Training in progress, epoch 5, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 37789864
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:52b37145078d88e9a343bd22617f50ead8d019b0aed1436bbb79f1fd49c66b6e
|
| 3 |
size 37789864
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2622266
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4bcc4bfa26449a669fcf4e2bd2006218fa37d9141b3c0e6ed1720dba59a9fd65
|
| 3 |
size 2622266
|
last-checkpoint/rng_state.pth
CHANGED
|
Binary files a/last-checkpoint/rng_state.pth and b/last-checkpoint/rng_state.pth differ
|
|
|
last-checkpoint/scheduler.pt
CHANGED
|
Binary files a/last-checkpoint/scheduler.pt and b/last-checkpoint/scheduler.pt differ
|
|
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "dq158/coqui/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2015,13 +2015,518 @@
|
|
| 2015 |
"eval_steps_per_second": 0.544,
|
| 2016 |
"eval_translation_length": 4591104,
|
| 2017 |
"step": 161392
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2018 |
}
|
| 2019 |
],
|
| 2020 |
"logging_steps": 500,
|
| 2021 |
"max_steps": 1210440,
|
| 2022 |
"num_train_epochs": 30,
|
| 2023 |
"save_steps": 1000,
|
| 2024 |
-
"total_flos":
|
| 2025 |
"trial_name": null,
|
| 2026 |
"trial_params": null
|
| 2027 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.995251417160034,
|
| 3 |
+
"best_model_checkpoint": "dq158/coqui/checkpoint-201740",
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 201740,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2015 |
"eval_steps_per_second": 0.544,
|
| 2016 |
"eval_translation_length": 4591104,
|
| 2017 |
"step": 161392
|
| 2018 |
+
},
|
| 2019 |
+
{
|
| 2020 |
+
"epoch": 4.0,
|
| 2021 |
+
"learning_rate": 7.655554846753297e-05,
|
| 2022 |
+
"loss": 3.1442,
|
| 2023 |
+
"step": 161500
|
| 2024 |
+
},
|
| 2025 |
+
{
|
| 2026 |
+
"epoch": 4.02,
|
| 2027 |
+
"learning_rate": 7.653443602624745e-05,
|
| 2028 |
+
"loss": 3.079,
|
| 2029 |
+
"step": 162000
|
| 2030 |
+
},
|
| 2031 |
+
{
|
| 2032 |
+
"epoch": 4.03,
|
| 2033 |
+
"learning_rate": 7.651326200854252e-05,
|
| 2034 |
+
"loss": 3.1155,
|
| 2035 |
+
"step": 162500
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 4.04,
|
| 2039 |
+
"learning_rate": 7.64920264501056e-05,
|
| 2040 |
+
"loss": 3.034,
|
| 2041 |
+
"step": 163000
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 4.05,
|
| 2045 |
+
"learning_rate": 7.647072938672785e-05,
|
| 2046 |
+
"loss": 3.1226,
|
| 2047 |
+
"step": 163500
|
| 2048 |
+
},
|
| 2049 |
+
{
|
| 2050 |
+
"epoch": 4.06,
|
| 2051 |
+
"learning_rate": 7.644937085430409e-05,
|
| 2052 |
+
"loss": 3.1027,
|
| 2053 |
+
"step": 164000
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 4.08,
