Upload folder using huggingface_hub
Browse files- global_step5000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt +3 -0
- global_step5000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt +3 -0
- global_step5000/zero_pp_rank_0_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_1_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_2_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_3_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_4_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_5_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_6_mp_rank_00_model_states.pt +3 -0
- global_step5000/zero_pp_rank_7_mp_rank_00_model_states.pt +3 -0
- latest +1 -1
- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- scheduler.pt +1 -1
- trainer_state.json +3503 -3
global_step5000/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:06466086dee701cbe722ae107fb8ec1a80970d141bea0d18508fbfe1ad670a05
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ea84d69e24e3c321c331e60029e3541824eec5931ea5012d3bac14515d817391
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2a076cbb8b3b6c0c8f4817cc1c07fc3c9baa867470480278f4bfeb0bc369541d
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b6f2ce026059fb91e597626bb7c57101df78b9f1293442290c1d19e28461b6f
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_4_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:72544dffe76ae23965289b4cc067f4cde4285a83f467893b40633b93d1d0d278
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_5_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:447808c131d563e37d25a824342c1b66d012fcd484cbb838a5f64d8a2ff41910
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_6_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62b232a5831c1233293017065e950b48a46baf6e4c8857ace6868ef49c17f032
|
| 3 |
+
size 5939820828
|
global_step5000/bf16_zero_pp_rank_7_mp_rank_00_optim_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5b72b5665ae4d2185c4c6416ce28f5a0ef7e553c2f0045d2258fd95eb0158d6e
|
| 3 |
+
size 5939820828
|
global_step5000/zero_pp_rank_0_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:452e4b05e3c56dc57b8e75c9c4f97810420fa5a599cfd0f71f474e088723677b
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_1_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3f97d5fe0b381c3da18941c4a226c9199f6da1a78190a5166271a97dcb5b5516
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_2_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8946799e3c78f06898c0cee757f52d4a3d67e3098ccd04ebe46592c36259b684
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_3_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:74e01dd094883b1b5ac19a670aeb18171b9c211b46315c415b0b9c14fa827908
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_4_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:614331d549444f501f441e7673f7f00ec63b946fe6c5727beeb43d8168519ff6
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_5_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c45f9b8ba995f620647bb93b8cdf9e54cf762aaf03b8f4bbe120ea7d0b768475
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_6_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:320e3f915838a1188b740ff78746b8e669856dd11ef0056722e42dc7b526b057
|
| 3 |
+
size 107706008
|
global_step5000/zero_pp_rank_7_mp_rank_00_model_states.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5a42e9f9aacd9ce8f5674cee935e6443fe71253eb88c057c3274e524570b5772
|
| 3 |
+
size 107706008
|
latest
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
|
|
|
|
| 1 |
+
global_step5000
|
model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4989973456
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2407ec1ec8fa85107e8ebc1b659b5435c3a5bc8ca6bb1741dd92c7139fec22e8
|
| 3 |
size 4989973456
|
model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3786358064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ff1c1883d58d4062b6ff60c91c98c018617933e570cbcbcdc665b56143a0ece
|
| 3 |
size 3786358064
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:38371fe746a8513231631b5bd32374702b655c44b4e2865b7a45576402f7015f
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 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,
|
|
@@ -31507,6 +31507,3506 @@
|
|
| 31507 |
"learning_rate": 9.19645679769382e-07,
|
| 31508 |
"loss": 0.8809,
|
| 31509 |
"step": 4500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 31510 |
}
|
| 31511 |
],
|
| 31512 |
"logging_steps": 1.0,
|
|
@@ -31514,7 +35014,7 @@
|
|
| 31514 |
"num_input_tokens_seen": 0,
|
| 31515 |
"num_train_epochs": 1,
|
| 31516 |
"save_steps": 500,
|
| 31517 |
-
"total_flos":
|
| 31518 |
"train_batch_size": 4,
|
| 31519 |
"trial_name": null,
|
| 31520 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9628345850182939,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 31507 |
"learning_rate": 9.19645679769382e-07,
|
| 31508 |
"loss": 0.8809,
|
| 31509 |
"step": 4500
|
| 31510 |
+
},
|
| 31511 |
+
{
|
| 31512 |
+
"epoch": 0.87,
|
| 31513 |
+
"grad_norm": 1.594906527805702,
|
| 31514 |
+
"learning_rate": 9.170347950523506e-07,
|
| 31515 |
+
"loss": 0.8415,
|
| 31516 |
+
"step": 4501
|
| 31517 |
+
},
|
| 31518 |
+
{
|
| 31519 |
+
"epoch": 0.87,
|
| 31520 |
+
"grad_norm": 1.6839026327500048,
|
| 31521 |
+
"learning_rate": 9.144274436594558e-07,
|
| 31522 |
+
"loss": 0.8076,
|
| 31523 |
+
"step": 4502
|
| 31524 |
+
},
|
| 31525 |
+
{
|
| 31526 |
+
"epoch": 0.87,
|
| 31527 |
+
"grad_norm": 1.5852937015977455,
|
| 31528 |
+
"learning_rate": 9.118236266049707e-07,
|
| 31529 |
+
"loss": 0.8393,
|
| 31530 |
+
"step": 4503
|
| 31531 |
+
},
|
| 31532 |
+
{
|
| 31533 |
+
"epoch": 0.87,
|
| 31534 |
+
"grad_norm": 1.482828251223841,
|
| 31535 |
+
"learning_rate": 9.092233449017962e-07,
|
| 31536 |
+
"loss": 0.8594,
|
| 31537 |
+
"step": 4504
|
| 31538 |
+
},
|
| 31539 |
+
{
|
| 31540 |
+
"epoch": 0.87,
|
| 31541 |
+
"grad_norm": 1.454387680780983,
|
| 31542 |
+
"learning_rate": 9.066265995614554e-07,
|
| 31543 |
+
"loss": 0.7132,
|
| 31544 |
+
"step": 4505
|
| 31545 |
+
},
|
| 31546 |
+
{
|
| 31547 |
+
"epoch": 0.87,
|
| 31548 |
+
"grad_norm": 1.6223084567765689,
|
| 31549 |
+
"learning_rate": 9.040333915940991e-07,
|
| 31550 |
+
"loss": 0.7961,
|
| 31551 |
+
"step": 4506
|
| 31552 |
+
},
|
| 31553 |
+
{
|
| 31554 |
+
"epoch": 0.87,
|
| 31555 |
+
"grad_norm": 1.6762400840034626,
|
| 31556 |
+
"learning_rate": 9.014437220084948e-07,
|
| 31557 |
+
"loss": 0.8379,
|
| 31558 |
+
"step": 4507
|
| 31559 |
+
},
|
| 31560 |
+
{
|
| 31561 |
+
"epoch": 0.87,
|
| 31562 |
+
"grad_norm": 1.5407728498704398,
|
| 31563 |
+
"learning_rate": 8.988575918120413e-07,
|
| 31564 |
+
"loss": 0.7942,
|
| 31565 |
+
"step": 4508
|
| 31566 |
+
},
|
| 31567 |
+
{
|
| 31568 |
+
"epoch": 0.87,
|
| 31569 |
+
"grad_norm": 1.8192484009488412,
|
| 31570 |
+
"learning_rate": 8.96275002010758e-07,
|
| 31571 |
+
"loss": 0.8772,
|
| 31572 |
+
"step": 4509
|
| 31573 |
+
},
|
| 31574 |
+
{
|
| 31575 |
+
"epoch": 0.87,
|
| 31576 |
+
"grad_norm": 1.7400925959795057,
|
| 31577 |
+
"learning_rate": 8.93695953609286e-07,
|
| 31578 |
+
"loss": 0.8795,
|
| 31579 |
+
"step": 4510
|
| 31580 |
+
},
|
| 31581 |
+
{
|
| 31582 |
+
"epoch": 0.87,
|
| 31583 |
+
"grad_norm": 1.6922514370220625,
|
| 31584 |
+
"learning_rate": 8.911204476108892e-07,
|
| 31585 |
+
"loss": 0.8421,
|
| 31586 |
+
"step": 4511
|
| 31587 |
+
},
|
| 31588 |
+
{
|
| 31589 |
+
"epoch": 0.87,
|
| 31590 |
+
"grad_norm": 1.5665789156211634,
|
| 31591 |
+
"learning_rate": 8.885484850174541e-07,
|
| 31592 |
+
"loss": 0.7943,
|
| 31593 |
+
"step": 4512
|
| 31594 |
+
},
|
| 31595 |
+
{
|
| 31596 |
+
"epoch": 0.87,
|
| 31597 |
+
"grad_norm": 1.4469490149019075,
|
| 31598 |
+
"learning_rate": 8.859800668294916e-07,
|
| 31599 |
+
"loss": 0.8074,
|
| 31600 |
+
"step": 4513
|
| 31601 |
+
},
|
| 31602 |
+
{
|
| 31603 |
+
"epoch": 0.87,
|
| 31604 |
+
"grad_norm": 1.753138597801452,
|
| 31605 |
+
"learning_rate": 8.834151940461255e-07,
|
| 31606 |
+
"loss": 0.8427,
|
| 31607 |
+
"step": 4514
|
| 31608 |
+
},
|
| 31609 |
+
{
|
| 31610 |
+
"epoch": 0.87,
|
| 31611 |
+
"grad_norm": 1.76427896737675,
|
| 31612 |
+
"learning_rate": 8.808538676651079e-07,
|
| 31613 |
+
"loss": 0.8607,
|
| 31614 |
+
"step": 4515
|
| 31615 |
+
},
|
| 31616 |
+
{
|
| 31617 |
+
"epoch": 0.87,
|
| 31618 |
+
"grad_norm": 1.6686741522110797,
|
| 31619 |
+
"learning_rate": 8.782960886828084e-07,
|
| 31620 |
+
"loss": 0.8144,
|
| 31621 |
+
"step": 4516
|
| 31622 |
+
},
|
| 31623 |
+
{
|
| 31624 |
+
"epoch": 0.87,
|
| 31625 |
+
"grad_norm": 1.5827402549651326,
|
| 31626 |
+
"learning_rate": 8.75741858094219e-07,
|
| 31627 |
+
"loss": 0.8336,
|
| 31628 |
+
"step": 4517
|
| 31629 |
+
},
|
| 31630 |
+
{
|
| 31631 |
+
"epoch": 0.87,
|
| 31632 |
+
"grad_norm": 1.6691552735982413,
|
| 31633 |
+
"learning_rate": 8.73191176892948e-07,
|
| 31634 |
+
"loss": 0.7777,
|
