Commit
β’
3dd94b9
1
Parent(s):
10b336a
Training in progress, step 1900
Browse files- checkpoint-1500/latest +0 -1
- {checkpoint-1500 β checkpoint-1900}/config.json +0 -0
- {checkpoint-1500 β checkpoint-1900}/generation_config.json +0 -0
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_0_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_1_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_2_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_3_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_4_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_5_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_6_mp_rank_00_model_states.pt +1 -1
- {checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_7_mp_rank_00_model_states.pt +1 -1
- checkpoint-1900/latest +1 -0
- {checkpoint-1500 β checkpoint-1900}/model-00001-of-00002.safetensors +1 -1
- {checkpoint-1500 β checkpoint-1900}/model-00002-of-00002.safetensors +1 -1
- {checkpoint-1500 β checkpoint-1900}/model.safetensors.index.json +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_0.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_1.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_2.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_3.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_4.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_5.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_6.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/rng_state_7.pth +0 -0
- {checkpoint-1500 β checkpoint-1900}/special_tokens_map.json +0 -0
- {checkpoint-1500 β checkpoint-1900}/tokenizer.json +0 -0
- {checkpoint-1500 β checkpoint-1900}/tokenizer.model +0 -0
- {checkpoint-1500 β checkpoint-1900}/tokenizer_config.json +0 -0
- {checkpoint-1500 β checkpoint-1900}/trainer_state.json +243 -3
- {checkpoint-1500 β checkpoint-1900}/training_args.bin +0 -0
- {checkpoint-1500 β checkpoint-1900}/zero_to_fp32.py +0 -0
- runs/Aug22_18-42-03_ip-26-0-150-12/events.out.tfevents.1692729850.ip-26-0-150-12.2895584.0 +2 -2
checkpoint-1500/latest
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
global_step1500
|
|
|
|
{checkpoint-1500 β checkpoint-1900}/config.json
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/generation_config.json
RENAMED
File without changes
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:47aebfb81f8363ed05b74e6e078681b273c5c586465aaeef3d06ab2733851a41
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e2c4b9eee899438e56990c5b63498d9dcb5a31753aaf89888cc5b0f2ae771706
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4ea0f41d07ac677f422404956c83f402b623d1e9d38f80d96aa7df1573be8ff3
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ff2158c21bc75a93ba75193e8cd8dfad392a2398d8707967fcb537e61891b1a3
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c811cf5b1861c8cb662c054b12a7c96a038902adb71f04de6bf96a65b6246ec4
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a9369b4ef2bf8dddb4168a4446f9088e8f3fb3b051aff7a7e19a3049c2070e54
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5ef94ff510dd0d525c7d5571350b2b024f9cab3ccce19b0a137485160166c1b8
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 10107626487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fa3a389cbb94be741314103205c427d4a3571699efd16e0fd5f242bcb0c9f9f8
|
