Upload folder using huggingface_hub
Browse files- .gitattributes +1 -0
- checkpoints-v3.0-discrete/checkpoint-20480/eval_state.json +3 -0
- checkpoints-v3.0-discrete/checkpoint-20480/model.safetensors +3 -0
- checkpoints-v3.0-discrete/checkpoint-20480/optimizer.pt +3 -0
- checkpoints-v3.0-discrete/checkpoint-20480/rng_state.pth +3 -0
- checkpoints-v3.0-discrete/checkpoint-20480/scaler.pt +3 -0
- checkpoints-v3.0-discrete/checkpoint-20480/scheduler.pt +3 -0
- checkpoints-v3.0-discrete/checkpoint-20480/trainer_state.json +679 -0
- checkpoints-v3.0-discrete/checkpoint-20480/training_args.bin +3 -0
.gitattributes
CHANGED
|
@@ -89,3 +89,4 @@ checkpoints-d3.2/checkpoint-12288/eval_state.json filter=lfs diff=lfs merge=lfs
|
|
| 89 |
checkpoints-v1.0-discrete/checkpoint-21504/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
| 90 |
checkpoints-v1.0-discrete/checkpoint-21504-2/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
| 91 |
checkpoints-v1.1-discrete/checkpoint-21504/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 89 |
checkpoints-v1.0-discrete/checkpoint-21504/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
| 90 |
checkpoints-v1.0-discrete/checkpoint-21504-2/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
| 91 |
checkpoints-v1.1-discrete/checkpoint-21504/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
| 92 |
+
checkpoints-v3.0-discrete/checkpoint-20480/eval_state.json filter=lfs diff=lfs merge=lfs -text
|
checkpoints-v3.0-discrete/checkpoint-20480/eval_state.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e5bf49b9b11678346ca731913ba552ea24d6e3e390805d09878bdc11fe2f9e64
|
| 3 |
+
size 55222547
|
checkpoints-v3.0-discrete/checkpoint-20480/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d15aa39217fdd991a5567e121db82d4e4f1a373eaa1e6de3146ee2acaf7689c5
|
| 3 |
+
size 24033312
|
checkpoints-v3.0-discrete/checkpoint-20480/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68897590fbd3f1d0f97cfb7f36cb732e0ac8cd24a9d961106cee4db24e18055d
|
| 3 |
+
size 1641995
|
checkpoints-v3.0-discrete/checkpoint-20480/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:37114774d777269916aa3cb7a26ea0cbdc20a3ba2c9e77c2e8c673303b513aba
|
| 3 |
+
size 14645
|
checkpoints-v3.0-discrete/checkpoint-20480/scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2d84737ca1f2a52ae6a937e7f8dd7b5e2b71bed6fae33e8f88369e6f1f1874c
|
| 3 |
+
size 1383
|
checkpoints-v3.0-discrete/checkpoint-20480/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:205d765e13e1be2966644dd0f0eaaf87a8a81b7f8860b6a9dc2ebf25b30616e3
|
| 3 |
+
size 1465
|
checkpoints-v3.0-discrete/checkpoint-20480/trainer_state.json
ADDED
|
@@ -0,0 +1,679 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.945914738349268,
|
| 6 |
+
"eval_steps": 4096,
|
| 7 |
+
"global_step": 20480,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.011823934229365849,
|
| 14 |
+
"grad_norm": 2.4639666080474854,
|
| 15 |
+
"learning_rate": 0.000498046875,
|
| 16 |
+
"loss": 0.1339670717716217,
|
| 17 |
+
"step": 256
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.023647868458731697,
|
| 21 |
+
"grad_norm": 1.5375629663467407,
|
| 22 |
+
"learning_rate": 0.000998046875,