| 31635 |
+
"step": 4518
|
| 31636 |
+
},
|
| 31637 |
+
{
|
| 31638 |
+
"epoch": 0.87,
|
| 31639 |
+
"grad_norm": 1.569135781232643,
|
| 31640 |
+
"learning_rate": 8.706440460712251e-07,
|
| 31641 |
+
"loss": 0.8517,
|
| 31642 |
+
"step": 4519
|
| 31643 |
+
},
|
| 31644 |
+
{
|
| 31645 |
+
"epoch": 0.87,
|
| 31646 |
+
"grad_norm": 1.7146933616106483,
|
| 31647 |
+
"learning_rate": 8.681004666199011e-07,
|
| 31648 |
+
"loss": 0.8775,
|
| 31649 |
+
"step": 4520
|
| 31650 |
+
},
|
| 31651 |
+
{
|
| 31652 |
+
"epoch": 0.87,
|
| 31653 |
+
"grad_norm": 1.6239984857466807,
|
| 31654 |
+
"learning_rate": 8.655604395284378e-07,
|
| 31655 |
+
"loss": 0.8086,
|
| 31656 |
+
"step": 4521
|
| 31657 |
+
},
|
| 31658 |
+
{
|
| 31659 |
+
"epoch": 0.87,
|
| 31660 |
+
"grad_norm": 1.656663129562715,
|
| 31661 |
+
"learning_rate": 8.630239657849215e-07,
|
| 31662 |
+
"loss": 0.8586,
|
| 31663 |
+
"step": 4522
|
| 31664 |
+
},
|
| 31665 |
+
{
|
| 31666 |
+
"epoch": 0.87,
|
| 31667 |
+
"grad_norm": 1.6751159708848744,
|
| 31668 |
+
"learning_rate": 8.604910463760585e-07,
|
| 31669 |
+
"loss": 0.8371,
|
| 31670 |
+
"step": 4523
|
| 31671 |
+
},
|
| 31672 |
+
{
|
| 31673 |
+
"epoch": 0.87,
|
| 31674 |
+
"grad_norm": 1.5923042899442412,
|
| 31675 |
+
"learning_rate": 8.579616822871628e-07,
|
| 31676 |
+
"loss": 0.8247,
|
| 31677 |
+
"step": 4524
|
| 31678 |
+
},
|
| 31679 |
+
{
|
| 31680 |
+
"epoch": 0.87,
|
| 31681 |
+
"grad_norm": 1.6631348630525378,
|
| 31682 |
+
"learning_rate": 8.554358745021741e-07,
|
| 31683 |
+
"loss": 0.8895,
|
| 31684 |
+
"step": 4525
|
| 31685 |
+
},
|
| 31686 |
+
{
|
| 31687 |
+
"epoch": 0.87,
|
| 31688 |
+
"grad_norm": 1.7292854021155024,
|
| 31689 |
+
"learning_rate": 8.529136240036439e-07,
|
| 31690 |
+
"loss": 0.8041,
|
| 31691 |
+
"step": 4526
|
| 31692 |
+
},
|
| 31693 |
+
{
|
| 31694 |
+
"epoch": 0.87,
|
| 31695 |
+
"grad_norm": 1.6301954531826386,
|
| 31696 |
+
"learning_rate": 8.503949317727444e-07,
|
| 31697 |
+
"loss": 0.8314,
|
| 31698 |
+
"step": 4527
|
| 31699 |
+
},
|
| 31700 |
+
{
|
| 31701 |
+
"epoch": 0.87,
|
| 31702 |
+
"grad_norm": 1.8784379544296637,
|
| 31703 |
+
"learning_rate": 8.478797987892595e-07,
|
| 31704 |
+
"loss": 0.7901,
|
| 31705 |
+
"step": 4528
|
| 31706 |
+
},
|
| 31707 |
+
{
|
| 31708 |
+
"epoch": 0.87,
|
| 31709 |
+
"grad_norm": 1.5767654640020843,
|
| 31710 |
+
"learning_rate": 8.45368226031592e-07,
|
| 31711 |
+
"loss": 0.8215,
|
| 31712 |
+
"step": 4529
|
| 31713 |
+
},
|
| 31714 |
+
{
|
| 31715 |
+
"epoch": 0.87,
|
| 31716 |
+
"grad_norm": 1.5956433850813396,
|
| 31717 |
+
"learning_rate": 8.42860214476754e-07,
|
| 31718 |
+
"loss": 0.7906,
|
| 31719 |
+
"step": 4530
|
| 31720 |
+
},
|
| 31721 |
+
{
|
| 31722 |
+
"epoch": 0.87,
|
| 31723 |
+
"grad_norm": 1.9037041961425434,
|
| 31724 |
+
"learning_rate": 8.403557651003779e-07,
|
| 31725 |
+
"loss": 0.8136,
|
| 31726 |
+
"step": 4531
|
| 31727 |
+
},
|
| 31728 |
+
{
|
| 31729 |
+
"epoch": 0.87,
|
| 31730 |
+
"grad_norm": 1.7133031905826972,
|
| 31731 |
+
"learning_rate": 8.378548788767083e-07,
|
| 31732 |
+
"loss": 0.7756,
|
| 31733 |
+
"step": 4532
|
| 31734 |
+
},
|
| 31735 |
+
{
|
| 31736 |
+
"epoch": 0.87,
|
| 31737 |
+
"grad_norm": 1.6653051307639297,
|
| 31738 |
+
"learning_rate": 8.353575567786032e-07,
|
| 31739 |
+
"loss": 0.8682,
|
| 31740 |
+
"step": 4533
|
| 31741 |
+
},
|
| 31742 |
+
{
|
| 31743 |
+
"epoch": 0.87,
|
| 31744 |
+
"grad_norm": 1.6041819040966523,
|
| 31745 |
+
"learning_rate": 8.328637997775368e-07,
|
| 31746 |
+
"loss": 0.8277,
|
| 31747 |
+
"step": 4534
|
| 31748 |
+
},
|
| 31749 |
+
{
|
| 31750 |
+
"epoch": 0.87,
|
| 31751 |
+
"grad_norm": 1.788934035577091,
|
| 31752 |
+
"learning_rate": 8.303736088435921e-07,
|
| 31753 |
+
"loss": 0.9114,
|
| 31754 |
+
"step": 4535
|
| 31755 |
+
},
|
| 31756 |
+
{
|
| 31757 |
+
"epoch": 0.87,
|
| 31758 |
+
"grad_norm": 1.5686609721094282,
|
| 31759 |
+
"learning_rate": 8.278869849454718e-07,
|
| 31760 |
+
"loss": 0.809,
|
| 31761 |
+
"step": 4536
|
| 31762 |
+
},
|
| 31763 |
+
{
|
| 31764 |
+
"epoch": 0.87,
|
| 31765 |
+
"grad_norm": 1.4676365965258953,
|
| 31766 |
+
"learning_rate": 8.2540392905048e-07,
|
| 31767 |
+
"loss": 0.8122,
|
| 31768 |
+
"step": 4537
|
| 31769 |
+
},
|
| 31770 |
+
{
|
| 31771 |
+
"epoch": 0.87,
|
| 31772 |
+
"grad_norm": 1.6464117299937717,
|
| 31773 |
+
"learning_rate": 8.22924442124543e-07,
|
| 31774 |
+
"loss": 0.8284,
|
| 31775 |
+
"step": 4538
|
| 31776 |
+
},
|
| 31777 |
+
{
|
| 31778 |
+
"epoch": 0.87,
|
| 31779 |
+
"grad_norm": 1.5900447488545841,
|
| 31780 |
+
"learning_rate": 8.204485251321947e-07,
|
| 31781 |
+
"loss": 0.8182,
|
| 31782 |
+
"step": 4539
|
| 31783 |
+
},
|
| 31784 |
+
{
|
| 31785 |
+
"epoch": 0.87,
|
| 31786 |
+
"grad_norm": 1.5356577844841082,
|
| 31787 |
+
"learning_rate": 8.179761790365803e-07,
|
| 31788 |
+
"loss": 0.7832,
|
| 31789 |
+
"step": 4540
|
| 31790 |
+
},
|
| 31791 |
+
{
|
| 31792 |
+
"epoch": 0.87,
|
| 31793 |
+
"grad_norm": 1.692243945185495,
|
| 31794 |
+
"learning_rate": 8.15507404799456e-07,
|
| 31795 |
+
"loss": 0.9026,
|
| 31796 |
+
"step": 4541
|
| 31797 |
+
},
|
| 31798 |
+
{
|
| 31799 |
+
"epoch": 0.87,
|
| 31800 |
+
"grad_norm": 1.6824591329207468,
|
| 31801 |
+
"learning_rate": 8.130422033811892e-07,
|
| 31802 |
+
"loss": 0.8577,
|
| 31803 |
+
"step": 4542
|
| 31804 |
+
},
|
| 31805 |
+
{
|
| 31806 |
+
"epoch": 0.87,
|
| 31807 |
+
"grad_norm": 1.6332709059097492,
|
| 31808 |
+
"learning_rate": 8.105805757407592e-07,
|
| 31809 |
+
"loss": 0.8176,
|
| 31810 |
+
"step": 4543
|
| 31811 |
+
},
|
| 31812 |
+
{
|
| 31813 |
+
"epoch": 0.88,
|
| 31814 |
+
"grad_norm": 1.0034749651827926,
|
| 31815 |
+
"learning_rate": 8.081225228357481e-07,
|
| 31816 |
+
"loss": 0.7566,
|
| 31817 |
+
"step": 4544
|
| 31818 |
+
},
|
| 31819 |
+
{
|
| 31820 |
+
"epoch": 0.88,
|
| 31821 |
+
"grad_norm": 1.5996492708651493,
|
| 31822 |
+
"learning_rate": 8.056680456223553e-07,
|
| 31823 |
+
"loss": 0.8032,
|
| 31824 |
+
"step": 4545
|
| 31825 |
+
},
|
| 31826 |
+
{
|
| 31827 |
+
"epoch": 0.88,
|
| 31828 |
+
"grad_norm": 0.9290414679374163,
|
| 31829 |
+
"learning_rate": 8.03217145055385e-07,
|
| 31830 |
+
"loss": 0.7635,
|
| 31831 |
+
"step": 4546
|
| 31832 |
+
},
|
| 31833 |
+
{
|
| 31834 |
+
"epoch": 0.88,
|
| 31835 |
+
"grad_norm": 1.637370705811582,
|
| 31836 |
+
"learning_rate": 8.007698220882521e-07,
|
| 31837 |
+
"loss": 0.8748,
|
| 31838 |
+
"step": 4547
|
| 31839 |
+
},
|
| 31840 |
+
{
|
| 31841 |
+
"epoch": 0.88,
|
| 31842 |
+
"grad_norm": 1.6635441096211938,
|
| 31843 |
+
"learning_rate": 7.983260776729773e-07,
|
| 31844 |
+
"loss": 0.8472,
|
| 31845 |
+
"step": 4548
|
| 31846 |
+
},
|
| 31847 |
+
{
|
| 31848 |
+
"epoch": 0.88,
|
| 31849 |
+
"grad_norm": 1.5998342354299384,
|
| 31850 |
+
"learning_rate": 7.958859127601937e-07,
|
| 31851 |
+
"loss": 0.8708,
|
| 31852 |
+
"step": 4549
|
| 31853 |
+
},
|
| 31854 |
+
{
|
| 31855 |
+
"epoch": 0.88,
|
| 31856 |
+
"grad_norm": 0.9729592102274293,
|
| 31857 |
+
"learning_rate": 7.934493282991373e-07,
|
| 31858 |
+
"loss": 0.7931,
|
| 31859 |
+
"step": 4550
|
| 31860 |
+
},
|
| 31861 |
+
{
|
| 31862 |
+
"epoch": 0.88,
|
| 31863 |
+
"grad_norm": 1.6456794482605581,
|
| 31864 |
+
"learning_rate": 7.910163252376524e-07,
|
| 31865 |
+
"loss": 0.7578,
|
| 31866 |
+
"step": 4551
|
| 31867 |
+
},
|
| 31868 |
+
{
|
| 31869 |
+
"epoch": 0.88,
|
| 31870 |
+
"grad_norm": 1.5501096019336218,
|
| 31871 |
+
"learning_rate": 7.885869045221917e-07,
|
| 31872 |
+
"loss": 0.8705,
|
| 31873 |
+
"step": 4552
|
| 31874 |
+
},
|
| 31875 |
+
{
|
| 31876 |
+
"epoch": 0.88,
|
| 31877 |
+
"grad_norm": 1.5009277862429256,
|
| 31878 |
+
"learning_rate": 7.861610670978126e-07,
|
| 31879 |
+
"loss": 0.8719,
|
| 31880 |
+
"step": 4553
|
| 31881 |
+
},
|
| 31882 |
+
{
|
| 31883 |
+
"epoch": 0.88,
|
| 31884 |
+
"grad_norm": 1.6530036454212895,
|
| 31885 |
+
"learning_rate": 7.837388139081803e-07,
|
| 31886 |
+
"loss": 0.839,
|
| 31887 |
+
"step": 4554
|
| 31888 |
+
},
|
| 31889 |
+
{
|
| 31890 |
+
"epoch": 0.88,
|
| 31891 |
+
"grad_norm": 1.728410895192731,
|
| 31892 |
+
"learning_rate": 7.813201458955644e-07,
|
| 31893 |
+
"loss": 0.7931,
|
| 31894 |
+
"step": 4555
|
| 31895 |
+
},
|
| 31896 |
+
{
|
| 31897 |
+
"epoch": 0.88,
|
| 31898 |
+
"grad_norm": 1.645971164332409,
|
| 31899 |
+
"learning_rate": 7.789050640008411e-07,
|
| 31900 |
+
"loss": 0.8329,
|
| 31901 |
+
"step": 4556
|
| 31902 |
+
},
|
| 31903 |
+
{
|
| 31904 |
+
"epoch": 0.88,
|
| 31905 |
+
"grad_norm": 1.8456320281211538,
|
| 31906 |
+
"learning_rate": 7.7649356916349e-07,
|
| 31907 |
+
"loss": 0.8624,
|
| 31908 |
+
"step": 4557
|
| 31909 |
+
},
|
| 31910 |
+
{
|
| 31911 |
+
"epoch": 0.88,
|
| 31912 |
+
"grad_norm": 1.5518984995792304,
|
| 31913 |
+
"learning_rate": 7.740856623215953e-07,
|
| 31914 |
+
"loss": 0.7869,
|
| 31915 |
+
"step": 4558
|
| 31916 |
+
},
|
| 31917 |
+
{
|
