| 2369 |
+
"grad_norm": 1.2421875,
|
| 2370 |
+
"learning_rate": 4.4381515542632274e-05,
|
| 2371 |
+
"loss": 6.5232,
|
| 2372 |
+
"step": 3330
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 0.2418478916753156,
|
| 2376 |
+
"grad_norm": 1.21875,
|
| 2377 |
+
"learning_rate": 4.434442680533417e-05,
|
| 2378 |
+
"loss": 6.518,
|
| 2379 |
+
"step": 3340
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 0.24257198715937342,
|
| 2383 |
+
"grad_norm": 1.2734375,
|
| 2384 |
+
"learning_rate": 4.430723166109069e-05,
|
| 2385 |
+
"loss": 6.5465,
|
| 2386 |
+
"step": 3350
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 0.24329608264343125,
|
| 2390 |
+
"grad_norm": 1.28125,
|
| 2391 |
+
"learning_rate": 4.426993031449934e-05,
|
| 2392 |
+
"loss": 6.5353,
|
| 2393 |
+
"step": 3360
|
| 2394 |
+
},
|
| 2395 |
+
{
|
| 2396 |
+
"epoch": 0.24402017812748908,
|
| 2397 |
+
"grad_norm": 1.234375,
|
| 2398 |
+
"learning_rate": 4.423252297074183e-05,
|
| 2399 |
+
"loss": 6.5499,
|
| 2400 |
+
"step": 3370
|
| 2401 |
+
},
|
| 2402 |
+
{
|
| 2403 |
+
"epoch": 0.2447442736115469,
|
| 2404 |
+
"grad_norm": 1.3203125,
|
| 2405 |
+
"learning_rate": 4.41950098355829e-05,
|
| 2406 |
+
"loss": 6.5203,
|
| 2407 |
+
"step": 3380
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 0.24546836909560474,
|
| 2411 |
+
"grad_norm": 1.3125,
|
| 2412 |
+
"learning_rate": 4.415739111536924e-05,
|
| 2413 |
+
"loss": 6.5392,
|
| 2414 |
+
"step": 3390
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.24619246457966257,
|
| 2418 |
+
"grad_norm": 1.3515625,
|
| 2419 |
+
"learning_rate": 4.4119667017028297e-05,
|
| 2420 |
+
"loss": 6.5197,
|
| 2421 |
+
"step": 3400
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 0.2469165600637204,
|
| 2425 |
+
"grad_norm": 1.2578125,
|
| 2426 |
+
"learning_rate": 4.4081837748067186e-05,
|
| 2427 |
+
"loss": 6.4986,
|
| 2428 |
+
"step": 3410
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 0.24764065554777823,
|
| 2432 |
+
"grad_norm": 1.21875,
|
| 2433 |
+
"learning_rate": 4.404390351657153e-05,
|
| 2434 |
+
"loss": 6.5238,
|
| 2435 |
+
"step": 3420
|
| 2436 |
+
},
|
| 2437 |
+
{
|
| 2438 |
+
"epoch": 0.24836475103183606,
|
| 2439 |
+
"grad_norm": 1.25,
|
| 2440 |
+
"learning_rate": 4.4005864531204285e-05,
|
| 2441 |
+
"loss": 6.5473,
|
| 2442 |
+
"step": 3430
|
| 2443 |
+
},
|
| 2444 |
+
{
|
| 2445 |
+
"epoch": 0.24908884651589389,
|
| 2446 |
+
"grad_norm": 1.1953125,
|
| 2447 |
+
"learning_rate": 4.396772100120466e-05,
|
| 2448 |
+
"loss": 6.5189,
|
| 2449 |
+
"step": 3440
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 0.24981294199995172,
|
| 2453 |
+
"grad_norm": 1.3046875,
|
| 2454 |
+
"learning_rate": 4.39294731363869e-05,
|
| 2455 |
+
"loss": 6.5228,
|
| 2456 |
+
"step": 3450
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 0.2505370374840096,
|
| 2460 |
+
"grad_norm": 1.296875,
|
| 2461 |
+
"learning_rate": 4.389112114713918e-05,
|
| 2462 |
+
"loss": 6.5224,
|
| 2463 |