|
| 23 |
+
"loss": 0.1467132717370987,
|
| 24 |
+
"step": 512
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.03547180268809755,
|
| 28 |
+
"grad_norm": 1.506445288658142,
|
| 29 |
+
"learning_rate": 0.000999640996023194,
|
| 30 |
+
"loss": 0.15337222814559937,
|
| 31 |
+
"step": 768
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.047295736917463395,
|
| 35 |
+
"grad_norm": 1.641593098640442,
|
| 36 |
+
"learning_rate": 0.0009985588674043958,
|
| 37 |
+
"loss": 0.15139617025852203,
|
| 38 |
+
"step": 1024
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.05911967114682925,
|
| 42 |
+
"grad_norm": 1.3137743473052979,
|
| 43 |
+
"learning_rate": 0.0009967551747861387,
|
| 44 |
+
"loss": 0.14983117580413818,
|
| 45 |
+
"step": 1280
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.0709436053761951,
|
| 49 |
+
"grad_norm": 1.3725732564926147,
|
| 50 |
+
"learning_rate": 0.000994232528651847,
|
| 51 |
+
"loss": 0.14979030191898346,
|
| 52 |
+
"step": 1536
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.08276753960556095,
|
| 56 |
+
"grad_norm": 2.0527191162109375,
|
| 57 |
+
"learning_rate": 0.0009909945800260092,
|
| 58 |
+
"loss": 0.14714637398719788,
|
| 59 |
+
"step": 1792
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.09459147383492679,
|
| 63 |
+
"grad_norm": 1.8648000955581665,
|
| 64 |
+
"learning_rate": 0.0009870460151900522,
|
| 65 |
+
"loss": 0.1472843438386917,
|
| 66 |
+
"step": 2048
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.10641540806429264,
|
| 70 |
+
"grad_norm": 1.025286078453064,
|
| 71 |
+
"learning_rate": 0.0009823925488998885,
|
| 72 |
+
"loss": 0.1452621966600418,
|
| 73 |
+
"step": 2304
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.1182393422936585,
|
| 77 |
+
"grad_norm": 0.9860194325447083,
|
| 78 |
+
"learning_rate": 0.0009770409161149525,
|
| 79 |
+
"loss": 0.1464366316795349,
|
| 80 |
+
"step": 2560
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.13006327652302435,
|
| 84 |
+
"grad_norm": 1.4449650049209595,
|
| 85 |
+
"learning_rate": 0.0009709988622506973,
|
| 86 |
+
"loss": 0.1476413458585739,
|
| 87 |
+
"step": 2816
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.1418872107523902,
|
| 91 |
+
"grad_norm": 1.548802375793457,
|
| 92 |
+
"learning_rate": 0.000964275131968659,
|
| 93 |
+
"loss": 0.14479056000709534,
|
| 94 |
+
"step": 3072
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.15371114498175603,
|
| 98 |
+
"grad_norm": 0.9572728872299194,
|
| 99 |
+
"learning_rate": 0.0009568794565203123,
|
| 100 |
+
"loss": 0.14605838060379028,
|
| 101 |
+
"step": 3328
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.1655350792111219,
|
| 105 |
+
"grad_norm": 1.05500066280365,
|
| 106 |
+
"learning_rate": 0.0009488225396630347,
|
| 107 |
+
"loss": 0.14362677931785583,
|
| 108 |
+
"step": 3584
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.17735901344048774,
|
| 112 |
+
"grad_norm": 0.7869608402252197,
|
| 113 |
+
"learning_rate": 0.0009401160421685646,
|
| 114 |
+
"loss": 0.14176346361637115,
|
| 115 |
+
"step": 3840
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.18918294766985358,
|
| 119 |
+
"grad_norm": 0.8075957298278809,