| 31918 |
+
"epoch": 0.88,
|
| 31919 |
+
"grad_norm": 1.701746462823959,
|
| 31920 |
+
"learning_rate": 7.716813444118476e-07,
|
| 31921 |
+
"loss": 0.8131,
|
| 31922 |
+
"step": 4559
|
| 31923 |
+
},
|
| 31924 |
+
{
|
| 31925 |
+
"epoch": 0.88,
|
| 31926 |
+
"grad_norm": 1.5597681704026225,
|
| 31927 |
+
"learning_rate": 7.692806163695377e-07,
|
| 31928 |
+
"loss": 0.8239,
|
| 31929 |
+
"step": 4560
|
| 31930 |
+
},
|
| 31931 |
+
{
|
| 31932 |
+
"epoch": 0.88,
|
| 31933 |
+
"grad_norm": 1.6461590533877206,
|
| 31934 |
+
"learning_rate": 7.668834791285651e-07,
|
| 31935 |
+
"loss": 0.9086,
|
| 31936 |
+
"step": 4561
|
| 31937 |
+
},
|
| 31938 |
+
{
|
| 31939 |
+
"epoch": 0.88,
|
| 31940 |
+
"grad_norm": 1.7135544196441648,
|
| 31941 |
+
"learning_rate": 7.644899336214273e-07,
|
| 31942 |
+
"loss": 0.899,
|
| 31943 |
+
"step": 4562
|
| 31944 |
+
},
|
| 31945 |
+
{
|
| 31946 |
+
"epoch": 0.88,
|
| 31947 |
+
"grad_norm": 1.6928781909018462,
|
| 31948 |
+
"learning_rate": 7.620999807792284e-07,
|
| 31949 |
+
"loss": 0.892,
|
| 31950 |
+
"step": 4563
|
| 31951 |
+
},
|
| 31952 |
+
{
|
| 31953 |
+
"epoch": 0.88,
|
| 31954 |
+
"grad_norm": 1.7778128461398606,
|
| 31955 |
+
"learning_rate": 7.597136215316737e-07,
|
| 31956 |
+
"loss": 0.8981,
|
| 31957 |
+
"step": 4564
|
| 31958 |
+
},
|
| 31959 |
+
{
|
| 31960 |
+
"epoch": 0.88,
|
| 31961 |
+
"grad_norm": 1.5417553658687158,
|
| 31962 |
+
"learning_rate": 7.573308568070681e-07,
|
| 31963 |
+
"loss": 0.8443,
|
| 31964 |
+
"step": 4565
|
| 31965 |
+
},
|
| 31966 |
+
{
|
| 31967 |
+
"epoch": 0.88,
|
| 31968 |
+
"grad_norm": 1.645822549835426,
|
| 31969 |
+
"learning_rate": 7.549516875323215e-07,
|
| 31970 |
+
"loss": 0.7982,
|
| 31971 |
+
"step": 4566
|
| 31972 |
+
},
|
| 31973 |
+
{
|
| 31974 |
+
"epoch": 0.88,
|
| 31975 |
+
"grad_norm": 1.5611574387528526,
|
| 31976 |
+
"learning_rate": 7.525761146329447e-07,
|
| 31977 |
+
"loss": 0.9295,
|
| 31978 |
+
"step": 4567
|
| 31979 |
+
},
|
| 31980 |
+
{
|
| 31981 |
+
"epoch": 0.88,
|
| 31982 |
+
"grad_norm": 1.611365281760663,
|
| 31983 |
+
"learning_rate": 7.502041390330472e-07,
|
| 31984 |
+
"loss": 0.7989,
|
| 31985 |
+
"step": 4568
|
| 31986 |
+
},
|
| 31987 |
+
{
|
| 31988 |
+
"epoch": 0.88,
|
| 31989 |
+
"grad_norm": 1.4307881929378798,
|
| 31990 |
+
"learning_rate": 7.478357616553433e-07,
|
| 31991 |
+
"loss": 0.8531,
|
| 31992 |
+
"step": 4569
|
| 31993 |
+
},
|
| 31994 |
+
{
|
| 31995 |
+
"epoch": 0.88,
|
| 31996 |
+
"grad_norm": 1.6613458669647516,
|
| 31997 |
+
"learning_rate": 7.454709834211438e-07,
|
| 31998 |
+
"loss": 0.8417,
|
| 31999 |
+
"step": 4570
|
| 32000 |
+
},
|
| 32001 |
+
{
|
| 32002 |
+
"epoch": 0.88,
|
| 32003 |
+
"grad_norm": 1.5988643310648785,
|
| 32004 |
+
"learning_rate": 7.431098052503594e-07,
|
| 32005 |
+
"loss": 0.8731,
|
| 32006 |
+
"step": 4571
|
| 32007 |
+
},
|
| 32008 |
+
{
|
| 32009 |
+
"epoch": 0.88,
|
| 32010 |
+
"grad_norm": 1.520700797069022,
|
| 32011 |
+
"learning_rate": 7.40752228061502e-07,
|
| 32012 |
+
"loss": 0.8015,
|
| 32013 |
+
"step": 4572
|
| 32014 |
+
},
|
| 32015 |
+
{
|
| 32016 |
+
"epoch": 0.88,
|
| 32017 |
+
"grad_norm": 1.6569071538578897,
|
| 32018 |
+
"learning_rate": 7.383982527716848e-07,
|
| 32019 |
+
"loss": 0.9232,
|
| 32020 |
+
"step": 4573
|
| 32021 |
+
},
|
| 32022 |
+
{
|
| 32023 |
+
"epoch": 0.88,
|
| 32024 |
+
"grad_norm": 1.6385053303931487,
|
| 32025 |
+
"learning_rate": 7.360478802966131e-07,
|
| 32026 |
+
"loss": 0.7837,
|
| 32027 |
+
"step": 4574
|
| 32028 |
+
},
|
| 32029 |
+
{
|
| 32030 |
+
"epoch": 0.88,
|
| 32031 |
+
"grad_norm": 1.58320715655672,
|
| 32032 |
+
"learning_rate": 7.337011115505976e-07,
|
| 32033 |
+
"loss": 0.7945,
|
| 32034 |
+
"step": 4575
|
| 32035 |
+
},
|
| 32036 |
+
{
|
| 32037 |
+
"epoch": 0.88,
|
| 32038 |
+
"grad_norm": 1.6038227379699823,
|
| 32039 |
+
"learning_rate": 7.313579474465437e-07,
|
| 32040 |
+
"loss": 0.8834,
|
| 32041 |
+
"step": 4576
|
| 32042 |
+
},
|
| 32043 |
+
{
|
| 32044 |
+
"epoch": 0.88,
|
| 32045 |
+
"grad_norm": 1.701399427841146,
|
| 32046 |
+
"learning_rate": 7.290183888959557e-07,
|
| 32047 |
+
"loss": 0.8813,
|
| 32048 |
+
"step": 4577
|
| 32049 |
+
},
|
| 32050 |
+
{
|
| 32051 |
+
"epoch": 0.88,
|
| 32052 |
+
"grad_norm": 1.671771680207799,
|
| 32053 |
+
"learning_rate": 7.266824368089342e-07,
|
| 32054 |
+
"loss": 0.7681,
|
| 32055 |
+
"step": 4578
|
| 32056 |
+
},
|
| 32057 |
+
{
|
| 32058 |
+
"epoch": 0.88,
|
| 32059 |
+
"grad_norm": 0.9405244410682378,
|
| 32060 |
+
"learning_rate": 7.243500920941793e-07,
|
| 32061 |
+
"loss": 0.8179,
|
| 32062 |
+
"step": 4579
|
| 32063 |
+
},
|
| 32064 |
+
{
|
| 32065 |
+
"epoch": 0.88,
|
| 32066 |
+
"grad_norm": 1.6971553092278484,
|
| 32067 |
+
"learning_rate": 7.220213556589851e-07,
|
| 32068 |
+
"loss": 0.8041,
|
| 32069 |
+
"step": 4580
|
| 32070 |
+
},
|
| 32071 |
+
{
|
| 32072 |
+
"epoch": 0.88,
|
| 32073 |
+
"grad_norm": 1.6092722366159369,
|
| 32074 |
+
"learning_rate": 7.196962284092423e-07,
|
| 32075 |
+
"loss": 0.906,
|
| 32076 |
+
"step": 4581
|
| 32077 |
+
},
|
| 32078 |
+
{
|
| 32079 |
+
"epoch": 0.88,
|
| 32080 |
+
"grad_norm": 1.5790028990958835,
|
| 32081 |
+
"learning_rate": 7.173747112494389e-07,
|
| 32082 |
+
"loss": 0.9017,
|
| 32083 |
+
"step": 4582
|
| 32084 |
+
},
|
| 32085 |
+
{
|
| 32086 |
+
"epoch": 0.88,
|
| 32087 |
+
"grad_norm": 1.6613994467242639,
|
| 32088 |
+
"learning_rate": 7.150568050826579e-07,
|
| 32089 |
+
"loss": 0.7846,
|
| 32090 |
+
"step": 4583
|
| 32091 |
+
},
|
| 32092 |
+
{
|
| 32093 |
+
"epoch": 0.88,
|
| 32094 |
+
"grad_norm": 1.7813916794522948,
|
| 32095 |
+
"learning_rate": 7.12742510810579e-07,
|
| 32096 |
+
"loss": 0.8844,
|
| 32097 |
+
"step": 4584
|
| 32098 |
+
},
|
| 32099 |
+
{
|
| 32100 |
+
"epoch": 0.88,
|
| 32101 |
+
"grad_norm": 1.6133888168185018,
|
| 32102 |
+
"learning_rate": 7.104318293334733e-07,
|
| 32103 |
+
"loss": 0.8234,
|
| 32104 |
+
"step": 4585
|
| 32105 |
+
},
|
| 32106 |
+
{
|
| 32107 |
+
"epoch": 0.88,
|
| 32108 |
+
"grad_norm": 1.6638283699302128,
|
| 32109 |
+
"learning_rate": 7.081247615502107e-07,
|
| 32110 |
+
"loss": 0.8494,
|
| 32111 |
+
"step": 4586
|
| 32112 |
+
},
|
| 32113 |
+
{
|
| 32114 |
+
"epoch": 0.88,
|
| 32115 |
+
"grad_norm": 1.713196915699892,
|
| 32116 |
+
"learning_rate": 7.058213083582532e-07,
|
| 32117 |
+
"loss": 0.7312,
|
| 32118 |
+
"step": 4587
|
| 32119 |
+
},
|
| 32120 |
+
{
|
| 32121 |
+
"epoch": 0.88,
|
| 32122 |
+
"grad_norm": 1.5004791256135879,
|
| 32123 |
+
"learning_rate": 7.035214706536564e-07,
|
| 32124 |
+
"loss": 0.7935,
|
| 32125 |
+
"step": 4588
|
| 32126 |
+
},
|
| 32127 |
+
{
|
| 32128 |
+
"epoch": 0.88,
|
| 32129 |
+
"grad_norm": 1.5682227642510727,
|
| 32130 |
+
"learning_rate": 7.012252493310689e-07,
|
| 32131 |
+
"loss": 0.762,
|
| 32132 |
+
"step": 4589
|
| 32133 |
+
},
|
| 32134 |
+
{
|
| 32135 |
+
"epoch": 0.88,
|
| 32136 |
+
"grad_norm": 1.6640041441800169,
|
| 32137 |
+
"learning_rate": 6.989326452837364e-07,
|
| 32138 |
+
"loss": 0.7774,
|
| 32139 |
+
"step": 4590
|
| 32140 |
+
},
|
| 32141 |
+
{
|
| 32142 |
+
"epoch": 0.88,
|
| 32143 |
+
"grad_norm": 1.7428000494600797,
|
| 32144 |
+
"learning_rate": 6.96643659403492e-07,
|
| 32145 |
+
"loss": 0.863,
|
| 32146 |
+
"step": 4591
|
| 32147 |
+
},
|
| 32148 |
+
{
|
| 32149 |
+
"epoch": 0.88,
|
| 32150 |
+
"grad_norm": 1.7553837389304079,
|
| 32151 |
+
"learning_rate": 6.943582925807646e-07,
|
| 32152 |
+
"loss": 0.8168,
|
| 32153 |
+
"step": 4592
|
| 32154 |
+
},
|
| 32155 |
+
{
|
| 32156 |
+
"epoch": 0.88,
|
| 32157 |
+
"grad_norm": 1.5911234884094496,
|
| 32158 |
+
"learning_rate": 6.920765457045753e-07,
|
| 32159 |
+
"loss": 0.7925,
|
| 32160 |
+
"step": 4593
|
| 32161 |
+
},
|
| 32162 |
+
{
|
| 32163 |
+
"epoch": 0.88,
|
| 32164 |
+
"grad_norm": 1.6509026867265733,
|
| 32165 |
+
"learning_rate": 6.897984196625385e-07,
|
| 32166 |
+
"loss": 0.871,
|
| 32167 |
+
"step": 4594
|
| 32168 |
+
},
|
| 32169 |
+
{
|
| 32170 |
+
"epoch": 0.88,
|
| 32171 |
+
"grad_norm": 0.9510394161604407,
|
| 32172 |
+
"learning_rate": 6.875239153408541e-07,
|
| 32173 |
+
"loss": 0.8161,
|
| 32174 |
+
"step": 4595
|
| 32175 |
+
},
|
| 32176 |
+
{
|
| 32177 |
+
"epoch": 0.89,
|
| 32178 |
+
"grad_norm": 1.536910091669346,
|
| 32179 |
+
"learning_rate": 6.852530336243179e-07,
|
| 32180 |
+
"loss": 0.8231,
|
| 32181 |
+
"step": 4596
|
| 32182 |
+
},
|
| 32183 |
+
{
|
| 32184 |
+
"epoch": 0.89,
|
| 32185 |
+
"grad_norm": 1.6302778327684713,
|
| 32186 |
+
"learning_rate": 6.829857753963154e-07,
|
| 32187 |
+
"loss": 0.8091,
|
| 32188 |
+
"step": 4597
|
| 32189 |
+
},
|
| 32190 |
+
{
|
| 32191 |
+
"epoch": 0.89,
|
| 32192 |
+
"grad_norm": 1.6899195310445059,
|
| 32193 |
+
"learning_rate": 6.807221415388243e-07,
|
| 32194 |
+
"loss": 0.8516,
|
| 32195 |
+
"step": 4598
|
| 32196 |
+
},
|
| 32197 |
+
{
|
| 32198 |
+
"epoch": 0.89,
|
| 32199 |
+
"grad_norm": 1.6688393652366873,