3 |
size 10107626487
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_0_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9ccec6639f6021228a72cacd939c896abbd5ea2c92885d42b8682b9ec5808e12
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_1_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:eb94590dee0991ff663b4bfea29d11e9fb5258d9f6b09f9dc2c4ed958a22cbbc
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_2_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fd1a137c362ef4d5e939a77f90b7f1238a607611aca192d48e5c14e5590f89c8
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_3_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:00c67c840ba12469a70530ece889168c13d1c236e3c47f561627f2fb998a5811
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_4_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c6cd53ec91a613c2b080addf94707266554f06bea53f22e5f2410638b7c2f359
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_5_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e8380e0444737b81d5c7bb2a3201de3b50244230313758be4830937fbfcdd62e
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_6_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:99848a79deee2a4543bbb55739697dc1b1e83d7124d2b9100b203e0aea8ede67
|
3 |
size 168086
|
{checkpoint-1500/global_step1500 β checkpoint-1900/global_step1900}/zero_pp_rank_7_mp_rank_00_model_states.pt
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 168086
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c98e12012ad590a146a8cf1c7bce43258a07f093310465ea9b884c26768dfbb4
|
3 |
size 168086
|
checkpoint-1900/latest
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
global_step1900
|
{checkpoint-1500 β checkpoint-1900}/model-00001-of-00002.safetensors
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 9976576392
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b0250158745c111e3a3bd8c0c63bb88e9f24ad6e3b56dea72fd77a43559e57cb
|
3 |
size 9976576392
|
{checkpoint-1500 β checkpoint-1900}/model-00002-of-00002.safetensors
RENAMED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3500296504
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bf8e87043471e756376e41c4bb6a0d60729025052f4bc413ae38b154f3eaa9d9
|
3 |
size 3500296504
|
{checkpoint-1500 β checkpoint-1900}/model.safetensors.index.json
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_0.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_1.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_2.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_3.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_4.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_5.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_6.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/rng_state_7.pth
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/special_tokens_map.json
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/tokenizer.json
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/tokenizer.model
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/tokenizer_config.json
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/trainer_state.json
RENAMED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -906,11 +906,251 @@
|
|
906 |
"learning_rate": 0.0003,
|
907 |
"loss": 0.1118,
|
908 |