|
| 120 |
+
"learning_rate": 0.0009307725649463714,
|
| 121 |
+
"loss": 0.14215654134750366,
|
| 122 |
+
"step": 4096
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.18918294766985358,
|
| 126 |
+
"eval_bleu": 0.0007156187561984728,
|
| 127 |
+
"eval_loss": 0.1407095858291404,
|
| 128 |
+
"eval_mse_loss": 0.1407095858291404,
|
| 129 |
+
"step": 4096
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.18918294766985358,
|
| 133 |
+
"eval_bleu": 0.0007156187561984728,
|
| 134 |
+
"eval_loss": 0.1407095858291404,
|
| 135 |
+
"eval_mse_loss": 0.1407095858291404,
|
| 136 |
+
"eval_runtime": 172.648,
|
| 137 |
+
"eval_samples_per_second": 162.139,
|
| 138 |
+
"eval_steps_per_second": 2.537,
|
| 139 |
+
"step": 4096
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.20100688189921945,
|
| 143 |
+
"grad_norm": 1.2594881057739258,
|
| 144 |
+
"learning_rate": 0.0009208056308063659,
|
| 145 |
+
"loss": 0.1409526914358139,
|
| 146 |
+
"step": 4352
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.2128308161285853,
|
| 150 |
+
"grad_norm": 1.2300289869308472,
|
| 151 |
+
"learning_rate": 0.0009102296648873445,
|
| 152 |
+
"loss": 0.14084666967391968,
|
| 153 |
+
"step": 4608
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.22465475035795113,
|
| 157 |
+
"grad_norm": 1.4511398077011108,
|
| 158 |
+
"learning_rate": 0.0008990599737794927,
|
| 159 |
+
"loss": 0.13930904865264893,
|
| 160 |
+
"step": 4864
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.236478684587317,
|
| 164 |
+
"grad_norm": 1.1964119672775269,
|
| 165 |
+
"learning_rate": 0.0008873127233711644,
|
| 166 |
+
"loss": 0.13842026889324188,
|
| 167 |
+
"step": 5120
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.24830261881668284,
|
| 171 |
+
"grad_norm": 0.9578610062599182,
|
| 172 |
+
"learning_rate": 0.0008750049154520011,
|
| 173 |
+
"loss": 0.1392611265182495,
|
| 174 |
+
"step": 5376
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.2601265530460487,
|
| 178 |
+
"grad_norm": 0.9959284067153931,
|
| 179 |
+
"learning_rate": 0.0008621543631062487,
|
| 180 |
+
"loss": 0.1368764489889145,
|
| 181 |
+
"step": 5632
|
| 182 |
+
},
|
| 183 |
+
{
|
| 184 |
+
"epoch": 0.27195048727541454,
|
| 185 |
+
"grad_norm": 1.342193603515625,
|
| 186 |
+
"learning_rate": 0.0008487796649318904,
|
| 187 |
+
"loss": 0.13701793551445007,
|
| 188 |
+
"step": 5888
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.2837744215047804,
|
| 192 |
+
"grad_norm": 0.5825753211975098,
|
| 193 |
+
"learning_rate": 0.0008349001781229053,
|
| 194 |
+
"loss": 0.13572891056537628,
|
| 195 |
+
"step": 6144
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.2955983557341462,
|
| 199 |
+
"grad_norm": 1.275922417640686,
|
| 200 |
+
"learning_rate": 0.0008205359904536107,
|
| 201 |
+
"loss": 0.134858176112175,
|
| 202 |
+
"step": 6400
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.30742228996351206,
|
| 206 |
+
"grad_norm": 1.0904085636138916,
|
| 207 |
+
"learning_rate": 0.0008057078912056363,
|
| 208 |
+
"loss": 0.13740675151348114,
|
| 209 |
+
"step": 6656
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.3192462241928779,
|
| 213 |
+
"grad_norm": 0.7365452647209167,