|
| 32200 |
+
"learning_rate": 6.784621329324104e-07,
|
| 32201 |
+
"loss": 0.8116,
|
| 32202 |
+
"step": 4599
|
| 32203 |
+
},
|
| 32204 |
+
{
|
| 32205 |
+
"epoch": 0.89,
|
| 32206 |
+
"grad_norm": 1.6968648933180437,
|
| 32207 |
+
"learning_rate": 6.76205750456228e-07,
|
| 32208 |
+
"loss": 0.8566,
|
| 32209 |
+
"step": 4600
|
| 32210 |
+
},
|
| 32211 |
+
{
|
| 32212 |
+
"epoch": 0.89,
|
| 32213 |
+
"grad_norm": 1.699073788028988,
|
| 32214 |
+
"learning_rate": 6.739529949880263e-07,
|
| 32215 |
+
"loss": 0.9198,
|
| 32216 |
+
"step": 4601
|
| 32217 |
+
},
|
| 32218 |
+
{
|
| 32219 |
+
"epoch": 0.89,
|
| 32220 |
+
"grad_norm": 1.0303363034885953,
|
| 32221 |
+
"learning_rate": 6.717038674041354e-07,
|
| 32222 |
+
"loss": 0.8691,
|
| 32223 |
+
"step": 4602
|
| 32224 |
+
},
|
| 32225 |
+
{
|
| 32226 |
+
"epoch": 0.89,
|
| 32227 |
+
"grad_norm": 1.7700745086786087,
|
| 32228 |
+
"learning_rate": 6.694583685794798e-07,
|
| 32229 |
+
"loss": 0.8631,
|
| 32230 |
+
"step": 4603
|
| 32231 |
+
},
|
| 32232 |
+
{
|
| 32233 |
+
"epoch": 0.89,
|
| 32234 |
+
"grad_norm": 1.7156230999545747,
|
| 32235 |
+
"learning_rate": 6.672164993875707e-07,
|
| 32236 |
+
"loss": 0.8802,
|
| 32237 |
+
"step": 4604
|
| 32238 |
+
},
|
| 32239 |
+
{
|
| 32240 |
+
"epoch": 0.89,
|
| 32241 |
+
"grad_norm": 1.7587124281022528,
|
| 32242 |
+
"learning_rate": 6.649782607005095e-07,
|
| 32243 |
+
"loss": 0.8197,
|
| 32244 |
+
"step": 4605
|
| 32245 |
+
},
|
| 32246 |
+
{
|
| 32247 |
+
"epoch": 0.89,
|
| 32248 |
+
"grad_norm": 1.5890448851947154,
|
| 32249 |
+
"learning_rate": 6.62743653388982e-07,
|
| 32250 |
+
"loss": 0.7421,
|
| 32251 |
+
"step": 4606
|
| 32252 |
+
},
|
| 32253 |
+
{
|
| 32254 |
+
"epoch": 0.89,
|
| 32255 |
+
"grad_norm": 1.6941583093864598,
|
| 32256 |
+
"learning_rate": 6.605126783222637e-07,
|
| 32257 |
+
"loss": 0.9055,
|
| 32258 |
+
"step": 4607
|
| 32259 |
+
},
|
| 32260 |
+
{
|
| 32261 |
+
"epoch": 0.89,
|
| 32262 |
+
"grad_norm": 1.9345631981398188,
|
| 32263 |
+
"learning_rate": 6.582853363682184e-07,
|
| 32264 |
+
"loss": 0.8264,
|
| 32265 |
+
"step": 4608
|
| 32266 |
+
},
|
| 32267 |
+
{
|
| 32268 |
+
"epoch": 0.89,
|
| 32269 |
+
"grad_norm": 1.5727895120331519,
|
| 32270 |
+
"learning_rate": 6.560616283932897e-07,
|
| 32271 |
+
"loss": 0.8729,
|
| 32272 |
+
"step": 4609
|
| 32273 |
+
},
|
| 32274 |
+
{
|
| 32275 |
+
"epoch": 0.89,
|
| 32276 |
+
"grad_norm": 1.7270552928379281,
|
| 32277 |
+
"learning_rate": 6.538415552625143e-07,
|
| 32278 |
+
"loss": 0.839,
|
| 32279 |
+
"step": 4610
|
| 32280 |
+
},
|
| 32281 |
+
{
|
| 32282 |
+
"epoch": 0.89,
|
| 32283 |
+
"grad_norm": 1.6423156022324654,
|
| 32284 |
+
"learning_rate": 6.51625117839515e-07,
|
| 32285 |
+
"loss": 0.9414,
|
| 32286 |
+
"step": 4611
|
| 32287 |
+
},
|
| 32288 |
+
{
|
| 32289 |
+
"epoch": 0.89,
|
| 32290 |
+
"grad_norm": 1.588099616516975,
|
| 32291 |
+
"learning_rate": 6.494123169864964e-07,
|
| 32292 |
+
"loss": 0.9061,
|
| 32293 |
+
"step": 4612
|
| 32294 |
+
},
|
| 32295 |
+
{
|
| 32296 |
+
"epoch": 0.89,
|
| 32297 |
+
"grad_norm": 1.6908422858313596,
|
| 32298 |
+
"learning_rate": 6.472031535642509e-07,
|
| 32299 |
+
"loss": 0.8322,
|
| 32300 |
+
"step": 4613
|
| 32301 |
+
},
|
| 32302 |
+
{
|
| 32303 |
+
"epoch": 0.89,
|
| 32304 |
+
"grad_norm": 1.6507473606262408,
|
| 32305 |
+
"learning_rate": 6.449976284321547e-07,
|
| 32306 |
+
"loss": 0.8885,
|
| 32307 |
+
"step": 4614
|
| 32308 |
+
},
|
| 32309 |
+
{
|
| 32310 |
+
"epoch": 0.89,
|
| 32311 |
+
"grad_norm": 1.6452922413130318,
|
| 32312 |
+
"learning_rate": 6.427957424481724e-07,
|
| 32313 |
+
"loss": 0.7852,
|
| 32314 |
+
"step": 4615
|
| 32315 |
+
},
|
| 32316 |
+
{
|
| 32317 |
+
"epoch": 0.89,
|
| 32318 |
+
"grad_norm": 1.5906868583803813,
|
| 32319 |
+
"learning_rate": 6.405974964688477e-07,
|
| 32320 |
+
"loss": 0.8616,
|
| 32321 |
+
"step": 4616
|
| 32322 |
+
},
|
| 32323 |
+
{
|
| 32324 |
+
"epoch": 0.89,
|
| 32325 |
+
"grad_norm": 1.8525527893589988,
|
| 32326 |
+
"learning_rate": 6.38402891349309e-07,
|
| 32327 |
+
"loss": 0.8991,
|
| 32328 |
+
"step": 4617
|
| 32329 |
+
},
|
| 32330 |
+
{
|
| 32331 |
+
"epoch": 0.89,
|
| 32332 |
+
"grad_norm": 1.77877164245849,
|
| 32333 |
+
"learning_rate": 6.36211927943271e-07,
|
| 32334 |
+
"loss": 0.92,
|
| 32335 |
+
"step": 4618
|
| 32336 |
+
},
|
| 32337 |
+
{
|
| 32338 |
+
"epoch": 0.89,
|
| 32339 |
+
"grad_norm": 1.679959528587235,
|
| 32340 |
+
"learning_rate": 6.34024607103032e-07,
|
| 32341 |
+
"loss": 0.8461,
|
| 32342 |
+
"step": 4619
|
| 32343 |
+
},
|
| 32344 |
+
{
|
| 32345 |
+
"epoch": 0.89,
|
| 32346 |
+
"grad_norm": 1.650275475813174,
|
| 32347 |
+
"learning_rate": 6.318409296794703e-07,
|
| 32348 |
+
"loss": 0.7834,
|
| 32349 |
+
"step": 4620
|
| 32350 |
+
},
|
| 32351 |
+
{
|
| 32352 |
+
"epoch": 0.89,
|
| 32353 |
+
"grad_norm": 1.7103844161253232,
|
| 32354 |
+
"learning_rate": 6.296608965220496e-07,
|
| 32355 |
+
"loss": 0.8533,
|
| 32356 |
+
"step": 4621
|
| 32357 |
+
},
|
| 32358 |
+
{
|
| 32359 |
+
"epoch": 0.89,
|
| 32360 |
+
"grad_norm": 1.5560480879884606,
|
| 32361 |
+
"learning_rate": 6.274845084788151e-07,
|
| 32362 |
+
"loss": 0.8134,
|
| 32363 |
+
"step": 4622
|
| 32364 |
+
},
|
| 32365 |
+
{
|
| 32366 |
+
"epoch": 0.89,
|
| 32367 |
+
"grad_norm": 1.6758532716694152,
|
| 32368 |
+
"learning_rate": 6.253117663963948e-07,
|
| 32369 |
+
"loss": 0.8064,
|
| 32370 |
+
"step": 4623
|
| 32371 |
+
},
|
| 32372 |
+
{
|
| 32373 |
+
"epoch": 0.89,
|
| 32374 |
+
"grad_norm": 1.703648351870057,
|
| 32375 |
+
"learning_rate": 6.23142671119995e-07,
|
| 32376 |
+
"loss": 0.8758,
|
| 32377 |
+
"step": 4624
|
| 32378 |
+
},
|
| 32379 |
+
{
|
| 32380 |
+
"epoch": 0.89,
|
| 32381 |
+
"grad_norm": 1.7443635783563851,
|
| 32382 |
+
"learning_rate": 6.209772234934075e-07,
|
| 32383 |
+
"loss": 0.9093,
|
| 32384 |
+
"step": 4625
|
| 32385 |
+
},
|
| 32386 |
+
{
|
| 32387 |
+
"epoch": 0.89,
|
| 32388 |
+
"grad_norm": 1.6349407103004254,
|
| 32389 |
+
"learning_rate": 6.188154243590028e-07,
|
| 32390 |
+
"loss": 0.8695,
|
| 32391 |
+
"step": 4626
|
| 32392 |
+
},
|
| 32393 |
+
{
|
| 32394 |
+
"epoch": 0.89,
|
| 32395 |
+
"grad_norm": 1.7170681228786828,
|
| 32396 |
+
"learning_rate": 6.166572745577326e-07,
|
| 32397 |
+
"loss": 0.8604,
|
| 32398 |
+
"step": 4627
|
| 32399 |
+
},
|
| 32400 |
+
{
|
| 32401 |
+
"epoch": 0.89,
|
| 32402 |
+
"grad_norm": 1.6149701805074526,
|
| 32403 |
+
"learning_rate": 6.145027749291299e-07,
|
| 32404 |
+
"loss": 0.7834,
|
| 32405 |
+
"step": 4628
|
| 32406 |
+
},
|
| 32407 |
+
{
|
| 32408 |
+
"epoch": 0.89,
|
| 32409 |
+
"grad_norm": 0.9629179919446664,
|
| 32410 |
+
"learning_rate": 6.123519263113075e-07,
|
| 32411 |
+
"loss": 0.7807,
|
| 32412 |
+
"step": 4629
|
| 32413 |
+
},
|
| 32414 |
+
{
|
| 32415 |
+
"epoch": 0.89,
|
| 32416 |
+
"grad_norm": 1.4310449112443988,
|
| 32417 |
+
"learning_rate": 6.102047295409585e-07,
|
| 32418 |
+
"loss": 0.8408,
|
| 32419 |
+
"step": 4630
|
| 32420 |
+
},
|
| 32421 |
+
{
|
| 32422 |
+
"epoch": 0.89,
|
| 32423 |
+
"grad_norm": 1.431591253406188,
|
| 32424 |
+
"learning_rate": 6.080611854533514e-07,
|
| 32425 |
+
"loss": 0.7416,
|
| 32426 |
+
"step": 4631
|
| 32427 |
+
},
|
| 32428 |
+
{
|
| 32429 |
+
"epoch": 0.89,
|
| 32430 |
+
"grad_norm": 1.452024940046649,
|
| 32431 |
+
"learning_rate": 6.059212948823379e-07,
|
| 32432 |
+
"loss": 0.7764,
|
| 32433 |
+
"step": 4632
|
| 32434 |
+
},
|
| 32435 |
+
{
|
| 32436 |
+
"epoch": 0.89,
|
| 32437 |
+
"grad_norm": 1.6641309102131858,
|
| 32438 |
+
"learning_rate": 6.037850586603478e-07,
|
| 32439 |
+
"loss": 0.9695,
|
| 32440 |
+
"step": 4633
|
| 32441 |
+
},
|
| 32442 |
+
{
|
| 32443 |
+
"epoch": 0.89,
|
| 32444 |
+
"grad_norm": 1.5963484121260585,
|
| 32445 |
+
"learning_rate": 6.016524776183886e-07,
|
| 32446 |
+
"loss": 0.77,
|
| 32447 |
+
"step": 4634
|
| 32448 |
+
},
|
| 32449 |
+
{
|
| 32450 |
+
"epoch": 0.89,
|
| 32451 |
+
"grad_norm": 1.7710370822654729,
|
| 32452 |
+
"learning_rate": 5.995235525860466e-07,
|
| 32453 |
+
"loss": 0.835,
|
| 32454 |
+
"step": 4635
|
| 32455 |
+
},
|
| 32456 |
+
{
|
| 32457 |
+
"epoch": 0.89,
|
| 32458 |
+
"grad_norm": 0.9238054758388379,
|
| 32459 |
+
"learning_rate": 5.97398284391486e-07,
|
| 32460 |
+
"loss": 0.8299,
|
| 32461 |
+
"step": 4636
|
| 32462 |
+
},
|
| 32463 |
+
{
|
| 32464 |
+
"epoch": 0.89,
|
| 32465 |
+
"grad_norm": 1.732216941168927,
|
| 32466 |
+
"learning_rate": 5.952766738614468e-07,
|
| 32467 |
+
"loss": 0.8778,
|
| 32468 |
+
"step": 4637
|
| 32469 |
+
},
|
| 32470 |
+
{
|
| 32471 |
+
"epoch": 0.89,
|
| 32472 |
+
"grad_norm": 1.618882221464819,
|
| 32473 |
+
"learning_rate": 5.931587218212498e-07,
|
| 32474 |
+
"loss": 0.7453,
|
| 32475 |
+
"step": 4638
|
| 32476 |
+
},
|
| 32477 |
+
{
|
| 32478 |
+
"epoch": 0.89,
|
| 32479 |
+
"grad_norm": 1.7275497499921202,
|
| 32480 |
+
"learning_rate": 5.91044429094787e-07,
|