|
| 2057 |
+
"learning_rate": 7.642795088883274e-05,
|
| 2058 |
+
"loss": 3.1736,
|
| 2059 |
+
"step": 164500
|
| 2060 |
+
},
|
| 2061 |
+
{
|
| 2062 |
+
"epoch": 4.09,
|
| 2063 |
+
"learning_rate": 7.640646952641577e-05,
|
| 2064 |
+
"loss": 2.9279,
|
| 2065 |
+
"step": 165000
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 4.1,
|
| 2069 |
+
"learning_rate": 7.638492680325862e-05,
|
| 2070 |
+
"loss": 3.0596,
|
| 2071 |
+
"step": 165500
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 4.11,
|
| 2075 |
+
"learning_rate": 7.636332275567012e-05,
|
| 2076 |
+
"loss": 3.0027,
|
| 2077 |
+
"step": 166000
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 4.13,
|
| 2081 |
+
"learning_rate": 7.634165742006251e-05,
|
| 2082 |
+
"loss": 3.1487,
|
| 2083 |
+
"step": 166500
|
| 2084 |
+
},
|
| 2085 |
+
{
|
| 2086 |
+
"epoch": 4.14,
|
| 2087 |
+
"learning_rate": 7.631993083295134e-05,
|
| 2088 |
+
"loss": 3.04,
|
| 2089 |
+
"step": 167000
|
| 2090 |
+
},
|
| 2091 |
+
{
|
| 2092 |
+
"epoch": 4.15,
|
| 2093 |
+
"learning_rate": 7.62981430309553e-05,
|
| 2094 |
+
"loss": 3.0355,
|
| 2095 |
+
"step": 167500
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 4.16,
|
| 2099 |
+
"learning_rate": 7.627629405079637e-05,
|
| 2100 |
+
"loss": 3.1222,
|
| 2101 |
+
"step": 168000
|
| 2102 |
+
},
|
| 2103 |
+
{
|
| 2104 |
+
"epoch": 4.18,
|
| 2105 |
+
"learning_rate": 7.625438392929956e-05,
|
| 2106 |
+
"loss": 3.0185,
|
| 2107 |
+
"step": 168500
|
| 2108 |
+
},
|
| 2109 |
+
{
|
| 2110 |
+
"epoch": 4.19,
|
| 2111 |
+
"learning_rate": 7.623241270339294e-05,
|
| 2112 |
+
"loss": 3.036,
|
| 2113 |
+
"step": 169000
|
| 2114 |
+
},
|
| 2115 |
+
{
|
| 2116 |
+
"epoch": 4.2,
|
| 2117 |
+
"learning_rate": 7.621038041010763e-05,
|
| 2118 |
+
"loss": 3.0182,
|
| 2119 |
+
"step": 169500
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 4.21,
|
| 2123 |
+
"learning_rate": 7.61882870865776e-05,
|
| 2124 |
+
"loss": 3.0501,
|
| 2125 |
+
"step": 170000
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 4.23,
|
| 2129 |
+
"learning_rate": 7.61661327700397e-05,
|
| 2130 |
+
"loss": 3.0935,
|
| 2131 |
+
"step": 170500
|
| 2132 |
+
},
|
| 2133 |
+
{
|
| 2134 |
+
"epoch": 4.24,
|
| 2135 |
+
"learning_rate": 7.614391749783361e-05,
|
| 2136 |
+
"loss": 2.9745,
|
| 2137 |
+
"step": 171000
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 4.25,
|
| 2141 |
+
"learning_rate": 7.612164130740175e-05,
|
| 2142 |
+
"loss": 3.0497,
|
| 2143 |
+
"step": 171500
|
| 2144 |
+
},
|
| 2145 |
+
{
|
| 2146 |
+
"epoch": 4.26,
|
| 2147 |
+
"learning_rate": 7.609930423628915e-05,
|
| 2148 |
+
"loss": 3.0207,
|
| 2149 |
+
"step": 172000
|
| 2150 |
+
},
|
| 2151 |
+
{
|
| 2152 |
+
"epoch": 4.28,