|
| 214 |
+
"learning_rate": 0.0007904373410796086,
|
| 215 |
+
"loss": 0.13476549088954926,
|
| 216 |
+
"step": 6912
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.3310701584222438,
|
| 220 |
+
"grad_norm": 0.9742300510406494,
|
| 221 |
+
"learning_rate": 0.0007747464411350876,
|
| 222 |
+
"loss": 0.1339239925146103,
|
| 223 |
+
"step": 7168
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.34289409265160964,
|
| 227 |
+
"grad_norm": 1.1681714057922363,
|
| 228 |
+
"learning_rate": 0.000758657900803716,
|
| 229 |
+
"loss": 0.1327570527791977,
|
| 230 |
+
"step": 7424
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.3547180268809755,
|
| 234 |
+
"grad_norm": 1.0932402610778809,
|
| 235 |
+
"learning_rate": 0.000742195005021869,
|
| 236 |
+
"loss": 0.13453976809978485,
|
| 237 |
+
"step": 7680
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.3665419611103413,
|
| 241 |
+
"grad_norm": 1.7933037281036377,
|
| 242 |
+
"learning_rate": 0.0007253815805303786,
|
| 243 |
+
"loss": 0.1339324563741684,
|
| 244 |
+
"step": 7936
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.37836589533970716,
|
| 248 |
+
"grad_norm": 1.258001446723938,
|
| 249 |
+
"learning_rate": 0.0007082419613901028,
|
| 250 |
+
"loss": 0.13321880996227264,
|
| 251 |
+
"step": 8192
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.37836589533970716,
|
| 255 |
+
"eval_bleu": 0.0009159004592492724,
|
| 256 |
+
"eval_loss": 0.13241331889953242,
|
| 257 |
+
"eval_mse_loss": 0.13241331889953242,
|
| 258 |
+
"step": 8192
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.37836589533970716,
|
| 262 |
+
"eval_bleu": 0.0009159004592492724,
|
| 263 |
+
"eval_loss": 0.13241331889953242,
|
| 264 |
+
"eval_mse_loss": 0.13241331889953242,
|
| 265 |
+
"eval_runtime": 171.6296,
|
| 266 |
+
"eval_samples_per_second": 163.101,
|
| 267 |
+
"eval_steps_per_second": 2.552,
|
| 268 |
+
"step": 8192
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.390189829569073,
|
| 272 |
+
"grad_norm": 2.0500435829162598,
|
| 273 |
+
"learning_rate": 0.0006908009537632514,
|
| 274 |
+
"loss": 0.13275721669197083,
|
| 275 |
+
"step": 8448
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.4020137637984389,
|
| 279 |
+
"grad_norm": 1.0164549350738525,
|
| 280 |
+
"learning_rate": 0.0006730838000114403,
|
| 281 |
+
"loss": 0.13082970678806305,
|
| 282 |
+
"step": 8704
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.41383769802780473,
|
| 286 |
+
"grad_norm": 0.9163410663604736,
|
| 287 |
+
"learning_rate": 0.0006551161421624341,
|
| 288 |
+
"loss": 0.1323855221271515,
|
| 289 |
+
"step": 8960
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.4256616322571706,
|
| 293 |
+
"grad_norm": 0.8464987874031067,
|
| 294 |
+
"learning_rate": 0.0006369239847984517,
|
| 295 |
+
"loss": 0.12907665967941284,
|
| 296 |
+
"step": 9216
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.4374855664865364,
|
| 300 |
+
"grad_norm": 0.790055513381958,
|
| 301 |
+
"learning_rate": 0.0006185336574197479,
|
| 302 |
+
"loss": 0.13018614053726196,
|
| 303 |
+
"step": 9472
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.44930950071590225,
|
| 307 |
+
"grad_norm": 1.4321770668029785,