+
"step": 3460
|
| 2464 |
+
},
|
| 2465 |
+
{
|
| 2466 |
+
"epoch": 0.2512611329680674,
|
| 2467 |
+
"grad_norm": 1.3359375,
|
| 2468 |
+
"learning_rate": 4.385266524442241e-05,
|
| 2469 |
+
"loss": 6.5229,
|
| 2470 |
+
"step": 3470
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 0.25198522845212523,
|
| 2474 |
+
"grad_norm": 1.2578125,
|
| 2475 |
+
"learning_rate": 4.3814105639769106e-05,
|
| 2476 |
+
"loss": 6.54,
|
| 2477 |
+
"step": 3480
|
| 2478 |
+
},
|
| 2479 |
+
{
|
| 2480 |
+
"epoch": 0.25270932393618306,
|
| 2481 |
+
"grad_norm": 1.34375,
|
| 2482 |
+
"learning_rate": 4.37754425452822e-05,
|
| 2483 |
+
"loss": 6.5226,
|
| 2484 |
+
"step": 3490
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 0.2534334194202409,
|
| 2488 |
+
"grad_norm": 1.21875,
|
| 2489 |
+
"learning_rate": 4.373667617363389e-05,
|
| 2490 |
+
"loss": 6.5276,
|
| 2491 |
+
"step": 3500
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 0.2541575149042987,
|
| 2495 |
+
"grad_norm": 1.3828125,
|
| 2496 |
+
"learning_rate": 4.369780673806447e-05,
|
| 2497 |
+
"loss": 6.534,
|
| 2498 |
+
"step": 3510
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 0.25488161038835655,
|
| 2502 |
+
"grad_norm": 1.203125,
|
| 2503 |
+
"learning_rate": 4.365883445238116e-05,
|
| 2504 |
+
"loss": 6.5317,
|
| 2505 |
+
"step": 3520
|
| 2506 |
+
},
|
| 2507 |
+
{
|
| 2508 |
+
"epoch": 0.2556057058724144,
|
| 2509 |
+
"grad_norm": 1.2734375,
|
| 2510 |
+
"learning_rate": 4.361975953095689e-05,
|
| 2511 |
+
"loss": 6.5119,
|
| 2512 |
+
"step": 3530
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 0.2563298013564722,
|
| 2516 |
+
"grad_norm": 1.265625,
|
| 2517 |
+
"learning_rate": 4.358058218872918e-05,
|
| 2518 |
+
"loss": 6.5281,
|
| 2519 |
+
"step": 3540
|
| 2520 |
+
},
|
| 2521 |
+
{
|
| 2522 |
+
"epoch": 0.25705389684053004,
|
| 2523 |
+
"grad_norm": 1.2734375,
|
| 2524 |
+
"learning_rate": 4.354130264119894e-05,
|
| 2525 |
+
"loss": 6.5284,
|
| 2526 |
+
"step": 3550
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 0.25777799232458787,
|
| 2530 |
+
"grad_norm": 1.3046875,
|
| 2531 |
+
"learning_rate": 4.350192110442926e-05,
|
| 2532 |
+
"loss": 6.5324,
|
| 2533 |
+
"step": 3560
|
| 2534 |
+
},
|
| 2535 |
+
{
|
| 2536 |
+
"epoch": 0.2585020878086457,
|
| 2537 |
+
"grad_norm": 1.2265625,
|
| 2538 |
+
"learning_rate": 4.346243779504421e-05,
|
| 2539 |
+
"loss": 6.523,
|
| 2540 |
+
"step": 3570
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 0.2592261832927035,
|
| 2544 |
+
"grad_norm": 1.28125,
|
| 2545 |
+
"learning_rate": 4.342285293022775e-05,
|
| 2546 |
+
"loss": 6.5267,
|
| 2547 |
+
"step": 3580
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 0.25995027877676136,
|
| 2551 |
+
"grad_norm": 1.2265625,
|
| 2552 |
+
"learning_rate": 4.338316672772238e-05,
|
| 2553 |
+
"loss": 6.5263,
|
| 2554 |
+
"step": 3590
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 0.2606743742608192,
|
| 2558 |
+
"grad_norm": 1.21875,