| 32481 |
+
"loss": 0.8808,
|
| 32482 |
+
"step": 4639
|
| 32483 |
+
},
|
| 32484 |
+
{
|
| 32485 |
+
"epoch": 0.89,
|
| 32486 |
+
"grad_norm": 1.6947725598628667,
|
| 32487 |
+
"learning_rate": 5.889337965045305e-07,
|
| 32488 |
+
"loss": 0.817,
|
| 32489 |
+
"step": 4640
|
| 32490 |
+
},
|
| 32491 |
+
{
|
| 32492 |
+
"epoch": 0.89,
|
| 32493 |
+
"grad_norm": 1.645115300544802,
|
| 32494 |
+
"learning_rate": 5.868268248715292e-07,
|
| 32495 |
+
"loss": 0.8123,
|
| 32496 |
+
"step": 4641
|
| 32497 |
+
},
|
| 32498 |
+
{
|
| 32499 |
+
"epoch": 0.89,
|
| 32500 |
+
"grad_norm": 1.552338918938507,
|
| 32501 |
+
"learning_rate": 5.847235150154074e-07,
|
| 32502 |
+
"loss": 0.8338,
|
| 32503 |
+
"step": 4642
|
| 32504 |
+
},
|
| 32505 |
+
{
|
| 32506 |
+
"epoch": 0.89,
|
| 32507 |
+
"grad_norm": 1.6695952907682257,
|
| 32508 |
+
"learning_rate": 5.826238677543628e-07,
|
| 32509 |
+
"loss": 0.8792,
|
| 32510 |
+
"step": 4643
|
| 32511 |
+
},
|
| 32512 |
+
{
|
| 32513 |
+
"epoch": 0.89,
|
| 32514 |
+
"grad_norm": 1.768852515536146,
|
| 32515 |
+
"learning_rate": 5.805278839051709e-07,
|
| 32516 |
+
"loss": 0.8514,
|
| 32517 |
+
"step": 4644
|
| 32518 |
+
},
|
| 32519 |
+
{
|
| 32520 |
+
"epoch": 0.89,
|
| 32521 |
+
"grad_norm": 1.4671139405698723,
|
| 32522 |
+
"learning_rate": 5.784355642831829e-07,
|
| 32523 |
+
"loss": 0.771,
|
| 32524 |
+
"step": 4645
|
| 32525 |
+
},
|
| 32526 |
+
{
|
| 32527 |
+
"epoch": 0.89,
|
| 32528 |
+
"grad_norm": 1.7568780187384645,
|
| 32529 |
+
"learning_rate": 5.763469097023189e-07,
|
| 32530 |
+
"loss": 0.8898,
|
| 32531 |
+
"step": 4646
|
| 32532 |
+
},
|
| 32533 |
+
{
|
| 32534 |
+
"epoch": 0.89,
|
| 32535 |
+
"grad_norm": 0.9703185609130268,
|
| 32536 |
+
"learning_rate": 5.742619209750788e-07,
|
| 32537 |
+
"loss": 0.8261,
|
| 32538 |
+
"step": 4647
|
| 32539 |
+
},
|
| 32540 |
+
{
|
| 32541 |
+
"epoch": 0.9,
|
| 32542 |
+
"grad_norm": 1.552615942105755,
|
| 32543 |
+
"learning_rate": 5.721805989125362e-07,
|
| 32544 |
+
"loss": 0.84,
|
| 32545 |
+
"step": 4648
|
| 32546 |
+
},
|
| 32547 |
+
{
|
| 32548 |
+
"epoch": 0.9,
|
| 32549 |
+
"grad_norm": 1.566663525662729,
|
| 32550 |
+
"learning_rate": 5.701029443243345e-07,
|
| 32551 |
+
"loss": 0.7267,
|
| 32552 |
+
"step": 4649
|
| 32553 |
+
},
|
| 32554 |
+
{
|
| 32555 |
+
"epoch": 0.9,
|
| 32556 |
+
"grad_norm": 1.8205065610738194,
|
| 32557 |
+
"learning_rate": 5.680289580186959e-07,
|
| 32558 |
+
"loss": 0.8879,
|
| 32559 |
+
"step": 4650
|
| 32560 |
+
},
|
| 32561 |
+
{
|
| 32562 |
+
"epoch": 0.9,
|
| 32563 |
+
"grad_norm": 1.654652335755769,
|
| 32564 |
+
"learning_rate": 5.659586408024098e-07,
|
| 32565 |
+
"loss": 0.8391,
|
| 32566 |
+
"step": 4651
|
| 32567 |
+
},
|
| 32568 |
+
{
|
| 32569 |
+
"epoch": 0.9,
|
| 32570 |
+
"grad_norm": 1.6905002178206892,
|
| 32571 |
+
"learning_rate": 5.63891993480844e-07,
|
| 32572 |
+
"loss": 0.8697,
|
| 32573 |
+
"step": 4652
|
| 32574 |
+
},
|
| 32575 |
+
{
|
| 32576 |
+
"epoch": 0.9,
|
| 32577 |
+
"grad_norm": 0.9546336563075242,
|
| 32578 |
+
"learning_rate": 5.618290168579333e-07,
|
| 32579 |
+
"loss": 0.7947,
|
| 32580 |
+
"step": 4653
|
| 32581 |
+
},
|
| 32582 |
+
{
|
| 32583 |
+
"epoch": 0.9,
|
| 32584 |
+
"grad_norm": 1.6926962772920793,
|
| 32585 |
+
"learning_rate": 5.597697117361877e-07,
|
| 32586 |
+
"loss": 0.8929,
|
| 32587 |
+
"step": 4654
|
| 32588 |
+
},
|
| 32589 |
+
{
|
| 32590 |
+
"epoch": 0.9,
|
| 32591 |
+
"grad_norm": 1.7332662858890597,
|
| 32592 |
+
"learning_rate": 5.57714078916689e-07,
|
| 32593 |
+
"loss": 0.9356,
|
| 32594 |
+
"step": 4655
|
| 32595 |
+
},
|
| 32596 |
+
{
|
| 32597 |
+
"epoch": 0.9,
|
| 32598 |
+
"grad_norm": 1.8157951920257163,
|
| 32599 |
+
"learning_rate": 5.556621191990907e-07,
|
| 32600 |
+
"loss": 0.9756,
|
| 32601 |
+
"step": 4656
|
| 32602 |
+
},
|
| 32603 |
+
{
|
| 32604 |
+
"epoch": 0.9,
|
| 32605 |
+
"grad_norm": 1.7425133604802003,
|
| 32606 |
+
"learning_rate": 5.536138333816144e-07,
|
| 32607 |
+
"loss": 0.7792,
|
| 32608 |
+
"step": 4657
|
| 32609 |
+
},
|
| 32610 |
+
{
|
| 32611 |
+
"epoch": 0.9,
|
| 32612 |
+
"grad_norm": 1.5621644307282403,
|
| 32613 |
+
"learning_rate": 5.515692222610558e-07,
|
| 32614 |
+
"loss": 0.7705,
|
| 32615 |
+
"step": 4658
|
| 32616 |
+
},
|
| 32617 |
+
{
|
| 32618 |
+
"epoch": 0.9,
|
| 32619 |
+
"grad_norm": 1.6356106516646511,
|
| 32620 |
+
"learning_rate": 5.495282866327812e-07,
|
| 32621 |
+
"loss": 0.8203,
|
| 32622 |
+
"step": 4659
|
| 32623 |
+
},
|
| 32624 |
+
{
|
| 32625 |
+
"epoch": 0.9,
|
| 32626 |
+
"grad_norm": 1.7357420054931105,
|
| 32627 |
+
"learning_rate": 5.474910272907242e-07,
|
| 32628 |
+
"loss": 0.8684,
|
| 32629 |
+
"step": 4660
|
| 32630 |
+
},
|
| 32631 |
+
{
|
| 32632 |
+
"epoch": 0.9,
|
| 32633 |
+
"grad_norm": 1.644574615849858,
|
| 32634 |
+
"learning_rate": 5.454574450273898e-07,
|
| 32635 |
+
"loss": 0.8575,
|
| 32636 |
+
"step": 4661
|
| 32637 |
+
},
|
| 32638 |
+
{
|
| 32639 |
+
"epoch": 0.9,
|
| 32640 |
+
"grad_norm": 1.5828708989675613,
|
| 32641 |
+
"learning_rate": 5.434275406338552e-07,
|
| 32642 |
+
"loss": 0.8204,
|
| 32643 |
+
"step": 4662
|
| 32644 |
+
},
|
| 32645 |
+
{
|
| 32646 |
+
"epoch": 0.9,
|
| 32647 |
+
"grad_norm": 1.4186623546018264,
|
| 32648 |
+
"learning_rate": 5.414013148997621e-07,
|
| 32649 |
+
"loss": 0.7704,
|
| 32650 |
+
"step": 4663
|
| 32651 |
+
},
|
| 32652 |
+
{
|
| 32653 |
+
"epoch": 0.9,
|
| 32654 |
+
"grad_norm": 1.6833154380110336,
|
| 32655 |
+
"learning_rate": 5.393787686133234e-07,
|
| 32656 |
+
"loss": 0.8795,
|
| 32657 |
+
"step": 4664
|
| 32658 |
+
},
|
| 32659 |
+
{
|
| 32660 |
+
"epoch": 0.9,
|
| 32661 |
+
"grad_norm": 1.0095255383815676,
|
| 32662 |
+
"learning_rate": 5.37359902561323e-07,
|
| 32663 |
+
"loss": 0.86,
|
| 32664 |
+
"step": 4665
|
| 32665 |
+
},
|
| 32666 |
+
{
|
| 32667 |
+
"epoch": 0.9,
|
| 32668 |
+
"grad_norm": 1.5918336391342192,
|
| 32669 |
+
"learning_rate": 5.353447175291104e-07,
|
| 32670 |
+
"loss": 0.8537,
|
| 32671 |
+
"step": 4666
|
| 32672 |
+
},
|
| 32673 |
+
{
|
| 32674 |
+
"epoch": 0.9,
|
| 32675 |
+
"grad_norm": 1.6224700248960138,
|
| 32676 |
+
"learning_rate": 5.333332143006043e-07,
|
| 32677 |
+
"loss": 0.8482,
|
| 32678 |
+
"step": 4667
|
| 32679 |
+
},
|
| 32680 |
+
{
|
| 32681 |
+
"epoch": 0.9,
|
| 32682 |
+
"grad_norm": 1.6704298241763658,
|
| 32683 |
+
"learning_rate": 5.313253936582885e-07,
|
| 32684 |
+
"loss": 0.7602,
|
| 32685 |
+
"step": 4668
|
| 32686 |
+
},
|
| 32687 |
+
{
|
| 32688 |
+
"epoch": 0.9,
|
| 32689 |
+
"grad_norm": 0.9501917910467306,
|
| 32690 |
+
"learning_rate": 5.293212563832173e-07,
|
| 32691 |
+
"loss": 0.7765,
|
| 32692 |
+
"step": 4669
|
| 32693 |
+
},
|
| 32694 |
+
{
|
| 32695 |
+
"epoch": 0.9,
|
| 32696 |
+
"grad_norm": 1.527283488559378,
|
| 32697 |
+
"learning_rate": 5.273208032550126e-07,
|
| 32698 |
+
"loss": 0.7678,
|
| 32699 |
+
"step": 4670
|
| 32700 |
+
},
|
| 32701 |
+
{
|
| 32702 |
+
"epoch": 0.9,
|
| 32703 |
+
"grad_norm": 1.6792519676900624,
|
| 32704 |
+
"learning_rate": 5.253240350518607e-07,
|
| 32705 |
+
"loss": 0.7365,
|
| 32706 |
+
"step": 4671
|
| 32707 |
+
},
|
| 32708 |
+
{
|
| 32709 |
+
"epoch": 0.9,
|
| 32710 |
+
"grad_norm": 1.64356865145821,
|
| 32711 |
+
"learning_rate": 5.233309525505159e-07,
|
| 32712 |
+
"loss": 0.8504,
|
| 32713 |
+
"step": 4672
|
| 32714 |
+
},
|
| 32715 |
+
{
|
| 32716 |
+
"epoch": 0.9,
|
| 32717 |
+
"grad_norm": 1.4855516463960838,
|
| 32718 |
+
"learning_rate": 5.213415565262981e-07,
|
| 32719 |
+
"loss": 0.7894,
|
| 32720 |
+
"step": 4673
|
| 32721 |
+
},
|
| 32722 |
+
{
|
| 32723 |
+
"epoch": 0.9,
|
| 32724 |
+
"grad_norm": 1.623938335284566,
|
| 32725 |
+
"learning_rate": 5.193558477530958e-07,
|
| 32726 |
+
"loss": 0.8903,
|
| 32727 |
+
"step": 4674
|
| 32728 |
+
},
|
| 32729 |
+
{
|
| 32730 |
+
"epoch": 0.9,
|
| 32731 |
+
"grad_norm": 1.6525083687168214,
|
| 32732 |
+
"learning_rate": 5.173738270033568e-07,
|
| 32733 |
+
"loss": 0.7674,
|
| 32734 |
+
"step": 4675
|
| 32735 |
+
},
|
| 32736 |
+
{
|
| 32737 |
+
"epoch": 0.9,
|
| 32738 |
+
"grad_norm": 1.471112511849222,
|
| 32739 |
+
"learning_rate": 5.153954950481e-07,
|
| 32740 |
+
"loss": 0.7344,
|
| 32741 |
+
"step": 4676
|
| 32742 |
+
},
|
| 32743 |
+
{
|
| 32744 |
+
"epoch": 0.9,
|
| 32745 |
+
"grad_norm": 1.6775328255178097,
|
| 32746 |
+
"learning_rate": 5.134208526569084e-07,
|
| 32747 |
+
"loss": 0.9133,
|
| 32748 |
+
"step": 4677
|
| 32749 |
+
},
|
| 32750 |
+
{
|
| 32751 |
+
"epoch": 0.9,
|
| 32752 |
+
"grad_norm": 1.687127162343473,
|
| 32753 |
+
"learning_rate": 5.114499005979279e-07,
|
| 32754 |
+
"loss": 0.8031,
|
| 32755 |
+
"step": 4678
|
| 32756 |
+
},
|
| 32757 |
+
{
|
| 32758 |
+
"epoch": 0.9,
|
| 32759 |
+
"grad_norm": 1.749454069288342,
|
| 32760 |
+
"learning_rate": 5.094826396378694e-07,
|
| 32761 |
+
"loss": 0.7052,
|
| 32762 |
+
"step": 4679
|
| 32763 |
+
},
|
| 32764 |
+
{
|
| 32765 |