|
| 308 |
+
"learning_rate": 0.0005999717763379407,
|
| 309 |
+
"loss": 0.13014769554138184,
|
| 310 |
+
"step": 9728
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.4611334349452681,
|
| 314 |
+
"grad_norm": 1.1316198110580444,
|
| 315 |
+
"learning_rate": 0.0005812652061542363,
|
| 316 |
+
"loss": 0.13009950518608093,
|
| 317 |
+
"step": 9984
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.472957369174634,
|
| 321 |
+
"grad_norm": 2.499011516571045,
|
| 322 |
+
"learning_rate": 0.0005624410208783071,
|
| 323 |
+
"loss": 0.13046522438526154,
|
| 324 |
+
"step": 10240
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.48478130340399983,
|
| 328 |
+
"grad_norm": 1.3082486391067505,
|
| 329 |
+
"learning_rate": 0.0005435264647440881,
|
| 330 |
+
"loss": 0.12951838970184326,
|
| 331 |
+
"step": 10496
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.49660523763336567,
|
| 335 |
+
"grad_norm": 0.8309692144393921,
|
| 336 |
+
"learning_rate": 0.000524548912779213,
|
| 337 |
+
"loss": 0.1311841756105423,
|
| 338 |
+
"step": 10752
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.5084291718627315,
|
| 342 |
+
"grad_norm": 2.236577272415161,
|
| 343 |
+
"learning_rate": 0.0005055358311851499,
|
| 344 |
+
"loss": 0.1328016221523285,
|
| 345 |
+
"step": 11008
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.5202531060920974,
|
| 349 |
+
"grad_norm": 1.8365247249603271,
|
| 350 |
+
"learning_rate": 0.0004865147375853812,
|
| 351 |
+
"loss": 0.13411875069141388,
|
| 352 |
+
"step": 11264
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.5320770403214632,
|
| 356 |
+
"grad_norm": 1.6082954406738281,
|
| 357 |
+
"learning_rate": 0.0004675131611991607,
|
| 358 |
+
"loss": 0.13458885252475739,
|
| 359 |
+
"step": 11520
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.5439009745508291,
|
| 363 |
+
"grad_norm": 1.031647801399231,
|
| 364 |
+
"learning_rate": 0.0004485586029984899,
|
| 365 |
+
"loss": 0.13649903237819672,
|
| 366 |
+
"step": 11776
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 0.5557249087801949,
|
| 370 |
+
"grad_norm": 1.5296710729599,
|
| 371 |
+
"learning_rate": 0.00042967849590597266,
|
| 372 |
+
"loss": 0.1377032846212387,
|
| 373 |
+
"step": 12032
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.5675488430095608,
|
| 377 |
+
"grad_norm": 1.8958959579467773,
|
| 378 |
+
"learning_rate": 0.0004109001650911621,
|
| 379 |
+
"loss": 0.13875064253807068,
|
| 380 |
+
"step": 12288
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.5675488430095608,
|
| 384 |
+
"eval_bleu": 0.0008588915725993688,
|
| 385 |
+
"eval_loss": 0.13989224594477648,
|
| 386 |
+
"eval_mse_loss": 0.13989224594477648,
|
| 387 |
+
"step": 12288
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.5675488430095608,
|
| 391 |
+
"eval_bleu": 0.0008588915725993688,
|
| 392 |
+
"eval_loss": 0.13989224594477648,
|
| 393 |
+
"eval_mse_loss": 0.13989224594477648,
|
| 394 |
+
"eval_runtime": 173.5527,
|
| 395 |
+
"eval_samples_per_second": 161.294,
|
| 396 |
+
"eval_steps_per_second": 2.524,
|
| 397 |
+
"step": 12288
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 0.5793727772389267,
|
| 401 |
+