|
| 2559 |
+
"learning_rate": 4.334337940582808e-05,
|
| 2560 |
+
"loss": 6.5438,
|
| 2561 |
+
"step": 3600
|
| 2562 |
+
},
|
| 2563 |
+
{
|
| 2564 |
+
"epoch": 0.261398469744877,
|
| 2565 |
+
"grad_norm": 1.21875,
|
| 2566 |
+
"learning_rate": 4.330349118340102e-05,
|
| 2567 |
+
"loss": 6.5315,
|
| 2568 |
+
"step": 3610
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 0.26212256522893485,
|
| 2572 |
+
"grad_norm": 1.203125,
|
| 2573 |
+
"learning_rate": 4.326350227985241e-05,
|
| 2574 |
+
"loss": 6.5183,
|
| 2575 |
+
"step": 3620
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 0.2628466607129927,
|
| 2579 |
+
"grad_norm": 1.28125,
|
| 2580 |
+
"learning_rate": 4.3223412915147254e-05,
|
| 2581 |
+
"loss": 6.5214,
|
| 2582 |
+
"step": 3630
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 0.2635707561970505,
|
| 2586 |
+
"grad_norm": 1.3515625,
|
| 2587 |
+
"learning_rate": 4.318322330980317e-05,
|
| 2588 |
+
"loss": 6.531,
|
| 2589 |
+
"step": 3640
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 0.26429485168110833,
|
| 2593 |
+
"grad_norm": 1.2578125,
|
| 2594 |
+
"learning_rate": 4.314293368488915e-05,
|
| 2595 |
+
"loss": 6.5133,
|
| 2596 |
+
"step": 3650
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 0.26501894716516616,
|
| 2600 |
+
"grad_norm": 1.25,
|
| 2601 |
+
"learning_rate": 4.3102544262024394e-05,
|
| 2602 |
+
"loss": 6.5288,
|
| 2603 |
+
"step": 3660
|
| 2604 |
+
},
|
| 2605 |
+
{
|
| 2606 |
+
"epoch": 0.265743042649224,
|
| 2607 |
+
"grad_norm": 1.1953125,
|
| 2608 |
+
"learning_rate": 4.3062055263377e-05,
|
| 2609 |
+
"loss": 6.4998,
|
| 2610 |
+
"step": 3670
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 0.2664671381332818,
|
| 2614 |
+
"grad_norm": 1.3203125,
|
| 2615 |
+
"learning_rate": 4.302146691166286e-05,
|
| 2616 |
+
"loss": 6.5302,
|
| 2617 |
+
"step": 3680
|
| 2618 |
+
},
|
| 2619 |
+
{
|
| 2620 |
+
"epoch": 0.26719123361733965,
|
| 2621 |
+
"grad_norm": 1.265625,
|
| 2622 |
+
"learning_rate": 4.298077943014431e-05,
|
| 2623 |
+
"loss": 6.5053,
|
| 2624 |
+
"step": 3690
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 0.2679153291013975,
|
| 2628 |
+
"grad_norm": 1.2734375,
|
| 2629 |
+
"learning_rate": 4.293999304262902e-05,
|
| 2630 |
+
"loss": 6.5242,
|
| 2631 |
+
"step": 3700
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 0.2686394245854553,
|
| 2635 |
+
"grad_norm": 1.21875,
|
| 2636 |
+
"learning_rate": 4.289910797346868e-05,
|
| 2637 |
+
"loss": 6.5061,
|
| 2638 |
+
"step": 3710
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 0.26936352006951314,
|
| 2642 |
+
"grad_norm": 1.25,
|
| 2643 |
+
"learning_rate": 4.285812444755779e-05,
|
| 2644 |
+
"loss": 6.5257,
|
| 2645 |
+
"step": 3720
|
| 2646 |
+
},
|
| 2647 |
+
{
|
| 2648 |
+
"epoch": 0.27008761555357097,
|
| 2649 |
+
"grad_norm": 1.1953125,
|
| 2650 |
+
"learning_rate": 4.281704269033242e-05,
|
| 2651 |
+
"loss": 6.5269,
|
| 2652 |
+
"step": 3730
|
| 2653 |
+
},
|
| 2654 |