+
"epoch": 0.9,
|
| 32766 |
+
"grad_norm": 1.7090967580697272,
|
| 32767 |
+
"learning_rate": 5.075190705420097e-07,
|
| 32768 |
+
"loss": 0.8282,
|
| 32769 |
+
"step": 4680
|
| 32770 |
+
},
|
| 32771 |
+
{
|
| 32772 |
+
"epoch": 0.9,
|
| 32773 |
+
"grad_norm": 1.6219991440125505,
|
| 32774 |
+
"learning_rate": 5.0555919407419e-07,
|
| 32775 |
+
"loss": 0.7687,
|
| 32776 |
+
"step": 4681
|
| 32777 |
+
},
|
| 32778 |
+
{
|
| 32779 |
+
"epoch": 0.9,
|
| 32780 |
+
"grad_norm": 1.712762287461227,
|
| 32781 |
+
"learning_rate": 5.036030109968082e-07,
|
| 32782 |
+
"loss": 0.9188,
|
| 32783 |
+
"step": 4682
|
| 32784 |
+
},
|
| 32785 |
+
{
|
| 32786 |
+
"epoch": 0.9,
|
| 32787 |
+
"grad_norm": 1.6465939602965196,
|
| 32788 |
+
"learning_rate": 5.016505220708334e-07,
|
| 32789 |
+
"loss": 0.91,
|
| 32790 |
+
"step": 4683
|
| 32791 |
+
},
|
| 32792 |
+
{
|
| 32793 |
+
"epoch": 0.9,
|
| 32794 |
+
"grad_norm": 1.6876805121272327,
|
| 32795 |
+
"learning_rate": 4.997017280557936e-07,
|
| 32796 |
+
"loss": 0.8541,
|
| 32797 |
+
"step": 4684
|
| 32798 |
+
},
|
| 32799 |
+
{
|
| 32800 |
+
"epoch": 0.9,
|
| 32801 |
+
"grad_norm": 1.5718236529911196,
|
| 32802 |
+
"learning_rate": 4.977566297097824e-07,
|
| 32803 |
+
"loss": 0.8048,
|
| 32804 |
+
"step": 4685
|
| 32805 |
+
},
|
| 32806 |
+
{
|
| 32807 |
+
"epoch": 0.9,
|
| 32808 |
+
"grad_norm": 1.7317068845916241,
|
| 32809 |
+
"learning_rate": 4.958152277894523e-07,
|
| 32810 |
+
"loss": 0.8608,
|
| 32811 |
+
"step": 4686
|
| 32812 |
+
},
|
| 32813 |
+
{
|
| 32814 |
+
"epoch": 0.9,
|
| 32815 |
+
"grad_norm": 1.5823403939243572,
|
| 32816 |
+
"learning_rate": 4.938775230500192e-07,
|
| 32817 |
+
"loss": 0.8097,
|
| 32818 |
+
"step": 4687
|
| 32819 |
+
},
|
| 32820 |
+
{
|
| 32821 |
+
"epoch": 0.9,
|
| 32822 |
+
"grad_norm": 1.7201556233252429,
|
| 32823 |
+
"learning_rate": 4.919435162452635e-07,
|
| 32824 |
+
"loss": 0.8875,
|
| 32825 |
+
"step": 4688
|
| 32826 |
+
},
|
| 32827 |
+
{
|
| 32828 |
+
"epoch": 0.9,
|
| 32829 |
+
"grad_norm": 1.5195980744396738,
|
| 32830 |
+
"learning_rate": 4.900132081275222e-07,
|
| 32831 |
+
"loss": 0.7755,
|
| 32832 |
+
"step": 4689
|
| 32833 |
+
},
|
| 32834 |
+
{
|
| 32835 |
+
"epoch": 0.9,
|
| 32836 |
+
"grad_norm": 1.6020899057466513,
|
| 32837 |
+
"learning_rate": 4.880865994476958e-07,
|
| 32838 |
+
"loss": 0.8562,
|
| 32839 |
+
"step": 4690
|
| 32840 |
+
},
|
| 32841 |
+
{
|
| 32842 |
+
"epoch": 0.9,
|
| 32843 |
+
"grad_norm": 1.806839673319644,
|
| 32844 |
+
"learning_rate": 4.861636909552469e-07,
|
| 32845 |
+
"loss": 0.7896,
|
| 32846 |
+
"step": 4691
|
| 32847 |
+
},
|
| 32848 |
+
{
|
| 32849 |
+
"epoch": 0.9,
|
| 32850 |
+
"grad_norm": 1.7078013031874621,
|
| 32851 |
+
"learning_rate": 4.842444833981985e-07,
|
| 32852 |
+
"loss": 0.8869,
|
| 32853 |
+
"step": 4692
|
| 32854 |
+
},
|
| 32855 |
+
{
|
| 32856 |
+
"epoch": 0.9,
|
| 32857 |
+
"grad_norm": 1.7614360589223628,
|
| 32858 |
+
"learning_rate": 4.82328977523131e-07,
|
| 32859 |
+
"loss": 0.8127,
|
| 32860 |
+
"step": 4693
|
| 32861 |
+
},
|
| 32862 |
+
{
|
| 32863 |
+
"epoch": 0.9,
|
| 32864 |
+
"grad_norm": 1.7649355206668866,
|
| 32865 |
+
"learning_rate": 4.804171740751873e-07,
|
| 32866 |
+
"loss": 0.9398,
|
| 32867 |
+
"step": 4694
|
| 32868 |
+
},
|
| 32869 |
+
{
|
| 32870 |
+
"epoch": 0.9,
|
| 32871 |
+
"grad_norm": 1.7582056140597686,
|
| 32872 |
+
"learning_rate": 4.785090737980725e-07,
|
| 32873 |
+
"loss": 0.9227,
|
| 32874 |
+
"step": 4695
|
| 32875 |
+
},
|
| 32876 |
+
{
|
| 32877 |
+
"epoch": 0.9,
|
| 32878 |
+
"grad_norm": 1.5040542605572478,
|
| 32879 |
+
"learning_rate": 4.766046774340438e-07,
|
| 32880 |
+
"loss": 0.8656,
|
| 32881 |
+
"step": 4696
|
| 32882 |
+
},
|
| 32883 |
+
{
|
| 32884 |
+
"epoch": 0.9,
|
| 32885 |
+
"grad_norm": 1.6405321539227562,
|
| 32886 |
+
"learning_rate": 4.747039857239244e-07,
|
| 32887 |
+
"loss": 0.797,
|
| 32888 |
+
"step": 4697
|
| 32889 |
+
},
|
| 32890 |
+
{
|
| 32891 |
+
"epoch": 0.9,
|
| 32892 |
+
"grad_norm": 0.9701252958307068,
|
| 32893 |
+
"learning_rate": 4.7280699940709275e-07,
|
| 32894 |
+
"loss": 0.8364,
|
| 32895 |
+
"step": 4698
|
| 32896 |
+
},
|
| 32897 |
+
{
|
| 32898 |
+
"epoch": 0.9,
|
| 32899 |
+
"grad_norm": 1.632269414247232,
|
| 32900 |
+
"learning_rate": 4.7091371922148966e-07,
|
| 32901 |
+
"loss": 0.8506,
|
| 32902 |
+
"step": 4699
|
| 32903 |
+
},
|
| 32904 |
+
{
|
| 32905 |
+
"epoch": 0.91,
|
| 32906 |
+
"grad_norm": 1.8161203221894053,
|
| 32907 |
+
"learning_rate": 4.6902414590360935e-07,
|
| 32908 |
+
"loss": 0.9253,
|
| 32909 |
+
"step": 4700
|
| 32910 |
+
},
|
| 32911 |
+
{
|
| 32912 |
+
"epoch": 0.91,
|
| 32913 |
+
"grad_norm": 1.591192668567659,
|
| 32914 |
+
"learning_rate": 4.6713828018850606e-07,
|
| 32915 |
+
"loss": 0.8823,
|
| 32916 |
+
"step": 4701
|
| 32917 |
+
},
|
| 32918 |
+
{
|
| 32919 |
+
"epoch": 0.91,
|
| 32920 |
+
"grad_norm": 1.753644938899693,
|
| 32921 |
+
"learning_rate": 4.6525612280979514e-07,
|
| 32922 |
+
"loss": 0.8554,
|
| 32923 |
+
"step": 4702
|
| 32924 |
+
},
|
| 32925 |
+
{
|
| 32926 |
+
"epoch": 0.91,
|
| 32927 |
+
"grad_norm": 1.7057079269067104,
|
| 32928 |
+
"learning_rate": 4.6337767449964323e-07,
|
| 32929 |
+
"loss": 0.8982,
|
| 32930 |
+
"step": 4703
|
| 32931 |
+
},
|
| 32932 |
+
{
|
| 32933 |
+
"epoch": 0.91,
|
| 32934 |
+
"grad_norm": 1.8081618193792555,
|
| 32935 |
+
"learning_rate": 4.6150293598877793e-07,
|
| 32936 |
+
"loss": 0.8193,
|
| 32937 |
+
"step": 4704
|
| 32938 |
+
},
|
| 32939 |
+
{
|
| 32940 |
+
"epoch": 0.91,
|
| 32941 |
+
"grad_norm": 1.7086633815503671,
|
| 32942 |
+
"learning_rate": 4.596319080064837e-07,
|
| 32943 |
+
"loss": 0.8331,
|
| 32944 |
+
"step": 4705
|
| 32945 |
+
},
|
| 32946 |
+
{
|
| 32947 |
+
"epoch": 0.91,
|
| 32948 |
+
"grad_norm": 1.6974864863612644,
|
| 32949 |
+
"learning_rate": 4.577645912806006e-07,
|
| 32950 |
+
"loss": 0.7984,
|
| 32951 |
+
"step": 4706
|
| 32952 |
+
},
|
| 32953 |
+
{
|
| 32954 |
+
"epoch": 0.91,
|
| 32955 |
+
"grad_norm": 1.643569617189755,
|
| 32956 |
+
"learning_rate": 4.559009865375241e-07,
|
| 32957 |
+
"loss": 0.8059,
|
| 32958 |
+
"step": 4707
|
| 32959 |
+
},
|
| 32960 |
+
{
|
| 32961 |
+
"epoch": 0.91,
|
| 32962 |
+
"grad_norm": 1.5728461903585766,
|
| 32963 |
+
"learning_rate": 4.5404109450221e-07,
|
| 32964 |
+
"loss": 0.7617,
|
| 32965 |
+
"step": 4708
|
| 32966 |
+
},
|
| 32967 |
+
{
|
| 32968 |
+
"epoch": 0.91,
|
| 32969 |
+
"grad_norm": 1.5168007933423657,
|
| 32970 |
+
"learning_rate": 4.521849158981628e-07,
|
| 32971 |
+
"loss": 0.7905,
|
| 32972 |
+
"step": 4709
|
| 32973 |
+
},
|
| 32974 |
+
{
|
| 32975 |
+
"epoch": 0.91,
|
| 32976 |
+
"grad_norm": 1.6654829931616189,
|
| 32977 |
+
"learning_rate": 4.503324514474483e-07,
|
| 32978 |
+
"loss": 0.8554,
|
| 32979 |
+
"step": 4710
|
| 32980 |
+
},
|
| 32981 |
+
{
|
| 32982 |
+
"epoch": 0.91,
|
| 32983 |
+
"grad_norm": 1.601183658916333,
|
| 32984 |
+
"learning_rate": 4.484837018706867e-07,
|
| 32985 |
+
"loss": 0.8237,
|
| 32986 |
+
"step": 4711
|
| 32987 |
+
},
|
| 32988 |
+
{
|
| 32989 |
+
"epoch": 0.91,
|
| 32990 |
+
"grad_norm": 1.7492383350646525,
|
| 32991 |
+
"learning_rate": 4.4663866788704823e-07,
|
| 32992 |
+
"loss": 0.9066,
|
| 32993 |
+
"step": 4712
|
| 32994 |
+
},
|
| 32995 |
+
{
|
| 32996 |
+
"epoch": 0.91,
|
| 32997 |
+
"grad_norm": 1.6652859699269564,
|
| 32998 |
+
"learning_rate": 4.447973502142644e-07,
|
| 32999 |
+
"loss": 0.8496,
|
| 33000 |
+
"step": 4713
|
| 33001 |
+
},
|
| 33002 |
+
{
|
| 33003 |
+
"epoch": 0.91,
|
| 33004 |
+
"grad_norm": 1.5176071936844036,
|
| 33005 |
+
"learning_rate": 4.4295974956861664e-07,
|
| 33006 |
+
"loss": 0.9279,
|
| 33007 |
+
"step": 4714
|
| 33008 |
+
},
|
| 33009 |
+
{
|
| 33010 |
+
"epoch": 0.91,
|
| 33011 |
+
"grad_norm": 1.7104988370209713,
|
| 33012 |
+
"learning_rate": 4.4112586666494317e-07,
|
| 33013 |
+
"loss": 0.8779,
|
| 33014 |
+
"step": 4715
|
| 33015 |
+
},
|
| 33016 |
+
{
|
| 33017 |
+
"epoch": 0.91,
|
| 33018 |
+
"grad_norm": 1.466684809772613,
|
| 33019 |
+
"learning_rate": 4.3929570221663444e-07,
|
| 33020 |
+
"loss": 0.8849,
|
| 33021 |
+
"step": 4716
|
| 33022 |
+
},
|
| 33023 |
+
{
|
| 33024 |
+
"epoch": 0.91,
|
| 33025 |
+
"grad_norm": 1.5064332377635345,
|
| 33026 |
+
"learning_rate": 4.3746925693563314e-07,
|
| 33027 |
+
"loss": 0.7313,
|
| 33028 |
+
"step": 4717
|
| 33029 |
+
},
|
| 33030 |
+
{
|
| 33031 |
+
"epoch": 0.91,
|
| 33032 |
+
"grad_norm": 1.6662312130875077,
|
| 33033 |
+
"learning_rate": 4.3564653153243875e-07,
|
| 33034 |
+
"loss": 0.8869,
|
| 33035 |
+
"step": 4718
|
| 33036 |
+
},
|
| 33037 |
+
{
|
| 33038 |
+
"epoch": 0.91,
|
| 33039 |
+
"grad_norm": 1.669135117581719,
|
| 33040 |
+
"learning_rate": 4.3382752671610075e-07,
|
| 33041 |
+
"loss": 0.8753,
|
| 33042 |
+
"step": 4719
|
| 33043 |
+
},
|
| 33044 |
+
{
|
| 33045 |
+
"epoch": 0.91,
|
| 33046 |
+
"grad_norm": 1.5739945493482388,
|