"grad_norm": 2.3443777561187744,
|
| 402 |
+
"learning_rate": 0.0003922507884228551,
|
| 403 |
+
"loss": 0.1418546885251999,
|
| 404 |
+
"step": 12544
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.5911967114682924,
|
| 408 |
+
"grad_norm": 1.1060763597488403,
|
| 409 |
+
"learning_rate": 0.00037375735713457723,
|
| 410 |
+
"loss": 0.1433623880147934,
|
| 411 |
+
"step": 12800
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 0.6030206456976583,
|
| 415 |
+
"grad_norm": 1.806127905845642,
|
| 416 |
+
"learning_rate": 0.00035544663676018276,
|
| 417 |
+
"loss": 0.14775635302066803,
|
| 418 |
+
"step": 13056
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.6148445799270241,
|
| 422 |
+
"grad_norm": 1.7125252485275269,
|
| 423 |
+
"learning_rate": 0.00033734512839611255,
|
| 424 |
+
"loss": 0.14960132539272308,
|
| 425 |
+
"step": 13312
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.62666851415639,
|
| 429 |
+
"grad_norm": 2.017378091812134,
|
| 430 |
+
"learning_rate": 0.0003194790303463687,
|
| 431 |
+
"loss": 0.15272539854049683,
|
| 432 |
+
"step": 13568
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.6384924483857558,
|
| 436 |
+
"grad_norm": 2.1264116764068604,
|
| 437 |
+
"learning_rate": 0.00030187420020572406,
|
| 438 |
+
"loss": 0.15218184888362885,
|
| 439 |
+
"step": 13824
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.6503163826151217,
|
| 443 |
+
"grad_norm": 1.8446365594863892,
|
| 444 |
+
"learning_rate": 0.00028455611743603626,
|
| 445 |
+
"loss": 0.1529884785413742,
|
| 446 |
+
"step": 14080
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.6621403168444876,
|
| 450 |
+
"grad_norm": 2.2964935302734375,
|
| 451 |
+
"learning_rate": 0.0002675498464898373,
|
| 452 |
+
"loss": 0.15051132440567017,
|
| 453 |
+
"step": 14336
|
| 454 |
+
},
|
| 455 |
+
{
|
| 456 |
+
"epoch": 0.6739642510738534,
|
| 457 |
+
"grad_norm": 2.4945242404937744,
|
| 458 |
+
"learning_rate": 0.0002508800005345623,
|
| 459 |
+
"loss": 0.14853885769844055,
|
| 460 |
+
"step": 14592
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.6857881853032193,
|
| 464 |
+
"grad_norm": 1.3437820672988892,
|
| 465 |
+
"learning_rate": 0.00023457070582992562,
|
| 466 |
+
"loss": 0.14716462790966034,
|
| 467 |
+
"step": 14848
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.6976121195325851,
|
| 471 |
+
"grad_norm": 1.631665587425232,
|
| 472 |
+
"learning_rate": 0.00021864556680999692,
|
| 473 |
+
"loss": 0.14523349702358246,
|
| 474 |
+
"step": 15104
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.709436053761951,
|
| 478 |
+
"grad_norm": 1.9889062643051147,
|
| 479 |
+
"learning_rate": 0.0002031276319205152,
|
| 480 |
+
"loss": 0.14392541348934174,
|
| 481 |
+
"step": 15360
|
| 482 |
+
},
|
| 483 |
+
{
|
| 484 |
+
"epoch": 0.7212599879913169,
|
| 485 |
+
"grad_norm": 2.4409310817718506,
|
| 486 |
+
"learning_rate": 0.00018803936026088542,
|
| 487 |
+
"loss": 0.14354340732097626,
|
| 488 |
+
"step": 15616
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.7330839222206826,
|
| 492 |
+
"grad_norm": 2.1943178176879883,
|
| 493 |
+
"learning_rate": 0.00017340258907913464,
|
| 494 |
+
"loss": 0.14236165583133698,