|
| 2153 |
+
"learning_rate": 7.607690632214351e-05,
|
| 2154 |
+
"loss": 3.0059,
|
| 2155 |
+
"step": 172500
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 4.29,
|
| 2159 |
+
"learning_rate": 7.605444760271507e-05,
|
| 2160 |
+
"loss": 3.113,
|
| 2161 |
+
"step": 173000
|
| 2162 |
+
},
|
| 2163 |
+
{
|
| 2164 |
+
"epoch": 4.3,
|
| 2165 |
+
"learning_rate": 7.603192811585654e-05,
|
| 2166 |
+
"loss": 3.0989,
|
| 2167 |
+
"step": 173500
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 4.31,
|
| 2171 |
+
"learning_rate": 7.600934789952304e-05,
|
| 2172 |
+
"loss": 3.1174,
|
| 2173 |
+
"step": 174000
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 4.32,
|
| 2177 |
+
"learning_rate": 7.598670699177207e-05,
|
| 2178 |
+
"loss": 3.0884,
|
| 2179 |
+
"step": 174500
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 4.34,
|
| 2183 |
+
"learning_rate": 7.596400543076339e-05,
|
| 2184 |
+
"loss": 2.973,
|
| 2185 |
+
"step": 175000
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 4.35,
|
| 2189 |
+
"learning_rate": 7.594124325475904e-05,
|
| 2190 |
+
"loss": 3.0956,
|
| 2191 |
+
"step": 175500
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 4.36,
|
| 2195 |
+
"learning_rate": 7.591842050212317e-05,
|
| 2196 |
+
"loss": 2.9274,
|
| 2197 |
+
"step": 176000
|
| 2198 |
+
},
|
| 2199 |
+
{
|
| 2200 |
+
"epoch": 4.37,
|
| 2201 |
+
"learning_rate": 7.589553721132205e-05,
|
| 2202 |
+
"loss": 3.0128,
|
| 2203 |
+
"step": 176500
|
| 2204 |
+
},
|
| 2205 |
+
{
|
| 2206 |
+
"epoch": 4.39,
|
| 2207 |
+
"learning_rate": 7.587259342092397e-05,
|
| 2208 |
+
"loss": 3.0429,
|
| 2209 |
+
"step": 177000
|
| 2210 |
+
},
|
| 2211 |
+
{
|
| 2212 |
+
"epoch": 4.4,
|
| 2213 |
+
"learning_rate": 7.584958916959923e-05,
|
| 2214 |
+
"loss": 3.0955,
|
| 2215 |
+
"step": 177500
|
| 2216 |
+
},
|
| 2217 |
+
{
|
| 2218 |
+
"epoch": 4.41,
|
| 2219 |
+
"learning_rate": 7.582652449611996e-05,
|
| 2220 |
+
"loss": 3.1124,
|
| 2221 |
+
"step": 178000
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 4.42,
|
| 2225 |
+
"learning_rate": 7.58033994393602e-05,
|
| 2226 |
+
"loss": 2.9723,
|
| 2227 |
+
"step": 178500
|
| 2228 |
+
},
|
| 2229 |
+
{
|
| 2230 |
+
"epoch": 4.44,
|
| 2231 |
+
"learning_rate": 7.578021403829572e-05,
|
| 2232 |
+
"loss": 2.981,
|
| 2233 |
+
"step": 179000
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 4.45,
|
| 2237 |
+
"learning_rate": 7.5756968332004e-05,
|
| 2238 |
+
"loss": 3.1174,
|
| 2239 |
+
"step": 179500
|
| 2240 |
+
},
|
| 2241 |
+
{
|
| 2242 |
+
"epoch": 4.46,
|
| 2243 |
+
"learning_rate": 7.57336623596642e-05,
|
| 2244 |
+
"loss": 3.0292,
|
| 2245 |
+
"step": 180000
|
| 2246 |
+
},
|
| 2247 |
+
{
|
| 2248 |
+