| 33047 |
+
"learning_rate": 4.3201224319422084e-07,
|
| 33048 |
+
"loss": 0.8615,
|
| 33049 |
+
"step": 4720
|
| 33050 |
+
},
|
| 33051 |
+
{
|
| 33052 |
+
"epoch": 0.91,
|
| 33053 |
+
"grad_norm": 1.6975091587403395,
|
| 33054 |
+
"learning_rate": 4.3020068167295646e-07,
|
| 33055 |
+
"loss": 0.8725,
|
| 33056 |
+
"step": 4721
|
| 33057 |
+
},
|
| 33058 |
+
{
|
| 33059 |
+
"epoch": 0.91,
|
| 33060 |
+
"grad_norm": 1.559228645302732,
|
| 33061 |
+
"learning_rate": 4.283928428570139e-07,
|
| 33062 |
+
"loss": 0.8113,
|
| 33063 |
+
"step": 4722
|
| 33064 |
+
},
|
| 33065 |
+
{
|
| 33066 |
+
"epoch": 0.91,
|
| 33067 |
+
"grad_norm": 1.4789161304081258,
|
| 33068 |
+
"learning_rate": 4.2658872744965273e-07,
|
| 33069 |
+
"loss": 0.8488,
|
| 33070 |
+
"step": 4723
|
| 33071 |
+
},
|
| 33072 |
+
{
|
| 33073 |
+
"epoch": 0.91,
|
| 33074 |
+
"grad_norm": 1.762455128063593,
|
| 33075 |
+
"learning_rate": 4.2478833615268386e-07,
|
| 33076 |
+
"loss": 0.8784,
|
| 33077 |
+
"step": 4724
|
| 33078 |
+
},
|
| 33079 |
+
{
|
| 33080 |
+
"epoch": 0.91,
|
| 33081 |
+
"grad_norm": 1.6819388276178813,
|
| 33082 |
+
"learning_rate": 4.2299166966647154e-07,
|
| 33083 |
+
"loss": 0.9057,
|
| 33084 |
+
"step": 4725
|
| 33085 |
+
},
|
| 33086 |
+
{
|
| 33087 |
+
"epoch": 0.91,
|
| 33088 |
+
"grad_norm": 1.5074004293968402,
|
| 33089 |
+
"learning_rate": 4.211987286899255e-07,
|
| 33090 |
+
"loss": 0.7763,
|
| 33091 |
+
"step": 4726
|
| 33092 |
+
},
|
| 33093 |
+
{
|
| 33094 |
+
"epoch": 0.91,
|
| 33095 |
+
"grad_norm": 1.6874112613914176,
|
| 33096 |
+
"learning_rate": 4.194095139205123e-07,
|
| 33097 |
+
"loss": 0.8848,
|
| 33098 |
+
"step": 4727
|
| 33099 |
+
},
|
| 33100 |
+
{
|
| 33101 |
+
"epoch": 0.91,
|
| 33102 |
+
"grad_norm": 1.4389099026076047,
|
| 33103 |
+
"learning_rate": 4.1762402605424526e-07,
|
| 33104 |
+
"loss": 0.8292,
|
| 33105 |
+
"step": 4728
|
| 33106 |
+
},
|
| 33107 |
+
{
|
| 33108 |
+
"epoch": 0.91,
|
| 33109 |
+
"grad_norm": 1.7523147070895562,
|
| 33110 |
+
"learning_rate": 4.1584226578568977e-07,
|
| 33111 |
+
"loss": 0.8041,
|
| 33112 |
+
"step": 4729
|
| 33113 |
+
},
|
| 33114 |
+
{
|
| 33115 |
+
"epoch": 0.91,
|
| 33116 |
+
"grad_norm": 1.5440277838586294,
|
| 33117 |
+
"learning_rate": 4.1406423380796037e-07,
|
| 33118 |
+
"loss": 0.842,
|
| 33119 |
+
"step": 4730
|
| 33120 |
+
},
|
| 33121 |
+
{
|
| 33122 |
+
"epoch": 0.91,
|
| 33123 |
+
"grad_norm": 1.683555768036824,
|
| 33124 |
+
"learning_rate": 4.122899308127215e-07,
|
| 33125 |
+
"loss": 0.7741,
|
| 33126 |
+
"step": 4731
|
| 33127 |
+
},
|
| 33128 |
+
{
|
| 33129 |
+
"epoch": 0.91,
|
| 33130 |
+
"grad_norm": 1.5537148224409472,
|
| 33131 |
+
"learning_rate": 4.105193574901878e-07,
|
| 33132 |
+
"loss": 0.8457,
|
| 33133 |
+
"step": 4732
|
| 33134 |
+
},
|
| 33135 |
+
{
|
| 33136 |
+
"epoch": 0.91,
|
| 33137 |
+
"grad_norm": 1.5393656799619777,
|
| 33138 |
+
"learning_rate": 4.087525145291205e-07,
|
| 33139 |
+
"loss": 0.8413,
|
| 33140 |
+
"step": 4733
|
| 33141 |
+
},
|
| 33142 |
+
{
|
| 33143 |
+
"epoch": 0.91,
|
| 33144 |
+
"grad_norm": 1.5918678095739118,
|
| 33145 |
+
"learning_rate": 4.0698940261683197e-07,
|
| 33146 |
+
"loss": 0.807,
|
| 33147 |
+
"step": 4734
|
| 33148 |
+
},
|
| 33149 |
+
{
|
| 33150 |
+
"epoch": 0.91,
|
| 33151 |
+
"grad_norm": 1.7728363838958756,
|
| 33152 |
+
"learning_rate": 4.052300224391825e-07,
|
| 33153 |
+
"loss": 0.7864,
|
| 33154 |
+
"step": 4735
|
| 33155 |
+
},
|
| 33156 |
+
{
|
| 33157 |
+
"epoch": 0.91,
|
| 33158 |
+
"grad_norm": 1.7482045193198998,
|
| 33159 |
+
"learning_rate": 4.0347437468058026e-07,
|
| 33160 |
+
"loss": 0.877,
|
| 33161 |
+
"step": 4736
|
| 33162 |
+
},
|
| 33163 |
+
{
|
| 33164 |
+
"epoch": 0.91,
|
| 33165 |
+
"grad_norm": 1.7525840635202183,
|
| 33166 |
+
"learning_rate": 4.017224600239833e-07,
|
| 33167 |
+
"loss": 0.8769,
|
| 33168 |
+
"step": 4737
|
| 33169 |
+
},
|
| 33170 |
+
{
|
| 33171 |
+
"epoch": 0.91,
|
| 33172 |
+
"grad_norm": 1.5071276409577965,
|
| 33173 |
+
"learning_rate": 3.999742791508965e-07,
|
| 33174 |
+
"loss": 0.8583,
|
| 33175 |
+
"step": 4738
|
| 33176 |
+
},
|
| 33177 |
+
{
|
| 33178 |
+
"epoch": 0.91,
|
| 33179 |
+
"grad_norm": 1.5494750382317546,
|
| 33180 |
+
"learning_rate": 3.9822983274137137e-07,
|
| 33181 |
+
"loss": 0.8042,
|
| 33182 |
+
"step": 4739
|
| 33183 |
+
},
|
| 33184 |
+
{
|
| 33185 |
+
"epoch": 0.91,
|
| 33186 |
+
"grad_norm": 1.7760013173719795,
|
| 33187 |
+
"learning_rate": 3.964891214740063e-07,
|
| 33188 |
+
"loss": 0.7957,
|
| 33189 |
+
"step": 4740
|
| 33190 |
+
},
|
| 33191 |
+
{
|
| 33192 |
+
"epoch": 0.91,
|
| 33193 |
+
"grad_norm": 1.616107440677217,
|
| 33194 |
+
"learning_rate": 3.9475214602594844e-07,
|
| 33195 |
+
"loss": 0.9369,
|
| 33196 |
+
"step": 4741
|
| 33197 |
+
},
|
| 33198 |
+
{
|
| 33199 |
+
"epoch": 0.91,
|
| 33200 |
+
"grad_norm": 1.6297856162850126,
|
| 33201 |
+
"learning_rate": 3.930189070728907e-07,
|
| 33202 |
+
"loss": 0.8166,
|
| 33203 |
+
"step": 4742
|
| 33204 |
+
},
|
| 33205 |
+
{
|
| 33206 |
+
"epoch": 0.91,
|
| 33207 |
+
"grad_norm": 1.5261404817116855,
|
| 33208 |
+
"learning_rate": 3.9128940528907256e-07,
|
| 33209 |
+
"loss": 0.882,
|
| 33210 |
+
"step": 4743
|
| 33211 |
+
},
|
| 33212 |
+
{
|
| 33213 |
+
"epoch": 0.91,
|
| 33214 |
+
"grad_norm": 1.6171829450573758,
|
| 33215 |
+
"learning_rate": 3.895636413472803e-07,
|
| 33216 |
+
"loss": 0.8086,
|
| 33217 |
+
"step": 4744
|
| 33218 |
+
},
|
| 33219 |
+
{
|
| 33220 |
+
"epoch": 0.91,
|
| 33221 |
+
"grad_norm": 1.65460452013808,
|
| 33222 |
+
"learning_rate": 3.878416159188458e-07,
|
| 33223 |
+
"loss": 0.9052,
|
| 33224 |
+
"step": 4745
|
| 33225 |
+
},
|
| 33226 |
+
{
|
| 33227 |
+
"epoch": 0.91,
|
| 33228 |
+
"grad_norm": 1.634016027582634,
|
| 33229 |
+
"learning_rate": 3.8612332967364776e-07,
|
| 33230 |
+
"loss": 0.8796,
|
| 33231 |
+
"step": 4746
|
| 33232 |
+
},
|
| 33233 |
+
{
|
| 33234 |
+
"epoch": 0.91,
|
| 33235 |
+
"grad_norm": 1.4705543286456317,
|
| 33236 |
+
"learning_rate": 3.844087832801069e-07,
|
| 33237 |
+
"loss": 0.8437,
|
| 33238 |
+
"step": 4747
|
| 33239 |
+
},
|
| 33240 |
+
{
|
| 33241 |
+
"epoch": 0.91,
|
| 33242 |
+
"grad_norm": 1.7151010171281107,
|
| 33243 |
+
"learning_rate": 3.826979774051909e-07,
|
| 33244 |
+
"loss": 0.8337,
|
| 33245 |
+
"step": 4748
|
| 33246 |
+
},
|
| 33247 |
+
{
|
| 33248 |
+
"epoch": 0.91,
|
| 33249 |
+
"grad_norm": 1.5843930520104088,
|
| 33250 |
+
"learning_rate": 3.809909127144151e-07,
|
| 33251 |
+
"loss": 0.8299,
|
| 33252 |
+
"step": 4749
|
| 33253 |
+
},
|
| 33254 |
+
{
|
| 33255 |
+
"epoch": 0.91,
|
| 33256 |
+
"grad_norm": 1.526416336262696,
|
| 33257 |
+
"learning_rate": 3.7928758987183624e-07,
|
| 33258 |
+
"loss": 0.7759,
|
| 33259 |
+
"step": 4750
|
| 33260 |
+
},
|
| 33261 |
+
{
|
| 33262 |
+
"epoch": 0.91,
|
| 33263 |
+
"grad_norm": 1.6428339947689987,
|
| 33264 |
+
"learning_rate": 3.7758800954005636e-07,
|
| 33265 |
+
"loss": 0.8362,
|
| 33266 |
+
"step": 4751
|
| 33267 |
+
},
|
| 33268 |
+
{
|
| 33269 |
+
"epoch": 0.92,
|
| 33270 |
+
"grad_norm": 1.6391771092396883,
|
| 33271 |
+
"learning_rate": 3.7589217238022113e-07,
|
| 33272 |
+
"loss": 0.8826,
|
| 33273 |
+
"step": 4752
|
| 33274 |
+
},
|
| 33275 |
+
{
|
| 33276 |
+
"epoch": 0.92,
|
| 33277 |
+
"grad_norm": 1.7667209649913396,
|
| 33278 |
+
"learning_rate": 3.7420007905202283e-07,
|
| 33279 |
+
"loss": 0.8631,
|
| 33280 |
+
"step": 4753
|
| 33281 |
+
},
|
| 33282 |
+
{
|
| 33283 |
+
"epoch": 0.92,
|
| 33284 |
+
"grad_norm": 1.5839302857640825,
|
| 33285 |
+
"learning_rate": 3.7251173021369156e-07,
|
| 33286 |
+
"loss": 0.8494,
|
| 33287 |
+
"step": 4754
|
| 33288 |
+
},
|
| 33289 |
+
{
|
| 33290 |
+
"epoch": 0.92,
|
| 33291 |
+
"grad_norm": 1.5713058747975237,
|
| 33292 |
+
"learning_rate": 3.708271265220087e-07,
|
| 33293 |
+
"loss": 0.7535,
|
| 33294 |
+
"step": 4755
|
| 33295 |
+
},
|
| 33296 |
+
{
|
| 33297 |
+
"epoch": 0.92,
|
| 33298 |
+
"grad_norm": 1.4581919015342857,
|
| 33299 |
+
"learning_rate": 3.6914626863229e-07,
|
| 33300 |
+
"loss": 0.7839,
|
| 33301 |
+
"step": 4756
|
| 33302 |
+
},
|
| 33303 |
+
{
|
| 33304 |
+
"epoch": 0.92,
|
| 33305 |
+
"grad_norm": 1.7593764793503315,
|
| 33306 |
+
"learning_rate": 3.674691571984013e-07,
|
| 33307 |
+
"loss": 0.8261,
|
| 33308 |
+
"step": 4757
|
| 33309 |
+
},
|
| 33310 |
+
{
|
| 33311 |
+
"epoch": 0.92,
|
| 33312 |
+
"grad_norm": 1.0229678649372016,
|
| 33313 |
+
"learning_rate": 3.657957928727474e-07,
|
| 33314 |
+
"loss": 0.8256,
|
| 33315 |
+
"step": 4758
|
| 33316 |
+
},
|
| 33317 |
+
{
|
| 33318 |
+
"epoch": 0.92,
|
| 33319 |
+
"grad_norm": 0.9535118258403272,
|
| 33320 |
+
"learning_rate": 3.6412617630627755e-07,
|
| 33321 |
+
"loss": 0.7838,
|
| 33322 |
+
"step": 4759
|
| 33323 |
+
},
|
| 33324 |
+
{
|
| 33325 |
+
"epoch": 0.92,
|
| 33326 |
+
"grad_norm": 1.7347847279444861,
|
| 33327 |
+
"learning_rate": 3.624603081484812e-07,