|
| 495 |
+
"step": 15872
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.7449078564500485,
|
| 499 |
+
"grad_norm": 7.1248555183410645,
|
| 500 |
+
"learning_rate": 0.0001592385021668743,
|
| 501 |
+
"loss": 0.14313820004463196,
|
| 502 |
+
"step": 16128
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 0.7567317906794143,
|
| 506 |
+
"grad_norm": 1.3160756826400757,
|
| 507 |
+
"learning_rate": 0.0001455675992000087,
|
| 508 |
+
"loss": 0.1404125690460205,
|
| 509 |
+
"step": 16384
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.7567317906794143,
|
| 513 |
+
"eval_bleu": 0.000725950642943874,
|
| 514 |
+
"eval_loss": 0.1416369776101145,
|
| 515 |
+
"eval_mse_loss": 0.1416369776101145,
|
| 516 |
+
"step": 16384
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.7567317906794143,
|
| 520 |
+
"eval_bleu": 0.000725950642943874,
|
| 521 |
+
"eval_loss": 0.1416369776101145,
|
| 522 |
+
"eval_mse_loss": 0.1416369776101145,
|
| 523 |
+
"eval_runtime": 172.4801,
|
| 524 |
+
"eval_samples_per_second": 162.297,
|
| 525 |
+
"eval_steps_per_second": 2.539,
|
| 526 |
+
"step": 16384
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 0.7685557249087802,
|
| 530 |
+
"grad_norm": 2.9946746826171875,
|
| 531 |
+
"learning_rate": 0.000132409666069565,
|
| 532 |
+
"loss": 0.13992173969745636,
|
| 533 |
+
"step": 16640
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 0.780379659138146,
|
| 537 |
+
"grad_norm": 1.9437847137451172,
|
| 538 |
+
"learning_rate": 0.0001197837462455823,
|
| 539 |
+
"loss": 0.1393483430147171,
|
| 540 |
+
"step": 16896
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 0.7922035933675119,
|
| 544 |
+
"grad_norm": 1.5408047437667847,
|
| 545 |
+
"learning_rate": 0.00010770811321550749,
|
| 546 |
+
"loss": 0.139127716422081,
|
| 547 |
+
"step": 17152
|
| 548 |
+
},
|
| 549 |
+
{
|
| 550 |
+
"epoch": 0.8040275275968778,
|
| 551 |
+
"grad_norm": 4.208256721496582,
|
| 552 |
+
"learning_rate": 9.620024403698591e-05,
|
| 553 |
+
"loss": 0.1388680338859558,
|
| 554 |
+
"step": 17408
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.8158514618262436,
|
| 558 |
+
"grad_norm": 1.5926990509033203,
|
| 559 |
+
"learning_rate": 8.527679404332429e-05,
|
| 560 |
+
"loss": 0.13790729641914368,
|
| 561 |
+
"step": 17664
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.8276753960556095,
|
| 565 |
+
"grad_norm": 1.5863239765167236,
|
| 566 |
+
"learning_rate": 7.495357273823544e-05,
|
| 567 |
+
"loss": 0.13762786984443665,
|
| 568 |
+
"step": 17920
|
| 569 |
+
},
|
| 570 |
+
{
|
| 571 |
+
"epoch": 0.8394993302849753,
|
| 572 |
+
"grad_norm": 0.7356535792350769,
|
| 573 |
+
"learning_rate": 6.524552091475183e-05,
|
| 574 |
+
"loss": 0.13656531274318695,
|
| 575 |
+
"step": 18176
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.8513232645143411,
|
| 579 |
+
"grad_norm": 1.0298752784729004,
|
| 580 |
+
"learning_rate": 5.6166689031422024e-05,
|
| 581 |
+
"loss": 0.13603614270687103,
|
| 582 |
+
"step": 18432
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.8631471987437069,
|
| 586 |
+
"grad_norm": 0.8704520463943481,
|
| 587 |
+
"learning_rate": 4.773021687709067e-05,
|
| 588 |
+