"epoch": 4.47,
|
| 2249 |
+
"learning_rate": 7.5710296160557e-05,
|
| 2250 |
+
"loss": 3.0106,
|
| 2251 |
+
"step": 180500
|
| 2252 |
+
},
|
| 2253 |
+
{
|
| 2254 |
+
"epoch": 4.49,
|
| 2255 |
+
"learning_rate": 7.568686977406459e-05,
|
| 2256 |
+
"loss": 2.9749,
|
| 2257 |
+
"step": 181000
|
| 2258 |
+
},
|
| 2259 |
+
{
|
| 2260 |
+
"epoch": 4.5,
|
| 2261 |
+
"learning_rate": 7.566338323967065e-05,
|
| 2262 |
+
"loss": 3.0591,
|
| 2263 |
+
"step": 181500
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 4.51,
|
| 2267 |
+
"learning_rate": 7.563983659696022e-05,
|
| 2268 |
+
"loss": 3.0228,
|
| 2269 |
+
"step": 182000
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 4.52,
|
| 2273 |
+
"learning_rate": 7.56162298856196e-05,
|
| 2274 |
+
"loss": 3.0134,
|
| 2275 |
+
"step": 182500
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 4.54,
|
| 2279 |
+
"learning_rate": 7.559256314543639e-05,
|
| 2280 |
+
"loss": 3.0624,
|
| 2281 |
+
"step": 183000
|
| 2282 |
+
},
|
| 2283 |
+
{
|
| 2284 |
+
"epoch": 4.55,
|
| 2285 |
+
"learning_rate": 7.556883641629936e-05,
|
| 2286 |
+
"loss": 3.0205,
|
| 2287 |
+
"step": 183500
|
| 2288 |
+
},
|
| 2289 |
+
{
|
| 2290 |
+
"epoch": 4.56,
|
| 2291 |
+
"learning_rate": 7.554504973819835e-05,
|
| 2292 |
+
"loss": 3.0067,
|
| 2293 |
+
"step": 184000
|
| 2294 |
+
},
|
| 2295 |
+
{
|
| 2296 |
+
"epoch": 4.57,
|
| 2297 |
+
"learning_rate": 7.552120315122426e-05,
|
| 2298 |
+
"loss": 3.0488,
|
| 2299 |
+
"step": 184500
|
| 2300 |
+
},
|
| 2301 |
+
{
|
| 2302 |
+
"epoch": 4.59,
|
| 2303 |
+
"learning_rate": 7.549729669556898e-05,
|
| 2304 |
+
"loss": 3.0992,
|
| 2305 |
+
"step": 185000
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 4.6,
|
| 2309 |
+
"learning_rate": 7.547333041152526e-05,
|
| 2310 |
+
"loss": 3.0137,
|
| 2311 |
+
"step": 185500
|
| 2312 |
+
},
|
| 2313 |
+
{
|
| 2314 |
+
"epoch": 4.61,
|
| 2315 |
+
"learning_rate": 7.544930433948676e-05,
|
| 2316 |
+
"loss": 3.1379,
|
| 2317 |
+
"step": 186000
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 4.62,
|
| 2321 |
+
"learning_rate": 7.542521851994781e-05,
|
| 2322 |
+
"loss": 3.0818,
|
| 2323 |
+
"step": 186500
|
| 2324 |
+
},
|
| 2325 |
+
{
|
| 2326 |
+
"epoch": 4.63,
|
| 2327 |
+
"learning_rate": 7.540107299350354e-05,
|
| 2328 |
+
"loss": 3.0634,
|
| 2329 |
+
"step": 187000
|
| 2330 |
+
},
|
| 2331 |
+
{
|
| 2332 |
+
"epoch": 4.65,
|
| 2333 |
+
"learning_rate": 7.537686780084966e-05,
|
| 2334 |
+
"loss": 3.0984,
|
| 2335 |
+
"step": 187500
|
| 2336 |
+
},
|
| 2337 |
+
{
|
| 2338 |
+
"epoch": 4.66,
|
| 2339 |
+
"learning_rate": 7.53526029827824e-05,
|
| 2340 |
+
"loss": 3.1238,
|
| 2341 |
+
"step": 188000
|
| 2342 |
+
},
|
| 2343 |
+
{
|
| 2344 |
+