"step": 1500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
909 |
}
|
910 |
],
|
911 |
"max_steps": 3000,
|
912 |
"num_train_epochs": 9223372036854775807,
|
913 |
-
"total_flos":
|
914 |
"trial_name": null,
|
915 |
"trial_params": null
|
916 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.6333333333333333,
|
5 |
+
"global_step": 1900,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
906 |
"learning_rate": 0.0003,
|
907 |
"loss": 0.1118,
|
908 |
"step": 1500
|
909 |
+
},
|
910 |
+
{
|
911 |
+
"epoch": 0.5,
|
912 |
+
"learning_rate": 0.0003,
|
913 |
+
"loss": 0.113,
|
914 |
+
"step": 1510
|
915 |
+
},
|
916 |
+
{
|
917 |
+
"epoch": 0.51,
|
918 |
+
"learning_rate": 0.0003,
|
919 |
+
"loss": 0.3687,
|
920 |
+
"step": 1520
|
921 |
+
},
|
922 |
+
{
|
923 |
+
"epoch": 0.51,
|
924 |
+
"learning_rate": 0.0003,
|
925 |
+
"loss": 0.1801,
|
926 |
+
"step": 1530
|
927 |
+
},
|
928 |
+
{
|
929 |
+
"epoch": 0.51,
|
930 |
+
"learning_rate": 0.0003,
|
931 |
+
"loss": 0.1534,
|
932 |
+
"step": 1540
|
933 |
+
},
|
934 |
+
{
|
935 |
+
"epoch": 0.52,
|
936 |
+
"learning_rate": 0.0003,
|
937 |
+
"loss": 0.1248,
|
938 |
+
"step": 1550
|
939 |
+
},
|
940 |
+
{
|
941 |
+
"epoch": 0.52,
|
942 |
+
"learning_rate": 0.0003,
|
943 |
+
"loss": 0.1091,
|
944 |
+
"step": 1560
|
945 |
+
},
|
946 |
+
{
|
947 |
+
"epoch": 0.52,
|
948 |
+
"learning_rate": 0.0003,
|
949 |
+
"loss": 0.0965,
|
950 |
+
"step": 1570
|
951 |
+
},
|
952 |
+
{
|
953 |
+
"epoch": 0.53,
|
954 |
+
"learning_rate": 0.0003,
|
955 |
+
"loss": 0.098,
|
956 |
+
"step": 1580
|
957 |
+
},
|
958 |
+
{
|
959 |
+
"epoch": 0.53,
|
960 |
+
"learning_rate": 0.0003,
|
961 |
+
"loss": 0.097,
|
962 |
+
"step": 1590
|
963 |
+
},
|
964 |
+
{
|
965 |
+
"epoch": 0.53,
|
966 |
+
"learning_rate": 0.0003,
|
967 |
+
"loss": 0.0885,
|
968 |
+
"step": 1600
|
969 |
+
},
|
970 |
+
{
|
971 |
+
"epoch": 0.54,
|
972 |
+
"learning_rate": 0.0003,
|
973 |
+
"loss": 0.0795,
|
974 |
+
"step": 1610
|
975 |
+
},
|
976 |
+
{
|
977 |
+
"epoch": 0.54,
|
978 |
+
"learning_rate": 0.0003,
|
979 |
+
"loss": 0.0843,
|
980 |
+
"step": 1620
|
981 |
+
},
|
982 |
+
{
|
983 |
+
"epoch": 0.54,
|
984 |
+
"learning_rate": 0.0003,
|
985 |
+
"loss": 0.0734,
|
986 |
+
"step": 1630
|
987 |
+
},
|
988 |
+
{
|
989 |
+
"epoch": 0.55,
|
990 |
+
"learning_rate": 0.0003,
|
991 |
+
"loss": 0.0744,
|
992 |
+
"step": 1640
|
993 |
+
},
|
994 |
+
{
|
995 |
+
"epoch": 0.55,
|
996 |
+
"learning_rate": 0.0003,
|
997 |
+
"loss": 0.0794,
|
998 |
+
"step": 1650
|
999 |
+
},
|
1000 |
+
{
|
1001 |
+
"epoch": 0.55,
|
1002 |
+
"learning_rate": 0.0003,
|
1003 |
+
"loss": 0.0708,
|
1004 |
+
"step": 1660
|
1005 |
+
},
|
1006 |
+
{
|
1007 |
+
"epoch": 0.56,
|
1008 |
+
"learning_rate": 0.0003,
|
1009 |
+
"loss": 0.1133,
|
1010 |
+
"step": 1670
|
1011 |
+
},
|
1012 |
+
{
|
1013 |
+
"epoch": 0.56,
|
1014 |
+
"learning_rate": 0.0003,
|
1015 |
+
"loss": 0.1582,
|
1016 |
+
"step": 1680
|
1017 |
+
},
|
1018 |
+
{
|
1019 |
+
"epoch": 0.56,
|
1020 |
+
"learning_rate": 0.0003,
|
1021 |
+
"loss": 0.1106,
|
1022 |
+
"step": 1690
|
1023 |
+
},
|
1024 |
+
{
|
1025 |
+
"epoch": 0.57,
|
1026 |
+
"learning_rate": 0.0003,
|
1027 |
+
"loss": 0.0904,
|
1028 |
+
"step": 1700
|
1029 |
+
},
|
1030 |
+
{
|
1031 |
+
"epoch": 0.57,
|
1032 |
+
"learning_rate": 0.0003,
|
1033 |
+
"loss": 0.0785,
|
1034 |