"loss": 0.13718988001346588,
|
| 589 |
+
"step": 18688
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 0.8749711329730728,
|
| 593 |
+
"grad_norm": 1.7993870973587036,
|
| 594 |
+
"learning_rate": 3.994831455368719e-05,
|
| 595 |
+
"loss": 0.13554991781711578,
|
| 596 |
+
"step": 18944
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.8867950672024387,
|
| 600 |
+
"grad_norm": 1.9661895036697388,
|
| 601 |
+
"learning_rate": 3.283224480455282e-05,
|
| 602 |
+
"loss": 0.13669779896736145,
|
| 603 |
+
"step": 19200
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.8986190014318045,
|
| 607 |
+
"grad_norm": 2.396397113800049,
|
| 608 |
+
"learning_rate": 2.639230671387627e-05,
|
| 609 |
+
"loss": 0.13444113731384277,
|
| 610 |
+
"step": 19456
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.9104429356611704,
|
| 614 |
+
"grad_norm": 5.158544063568115,
|
| 615 |
+
"learning_rate": 2.063782080083576e-05,
|
| 616 |
+
"loss": 0.13486656546592712,
|
| 617 |
+
"step": 19712
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.9222668698905362,
|
| 621 |
+
"grad_norm": 2.635748863220215,
|
| 622 |
+
"learning_rate": 1.557711553001523e-05,
|
| 623 |
+
"loss": 0.13480573892593384,
|
| 624 |
+
"step": 19968
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.9340908041199021,
|
| 628 |
+
"grad_norm": 1.383921504020691,
|
| 629 |
+
"learning_rate": 1.1217515257622269e-05,
|
| 630 |
+
"loss": 0.13407942652702332,
|
| 631 |
+
"step": 20224
|
| 632 |
+
},
|
| 633 |
+
{
|
| 634 |
+
"epoch": 0.945914738349268,
|
| 635 |
+
"grad_norm": 1.3553746938705444,
|
| 636 |
+
"learning_rate": 7.565329630950746e-06,
|
| 637 |
+
"loss": 0.13326187431812286,
|
| 638 |
+
"step": 20480
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.945914738349268,
|
| 642 |
+
"eval_bleu": 0.0007028356585600878,
|
| 643 |
+
"eval_loss": 0.13298223137039028,
|
| 644 |
+
"eval_mse_loss": 0.13298223137039028,
|
| 645 |
+
"step": 20480
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.945914738349268,
|
| 649 |
+
"eval_bleu": 0.0007028356585600878,
|
| 650 |
+
"eval_loss": 0.13298223137039028,
|
| 651 |
+
"eval_mse_loss": 0.13298223137039028,
|
| 652 |
+
"eval_runtime": 172.3836,
|
| 653 |
+
"eval_samples_per_second": 162.388,
|
| 654 |
+
"eval_steps_per_second": 2.541,
|
| 655 |
+
"step": 20480
|
| 656 |
+
}
|
| 657 |
+
],
|
| 658 |
+
"logging_steps": 256,
|
| 659 |
+
"max_steps": 21651,
|
| 660 |
+
"num_input_tokens_seen": 0,
|
| 661 |
+
"num_train_epochs": 1,
|
| 662 |
+
"save_steps": 4096,
|
| 663 |
+
"stateful_callbacks": {
|
| 664 |
+
"TrainerControl": {
|
| 665 |
+
"args": {
|
| 666 |
+
"should_epoch_stop": false,
|
| 667 |
+
"should_evaluate": false,
|
| 668 |
+
"should_log": false,
|
| 669 |
+
"should_save": true,
|
| 670 |
+
"should_training_stop": false
|
| 671 |
+
},
|
| 672 |
+
"attributes": {}
|
| 673 |
+
}
|
| 674 |
+
},
|
| 675 |
+
"total_flos": 0.0,
|
| 676 |
+
"train_batch_size": 64,
|
| 677 |
+
"trial_name": null,
|
| 678 |
+
"trial_params": null
|
| 679 |
+
}
|
checkpoints-v3.0-discrete/checkpoint-20480/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9ead218b695fa4b63d3e4ce43bbad945367d95b659c14f235f9c2095bf1b3c31
|
| 3 |
+
size 5137
|