"epoch": 4.67,
|
| 2345 |
+
"learning_rate": 7.532827858019862e-05,
|
| 2346 |
+
"loss": 3.0431,
|
| 2347 |
+
"step": 188500
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 4.68,
|
| 2351 |
+
"learning_rate": 7.530389463409545e-05,
|
| 2352 |
+
"loss": 3.0216,
|
| 2353 |
+
"step": 189000
|
| 2354 |
+
},
|
| 2355 |
+
{
|
| 2356 |
+
"epoch": 4.7,
|
| 2357 |
+
"learning_rate": 7.527945118557048e-05,
|
| 2358 |
+
"loss": 3.0448,
|
| 2359 |
+
"step": 189500
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 4.71,
|
| 2363 |
+
"learning_rate": 7.525494827582155e-05,
|
| 2364 |
+
"loss": 3.1713,
|
| 2365 |
+
"step": 190000
|
| 2366 |
+
},
|
| 2367 |
+
{
|
| 2368 |
+
"epoch": 4.72,
|
| 2369 |
+
"learning_rate": 7.523038594614671e-05,
|
| 2370 |
+
"loss": 3.0396,
|
| 2371 |
+
"step": 190500
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 4.73,
|
| 2375 |
+
"learning_rate": 7.52057642379442e-05,
|
| 2376 |
+
"loss": 3.1008,
|
| 2377 |
+
"step": 191000
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 4.75,
|
| 2381 |
+
"learning_rate": 7.518108319271228e-05,
|
| 2382 |
+
"loss": 3.0965,
|
| 2383 |
+
"step": 191500
|
| 2384 |
+
},
|
| 2385 |
+
{
|
| 2386 |
+
"epoch": 4.76,
|
| 2387 |
+
"learning_rate": 7.515634285204928e-05,
|
| 2388 |
+
"loss": 3.0407,
|
| 2389 |
+
"step": 192000
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 4.77,
|
| 2393 |
+
"learning_rate": 7.51315432576534e-05,
|
| 2394 |
+
"loss": 3.0669,
|
| 2395 |
+
"step": 192500
|
| 2396 |
+
},
|
| 2397 |
+
{
|
| 2398 |
+
"epoch": 4.78,
|
| 2399 |
+
"learning_rate": 7.510668445132279e-05,
|
| 2400 |
+
"loss": 3.0752,
|
| 2401 |
+
"step": 193000
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 4.8,
|
| 2405 |
+
"learning_rate": 7.508176647495532e-05,
|
| 2406 |
+
"loss": 2.9414,
|
| 2407 |
+
"step": 193500
|
| 2408 |
+
},
|
| 2409 |
+
{
|
| 2410 |
+
"epoch": 4.81,
|
| 2411 |
+
"learning_rate": 7.505678937054863e-05,
|
| 2412 |
+
"loss": 3.0746,
|
| 2413 |
+
"step": 194000
|
| 2414 |
+
},
|
| 2415 |
+
{
|
| 2416 |
+
"epoch": 4.82,
|
| 2417 |
+
"learning_rate": 7.503175318019999e-05,
|
| 2418 |
+
"loss": 2.9645,
|
| 2419 |
+
"step": 194500
|
| 2420 |
+
},
|
| 2421 |
+
{
|
| 2422 |
+
"epoch": 4.83,
|
| 2423 |
+
"learning_rate": 7.500665794610632e-05,
|
| 2424 |
+
"loss": 3.0942,
|
| 2425 |
+
"step": 195000
|
| 2426 |
+
},
|
| 2427 |
+
{
|
| 2428 |
+
"epoch": 4.85,
|
| 2429 |
+
"learning_rate": 7.498150371056396e-05,
|
| 2430 |
+
"loss": 3.0586,
|
| 2431 |
+
"step": 195500
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 4.86,
|
| 2435 |
+
"learning_rate": 7.495629051596876e-05,
|
| 2436 |
+
"loss": 3.0251,
|
| 2437 |
+
"step": 196000
|
| 2438 |
+
},
|
| 2439 |
+
{
|
| 2440 |