+
{
|
| 2655 |
+
"epoch": 0.27081171103762885,
|
| 2656 |
+
"grad_norm": 1.3046875,
|
| 2657 |
+
"learning_rate": 4.2775862927769025e-05,
|
| 2658 |
+
"loss": 6.4974,
|
| 2659 |
+
"step": 3740
|
| 2660 |
+
},
|
| 2661 |
+
{
|
| 2662 |
+
"epoch": 0.2715358065216867,
|
| 2663 |
+
"grad_norm": 1.3671875,
|
| 2664 |
+
"learning_rate": 4.2734585386383086e-05,
|
| 2665 |
+
"loss": 6.4876,
|
| 2666 |
+
"step": 3750
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.2722599020057445,
|
| 2670 |
+
"grad_norm": 1.2734375,
|
| 2671 |
+
"learning_rate": 4.269321029322797e-05,
|
| 2672 |
+
"loss": 6.5099,
|
| 2673 |
+
"step": 3760
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 0.27298399748980234,
|
| 2677 |
+
"grad_norm": 1.265625,
|
| 2678 |
+
"learning_rate": 4.265173787589364e-05,
|
| 2679 |
+
"loss": 6.5222,
|
| 2680 |
+
"step": 3770
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 0.2737080929738602,
|
| 2684 |
+
"grad_norm": 1.2109375,
|
| 2685 |
+
"learning_rate": 4.2610168362505395e-05,
|
| 2686 |
+
"loss": 6.5077,
|
| 2687 |
+
"step": 3780
|
| 2688 |
+
},
|
| 2689 |
+
{
|
| 2690 |
+
"epoch": 0.274432188457918,
|
| 2691 |
+
"grad_norm": 1.203125,
|
| 2692 |
+
"learning_rate": 4.256850198172263e-05,
|
| 2693 |
+
"loss": 6.5283,
|
| 2694 |
+
"step": 3790
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 0.27515628394197583,
|
| 2698 |
+
"grad_norm": 1.171875,
|
| 2699 |
+
"learning_rate": 4.252673896273758e-05,
|
| 2700 |
+
"loss": 6.5135,
|
| 2701 |
+
"step": 3800
|
| 2702 |
+
},
|
| 2703 |
+
{
|
| 2704 |
+
"epoch": 0.27588037942603366,
|
| 2705 |
+
"grad_norm": 1.203125,
|
| 2706 |
+
"learning_rate": 4.248487953527404e-05,
|
| 2707 |
+
"loss": 6.5123,
|
| 2708 |
+
"step": 3810
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 0.2766044749100915,
|
| 2712 |
+
"grad_norm": 1.21875,
|
| 2713 |
+
"learning_rate": 4.244292392958613e-05,
|
| 2714 |
+
"loss": 6.5163,
|
| 2715 |
+
"step": 3820
|
| 2716 |
+
},
|
| 2717 |
+
{
|
| 2718 |
+
"epoch": 0.2773285703941493,
|
| 2719 |
+
"grad_norm": 1.3203125,
|
| 2720 |
+
"learning_rate": 4.2400872376457e-05,
|
| 2721 |
+
"loss": 6.5063,
|
| 2722 |
+
"step": 3830
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 0.27805266587820715,
|
| 2726 |
+
"grad_norm": 1.2734375,
|
| 2727 |
+
"learning_rate": 4.2358725107197576e-05,
|
| 2728 |
+
"loss": 6.5214,
|
| 2729 |
+
"step": 3840
|
| 2730 |
+
},
|
| 2731 |
+
{
|
| 2732 |
+
"epoch": 0.278776761362265,
|
| 2733 |
+
"grad_norm": 1.3046875,
|
| 2734 |
+
"learning_rate": 4.231648235364529e-05,
|
| 2735 |
+
"loss": 6.5375,
|
| 2736 |
+
"step": 3850
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 0.2795008568463228,
|
| 2740 |
+
"grad_norm": 1.2578125,
|
| 2741 |
+
"learning_rate": 4.227414434816279e-05,
|
| 2742 |
+
"loss": 6.5003,
|
| 2743 |
+
"step": 3860
|
| 2744 |
+
},
|
| 2745 |
+
{
|
| 2746 |
+
"epoch": 0.28022495233038064,
|
| 2747 |
+
"grad_norm": 1.234375,
|
| 2748 |
+
"learning_rate": 4.22317113236367e-05,