+
"step": 1710
|
1035 |
+
},
|
1036 |
+
{
|
1037 |
+
"epoch": 0.57,
|
1038 |
+
"learning_rate": 0.0003,
|
1039 |
+
"loss": 0.0747,
|
1040 |
+
"step": 1720
|
1041 |
+
},
|
1042 |
+
{
|
1043 |
+
"epoch": 0.58,
|
1044 |
+
"learning_rate": 0.0003,
|
1045 |
+
"loss": 0.0704,
|
1046 |
+
"step": 1730
|
1047 |
+
},
|
1048 |
+
{
|
1049 |
+
"epoch": 0.58,
|
1050 |
+
"learning_rate": 0.0003,
|
1051 |
+
"loss": 0.0634,
|
1052 |
+
"step": 1740
|
1053 |
+
},
|
1054 |
+
{
|
1055 |
+
"epoch": 0.58,
|
1056 |
+
"learning_rate": 0.0003,
|
1057 |
+
"loss": 0.0629,
|
1058 |
+
"step": 1750
|
1059 |
+
},
|
1060 |
+
{
|
1061 |
+
"epoch": 0.59,
|
1062 |
+
"learning_rate": 0.0003,
|
1063 |
+
"loss": 0.057,
|
1064 |
+
"step": 1760
|
1065 |
+
},
|
1066 |
+
{
|
1067 |
+
"epoch": 0.59,
|
1068 |
+
"learning_rate": 0.0003,
|
1069 |
+
"loss": 0.0563,
|
1070 |
+
"step": 1770
|
1071 |
+
},
|
1072 |
+
{
|
1073 |
+
"epoch": 0.59,
|
1074 |
+
"learning_rate": 0.0003,
|
1075 |
+
"loss": 0.054,
|
1076 |
+
"step": 1780
|
1077 |
+
},
|
1078 |
+
{
|
1079 |
+
"epoch": 0.6,
|
1080 |
+
"learning_rate": 0.0003,
|
1081 |
+
"loss": 0.0532,
|
1082 |
+
"step": 1790
|
1083 |
+
},
|
1084 |
+
{
|
1085 |
+
"epoch": 0.6,
|
1086 |
+
"learning_rate": 0.0003,
|
1087 |
+
"loss": 0.0525,
|
1088 |
+
"step": 1800
|
1089 |
+
},
|
1090 |
+
{
|
1091 |
+
"epoch": 0.6,
|
1092 |
+
"learning_rate": 0.0003,
|
1093 |
+
"loss": 0.0519,
|
1094 |
+
"step": 1810
|
1095 |
+
},
|
1096 |
+
{
|
1097 |
+
"epoch": 0.61,
|
1098 |
+
"learning_rate": 0.0003,
|
1099 |
+
"loss": 0.0478,
|
1100 |
+
"step": 1820
|
1101 |
+
},
|
1102 |
+
{
|
1103 |
+
"epoch": 0.61,
|
1104 |
+
"learning_rate": 0.0003,
|
1105 |
+
"loss": 0.0473,
|
1106 |
+
"step": 1830
|
1107 |
+
},
|
1108 |
+
{
|
1109 |
+
"epoch": 0.61,
|
1110 |
+
"learning_rate": 0.0003,
|
1111 |
+
"loss": 0.0473,
|
1112 |
+
"step": 1840
|
1113 |
+
},
|
1114 |
+
{
|
1115 |
+
"epoch": 0.62,
|
1116 |
+
"learning_rate": 0.0003,
|
1117 |
+
"loss": 0.0492,
|
1118 |
+
"step": 1850
|
1119 |
+
},
|
1120 |
+
{
|
1121 |
+
"epoch": 0.62,
|
1122 |
+
"learning_rate": 0.0003,
|
1123 |
+
"loss": 0.0451,
|
1124 |
+
"step": 1860
|
1125 |
+
},
|
1126 |
+
{
|
1127 |
+
"epoch": 0.62,
|
1128 |
+
"learning_rate": 0.0003,
|
1129 |
+
"loss": 0.0453,
|
1130 |
+
"step": 1870
|
1131 |
+
},
|
1132 |
+
{
|
1133 |
+
"epoch": 0.63,
|
1134 |
+
"learning_rate": 0.0003,
|
1135 |
+
"loss": 0.0439,
|
1136 |
+
"step": 1880
|
1137 |
+
},
|
1138 |
+
{
|
1139 |
+
"epoch": 0.63,
|
1140 |
+
"learning_rate": 0.0003,
|
1141 |
+
"loss": 0.0435,
|
1142 |
+
"step": 1890
|
1143 |
+
},
|
1144 |
+
{
|
1145 |
+
"epoch": 0.63,
|
1146 |
+
"learning_rate": 0.0003,
|
1147 |
+
"loss": 0.0435,
|
1148 |
+
"step": 1900
|
1149 |
}
|
1150 |
],
|
1151 |
"max_steps": 3000,
|
1152 |
"num_train_epochs": 9223372036854775807,
|
1153 |
+
"total_flos": 795642691584000.0,
|
1154 |
"trial_name": null,
|
1155 |
"trial_params": null
|
1156 |
}
|
{checkpoint-1500 β checkpoint-1900}/training_args.bin
RENAMED
File without changes
|
{checkpoint-1500 β checkpoint-1900}/zero_to_fp32.py
RENAMED
File without changes
|
runs/Aug22_18-42-03_ip-26-0-150-12/events.out.tfevents.1692729850.ip-26-0-150-12.2895584.0
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1542e5f72a0deb50ea56f278bcbfb04cc7ef3686cad25fabc136934d1b2ae0e6
|
3 |
+
size 33981
|