+
"epoch": 4.87,
|
| 2441 |
+
"learning_rate": 7.493101840481594e-05,
|
| 2442 |
+
"loss": 3.0769,
|
| 2443 |
+
"step": 196500
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 4.88,
|
| 2447 |
+
"learning_rate": 7.490568741969997e-05,
|
| 2448 |
+
"loss": 3.0701,
|
| 2449 |
+
"step": 197000
|
| 2450 |
+
},
|
| 2451 |
+
{
|
| 2452 |
+
"epoch": 4.89,
|
| 2453 |
+
"learning_rate": 7.488029760331459e-05,
|
| 2454 |
+
"loss": 3.0893,
|
| 2455 |
+
"step": 197500
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 4.91,
|
| 2459 |
+
"learning_rate": 7.48548489984527e-05,
|
| 2460 |
+
"loss": 3.1527,
|
| 2461 |
+
"step": 198000
|
| 2462 |
+
},
|
| 2463 |
+
{
|
| 2464 |
+
"epoch": 4.92,
|
| 2465 |
+
"learning_rate": 7.482934164800626e-05,
|
| 2466 |
+
"loss": 3.0497,
|
| 2467 |
+
"step": 198500
|
| 2468 |
+
},
|
| 2469 |
+
{
|
| 2470 |
+
"epoch": 4.93,
|
| 2471 |
+
"learning_rate": 7.480377559496624e-05,
|
| 2472 |
+
"loss": 3.0665,
|
| 2473 |
+
"step": 199000
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 4.94,
|
| 2477 |
+
"learning_rate": 7.47781508824226e-05,
|
| 2478 |
+
"loss": 3.0842,
|
| 2479 |
+
"step": 199500
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 4.96,
|
| 2483 |
+
"learning_rate": 7.47524675535641e-05,
|
| 2484 |
+
"loss": 3.0961,
|
| 2485 |
+
"step": 200000
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 4.97,
|
| 2489 |
+
"learning_rate": 7.472672565167833e-05,
|
| 2490 |
+
"loss": 3.0756,
|
| 2491 |
+
"step": 200500
|
| 2492 |
+
},
|
| 2493 |
+
{
|
| 2494 |
+
"epoch": 4.98,
|
| 2495 |
+
"learning_rate": 7.470092522015158e-05,
|
| 2496 |
+
"loss": 3.1524,
|
| 2497 |
+
"step": 201000
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 4.99,
|
| 2501 |
+
"learning_rate": 7.467506630246882e-05,
|
| 2502 |
+
"loss": 3.1614,
|
| 2503 |
+
"step": 201500
|
| 2504 |
+
},
|
| 2505 |
+
{
|
| 2506 |
+
"epoch": 5.0,
|
| 2507 |
+
"eval_bleu": 1.0,
|
| 2508 |
+
"eval_brevity_penalty": 1.0,
|
| 2509 |
+
"eval_length_ratio": 1.0,
|
| 2510 |
+
"eval_loss": 2.995251417160034,
|
| 2511 |
+
"eval_precisions": [
|
| 2512 |
+
1.0,
|
| 2513 |
+
1.0,
|
| 2514 |
+
1.0,
|
| 2515 |
+
1.0
|
| 2516 |
+
],
|
| 2517 |
+
"eval_reference_length": 4591104,
|
| 2518 |
+
"eval_runtime": 8351.4463,
|
| 2519 |
+
"eval_samples_per_second": 1.074,
|
| 2520 |
+
"eval_steps_per_second": 0.537,
|
| 2521 |
+
"eval_translation_length": 4591104,
|
| 2522 |
+
"step": 201740
|
| 2523 |
}
|
| 2524 |
],
|
| 2525 |
"logging_steps": 500,
|
| 2526 |
"max_steps": 1210440,
|
| 2527 |
"num_train_epochs": 30,
|
| 2528 |
"save_steps": 1000,
|
| 2529 |
+
"total_flos": 3.462342804976435e+18,
|
| 2530 |
"trial_name": null,
|
| 2531 |
"trial_params": null
|
| 2532 |
}
|