|
| 2749 |
+
"loss": 6.5425,
|
| 2750 |
+
"step": 3870
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 0.28094904781443847,
|
| 2754 |
+
"grad_norm": 1.1953125,
|
| 2755 |
+
"learning_rate": 4.218918351347626e-05,
|
| 2756 |
+
"loss": 6.5073,
|
| 2757 |
+
"step": 3880
|
| 2758 |
+
},
|
| 2759 |
+
{
|
| 2760 |
+
"epoch": 0.2816731432984963,
|
| 2761 |
+
"grad_norm": 1.3125,
|
| 2762 |
+
"learning_rate": 4.214656115161215e-05,
|
| 2763 |
+
"loss": 6.5127,
|
| 2764 |
+
"step": 3890
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 0.2823972387825541,
|
| 2768 |
+
"grad_norm": 1.2265625,
|
| 2769 |
+
"learning_rate": 4.210384447249509e-05,
|
| 2770 |
+
"loss": 6.5198,
|
| 2771 |
+
"step": 3900
|
| 2772 |
+
},
|
| 2773 |
+
{
|
| 2774 |
+
"epoch": 0.28312133426661196,
|
| 2775 |
+
"grad_norm": 1.234375,
|
| 2776 |
+
"learning_rate": 4.2061033711094655e-05,
|
| 2777 |
+
"loss": 6.5206,
|
| 2778 |
+
"step": 3910
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 0.2838454297506698,
|
| 2782 |
+
"grad_norm": 1.3125,
|
| 2783 |
+
"learning_rate": 4.2018129102897904e-05,
|
| 2784 |
+
"loss": 6.4952,
|
| 2785 |
+
"step": 3920
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 0.2845695252347276,
|
| 2789 |
+
"grad_norm": 1.2265625,
|
| 2790 |
+
"learning_rate": 4.197513088390813e-05,
|
| 2791 |
+
"loss": 6.5065,
|
| 2792 |
+
"step": 3930
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 0.28529362071878545,
|
| 2796 |
+
"grad_norm": 1.265625,
|
| 2797 |
+
"learning_rate": 4.193203929064353e-05,
|
| 2798 |
+
"loss": 6.4917,
|
| 2799 |
+
"step": 3940
|
| 2800 |
+
},
|
| 2801 |
+
{
|
| 2802 |
+
"epoch": 0.2860177162028433,
|
| 2803 |
+
"grad_norm": 1.28125,
|
| 2804 |
+
"learning_rate": 4.1888854560135934e-05,
|
| 2805 |
+
"loss": 6.5252,
|
| 2806 |
+
"step": 3950
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 0.2867418116869011,
|
| 2810 |
+
"grad_norm": 1.2578125,
|
| 2811 |
+
"learning_rate": 4.1845576929929486e-05,
|
| 2812 |
+
"loss": 6.4974,
|
| 2813 |
+
"step": 3960
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 0.28746590717095893,
|
| 2817 |
+
"grad_norm": 1.234375,
|
| 2818 |
+
"learning_rate": 4.180220663807934e-05,
|
| 2819 |
+
"loss": 6.5253,
|
| 2820 |
+
"step": 3970
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 0.28819000265501676,
|
| 2824 |
+
"grad_norm": 1.265625,
|
| 2825 |
+
"learning_rate": 4.175874392315033e-05,
|
| 2826 |
+
"loss": 6.4986,
|
| 2827 |
+
"step": 3980
|
| 2828 |
+
},
|
| 2829 |
+
{
|
| 2830 |
+
"epoch": 0.2889140981390746,
|
| 2831 |
+
"grad_norm": 1.3203125,
|
| 2832 |
+
"learning_rate": 4.1715189024215716e-05,
|
| 2833 |
+
"loss": 6.5147,
|
| 2834 |
+
"step": 3990
|
| 2835 |
}
|
| 2836 |
],
|
| 2837 |
"logging_steps": 10,
|
|
|
|
| 2851 |
"attributes": {}
|
| 2852 |
}
|
| 2853 |
},
|
| 2854 |
+
"total_flos": 8.601164131082437e+18,
|
| 2855 |
"train_batch_size": 30,
|
| 2856 |
"trial_name": null,
|
| 2857 |
"trial_params": null
|