|
| 33328 |
+
"loss": 0.8139,
|
| 33329 |
+
"step": 4760
|
| 33330 |
+
},
|
| 33331 |
+
{
|
| 33332 |
+
"epoch": 0.92,
|
| 33333 |
+
"grad_norm": 1.622160056422886,
|
| 33334 |
+
"learning_rate": 3.6079818904738884e-07,
|
| 33335 |
+
"loss": 0.7831,
|
| 33336 |
+
"step": 4761
|
| 33337 |
+
},
|
| 33338 |
+
{
|
| 33339 |
+
"epoch": 0.92,
|
| 33340 |
+
"grad_norm": 0.9716276915514496,
|
| 33341 |
+
"learning_rate": 3.5913981964957766e-07,
|
| 33342 |
+
"loss": 0.8886,
|
| 33343 |
+
"step": 4762
|
| 33344 |
+
},
|
| 33345 |
+
{
|
| 33346 |
+
"epoch": 0.92,
|
| 33347 |
+
"grad_norm": 1.6441177659837407,
|
| 33348 |
+
"learning_rate": 3.5748520060015944e-07,
|
| 33349 |
+
"loss": 0.8599,
|
| 33350 |
+
"step": 4763
|
| 33351 |
+
},
|
| 33352 |
+
{
|
| 33353 |
+
"epoch": 0.92,
|
| 33354 |
+
"grad_norm": 1.6551197231737023,
|
| 33355 |
+
"learning_rate": 3.5583433254279155e-07,
|
| 33356 |
+
"loss": 0.8248,
|
| 33357 |
+
"step": 4764
|
| 33358 |
+
},
|
| 33359 |
+
{
|
| 33360 |
+
"epoch": 0.92,
|
| 33361 |
+
"grad_norm": 0.9760164605599994,
|
| 33362 |
+
"learning_rate": 3.541872161196691e-07,
|
| 33363 |
+
"loss": 0.7797,
|
| 33364 |
+
"step": 4765
|
| 33365 |
+
},
|
| 33366 |
+
{
|
| 33367 |
+
"epoch": 0.92,
|
| 33368 |
+
"grad_norm": 1.4504786189252332,
|
| 33369 |
+
"learning_rate": 3.5254385197153297e-07,
|
| 33370 |
+
"loss": 0.8549,
|
| 33371 |
+
"step": 4766
|
| 33372 |
+
},
|
| 33373 |
+
{
|
| 33374 |
+
"epoch": 0.92,
|
| 33375 |
+
"grad_norm": 1.6089778666465842,
|
| 33376 |
+
"learning_rate": 3.5090424073765725e-07,
|
| 33377 |
+
"loss": 0.9239,
|
| 33378 |
+
"step": 4767
|
| 33379 |
+
},
|
| 33380 |
+
{
|
| 33381 |
+
"epoch": 0.92,
|
| 33382 |
+
"grad_norm": 1.6161196157246704,
|
| 33383 |
+
"learning_rate": 3.4926838305586295e-07,
|
| 33384 |
+
"loss": 0.9207,
|
| 33385 |
+
"step": 4768
|
| 33386 |
+
},
|
| 33387 |
+
{
|
| 33388 |
+
"epoch": 0.92,
|
| 33389 |
+
"grad_norm": 1.6606225882806407,
|
| 33390 |
+
"learning_rate": 3.4763627956250654e-07,
|
| 33391 |
+
"loss": 0.7786,
|
| 33392 |
+
"step": 4769
|
| 33393 |
+
},
|
| 33394 |
+
{
|
| 33395 |
+
"epoch": 0.92,
|
| 33396 |
+
"grad_norm": 0.9505629813721634,
|
| 33397 |
+
"learning_rate": 3.4600793089248465e-07,
|
| 33398 |
+
"loss": 0.8017,
|
| 33399 |
+
"step": 4770
|
| 33400 |
+
},
|
| 33401 |
+
{
|
| 33402 |
+
"epoch": 0.92,
|
| 33403 |
+
"grad_norm": 1.8036875245945276,
|
| 33404 |
+
"learning_rate": 3.443833376792349e-07,
|
| 33405 |
+
"loss": 0.8881,
|
| 33406 |
+
"step": 4771
|
| 33407 |
+
},
|
| 33408 |
+
{
|
| 33409 |
+
"epoch": 0.92,
|
| 33410 |
+
"grad_norm": 1.694822329392725,
|
| 33411 |
+
"learning_rate": 3.4276250055473304e-07,
|
| 33412 |
+
"loss": 0.8805,
|
| 33413 |
+
"step": 4772
|
| 33414 |
+
},
|
| 33415 |
+
{
|
| 33416 |
+
"epoch": 0.92,
|
| 33417 |
+
"grad_norm": 1.5982949061577643,
|
| 33418 |
+
"learning_rate": 3.411454201494935e-07,
|
| 33419 |
+
"loss": 0.8112,
|
| 33420 |
+
"step": 4773
|
| 33421 |
+
},
|
| 33422 |
+
{
|
| 33423 |
+
"epoch": 0.92,
|
| 33424 |
+
"grad_norm": 1.575896533752538,
|
| 33425 |
+
"learning_rate": 3.3953209709256975e-07,
|
| 33426 |
+
"loss": 0.7669,
|
| 33427 |
+
"step": 4774
|
| 33428 |
+
},
|
| 33429 |
+
{
|
| 33430 |
+
"epoch": 0.92,
|
| 33431 |
+
"grad_norm": 1.7289475032328157,
|
| 33432 |
+
"learning_rate": 3.3792253201155313e-07,
|
| 33433 |
+
"loss": 0.8515,
|
| 33434 |
+
"step": 4775
|
| 33435 |
+
},
|
| 33436 |
+
{
|
| 33437 |
+
"epoch": 0.92,
|
| 33438 |
+
"grad_norm": 1.6772017195281679,
|
| 33439 |
+
"learning_rate": 3.36316725532575e-07,
|
| 33440 |
+
"loss": 0.9792,
|
| 33441 |
+
"step": 4776
|
| 33442 |
+
},
|
| 33443 |
+
{
|
| 33444 |
+
"epoch": 0.92,
|
| 33445 |
+
"grad_norm": 1.6483114346919296,
|
| 33446 |
+
"learning_rate": 3.3471467828030013e-07,
|
| 33447 |
+
"loss": 0.8247,
|
| 33448 |
+
"step": 4777
|
| 33449 |
+
},
|
| 33450 |
+
{
|
| 33451 |
+
"epoch": 0.92,
|
| 33452 |
+
"grad_norm": 1.7981315587309108,
|
| 33453 |
+
"learning_rate": 3.3311639087793556e-07,
|
| 33454 |
+
"loss": 0.9334,
|
| 33455 |
+
"step": 4778
|
| 33456 |
+
},
|
| 33457 |
+
{
|
| 33458 |
+
"epoch": 0.92,
|
| 33459 |
+
"grad_norm": 1.798395276634243,
|
| 33460 |
+
"learning_rate": 3.3152186394722506e-07,
|
| 33461 |
+
"loss": 0.8194,
|
| 33462 |
+
"step": 4779
|
| 33463 |
+
},
|
| 33464 |
+
{
|
| 33465 |
+
"epoch": 0.92,
|
| 33466 |
+
"grad_norm": 1.6884601725851665,
|
| 33467 |
+
"learning_rate": 3.299310981084469e-07,
|
| 33468 |
+
"loss": 0.8869,
|
| 33469 |
+
"step": 4780
|
| 33470 |
+
},
|
| 33471 |
+
{
|
| 33472 |
+
"epoch": 0.92,
|
| 33473 |
+
"grad_norm": 1.6532311253737053,
|
| 33474 |
+
"learning_rate": 3.283440939804172e-07,
|
| 33475 |
+
"loss": 0.7801,
|
| 33476 |
+
"step": 4781
|
| 33477 |
+
},
|
| 33478 |
+
{
|
| 33479 |
+
"epoch": 0.92,
|
| 33480 |
+
"grad_norm": 1.4639294831852823,
|
| 33481 |
+
"learning_rate": 3.2676085218049215e-07,
|
| 33482 |
+
"loss": 0.8718,
|
| 33483 |
+
"step": 4782
|
| 33484 |
+
},
|
| 33485 |
+
{
|
| 33486 |
+
"epoch": 0.92,
|
| 33487 |
+
"grad_norm": 1.688433245042595,
|
| 33488 |
+
"learning_rate": 3.251813733245601e-07,
|
| 33489 |
+
"loss": 0.7258,
|
| 33490 |
+
"step": 4783
|
| 33491 |
+
},
|
| 33492 |
+
{
|
| 33493 |
+
"epoch": 0.92,
|
| 33494 |
+
"grad_norm": 1.5282375927340333,
|
| 33495 |
+
"learning_rate": 3.2360565802704634e-07,
|
| 33496 |
+
"loss": 0.7776,
|
| 33497 |
+
"step": 4784
|
| 33498 |
+
},
|
| 33499 |
+
{
|
| 33500 |
+
"epoch": 0.92,
|
| 33501 |
+
"grad_norm": 1.4939423179271425,
|
| 33502 |
+
"learning_rate": 3.2203370690091385e-07,
|
| 33503 |
+
"loss": 0.8289,
|
| 33504 |
+
"step": 4785
|
| 33505 |
+
},
|
| 33506 |
+
{
|
| 33507 |
+
"epoch": 0.92,
|
| 33508 |
+
"grad_norm": 1.686477793286507,
|
| 33509 |
+
"learning_rate": 3.2046552055766014e-07,
|
| 33510 |
+
"loss": 0.8194,
|
| 33511 |
+
"step": 4786
|
| 33512 |
+
},
|
| 33513 |
+
{
|
| 33514 |
+
"epoch": 0.92,
|
| 33515 |
+
"grad_norm": 1.6118000752000259,
|
| 33516 |
+
"learning_rate": 3.189010996073183e-07,
|
| 33517 |
+
"loss": 0.8074,
|
| 33518 |
+
"step": 4787
|
| 33519 |
+
},
|
| 33520 |
+
{
|
| 33521 |
+
"epoch": 0.92,
|
| 33522 |
+
"grad_norm": 1.7914929045154324,
|
| 33523 |
+
"learning_rate": 3.17340444658456e-07,
|
| 33524 |
+
"loss": 0.846,
|
| 33525 |
+
"step": 4788
|
| 33526 |
+
},
|
| 33527 |
+
{
|
| 33528 |
+
"epoch": 0.92,
|
| 33529 |
+
"grad_norm": 1.6823572069118011,
|
| 33530 |
+
"learning_rate": 3.157835563181788e-07,
|
| 33531 |
+
"loss": 0.8335,
|
| 33532 |
+
"step": 4789
|
| 33533 |
+
},
|
| 33534 |
+
{
|
| 33535 |
+
"epoch": 0.92,
|
| 33536 |
+
"grad_norm": 1.6606169497019847,
|
| 33537 |
+
"learning_rate": 3.1423043519212546e-07,
|
| 33538 |
+
"loss": 0.7818,
|
| 33539 |
+
"step": 4790
|
| 33540 |
+
},
|
| 33541 |
+
{
|
| 33542 |
+
"epoch": 0.92,
|
| 33543 |
+
"grad_norm": 1.5313722890918302,
|
| 33544 |
+
"learning_rate": 3.12681081884465e-07,
|
| 33545 |
+
"loss": 0.8774,
|
| 33546 |
+
"step": 4791
|
| 33547 |
+
},
|
| 33548 |
+
{
|
| 33549 |
+
"epoch": 0.92,
|
| 33550 |
+
"grad_norm": 1.729672334609458,
|
| 33551 |
+
"learning_rate": 3.1113549699790745e-07,
|
| 33552 |
+
"loss": 0.7877,
|
| 33553 |
+
"step": 4792
|
| 33554 |
+
},
|
| 33555 |
+
{
|
| 33556 |
+
"epoch": 0.92,
|
| 33557 |
+
"grad_norm": 1.7275981350830443,
|
| 33558 |
+
"learning_rate": 3.0959368113369305e-07,
|
| 33559 |
+
"loss": 0.8518,
|
| 33560 |
+
"step": 4793
|
| 33561 |
+
},
|
| 33562 |
+
{
|
| 33563 |
+
"epoch": 0.92,
|
| 33564 |
+
"grad_norm": 1.5431849570033087,
|
| 33565 |
+
"learning_rate": 3.080556348915964e-07,
|
| 33566 |
+
"loss": 0.8513,
|
| 33567 |
+
"step": 4794
|
| 33568 |
+
},
|
| 33569 |
+
{
|
| 33570 |
+
"epoch": 0.92,
|
| 33571 |
+
"grad_norm": 1.5549587703130405,
|
| 33572 |
+
"learning_rate": 3.0652135886992674e-07,
|
| 33573 |
+
"loss": 0.845,
|
| 33574 |
+
"step": 4795
|
| 33575 |
+
},
|
| 33576 |
+
{
|
| 33577 |
+
"epoch": 0.92,
|
| 33578 |
+
"grad_norm": 1.6150632070900868,
|
| 33579 |
+
"learning_rate": 3.049908536655266e-07,
|
| 33580 |
+
"loss": 0.8748,
|
| 33581 |
+
"step": 4796
|
| 33582 |
+
},
|
| 33583 |
+
{
|
| 33584 |
+
"epoch": 0.92,
|
| 33585 |
+
"grad_norm": 1.7712407035313649,
|
| 33586 |
+
"learning_rate": 3.0346411987377087e-07,
|
| 33587 |
+
"loss": 0.8705,
|
| 33588 |
+
"step": 4797
|
| 33589 |
+
},
|
| 33590 |
+
{
|
| 33591 |
+
"epoch": 0.92,
|
| 33592 |
+
"grad_norm": 1.7296781843055429,
|
| 33593 |
+
"learning_rate": 3.019411580885656e-07,
|
| 33594 |
+
"loss": 0.8107,
|
| 33595 |
+
"step": 4798
|
| 33596 |
+
},
|
| 33597 |
+
{
|
| 33598 |
+
"epoch": 0.92,
|
| 33599 |
+
"grad_norm": 1.7321411729256602,
|
| 33600 |
+
"learning_rate": 3.0042196890235244e-07,
|
| 33601 |
+
"loss": 0.8312,
|
| 33602 |
+
"step": 4799
|
| 33603 |
+
},
|
| 33604 |
+
{
|
| 33605 |
+
"epoch": 0.92,
|
| 33606 |
+
"grad_norm": 1.6667705278133578,
|
| 33607 |
+
"learning_rate": 2.9890655290610524e-07,
|
| 33608 |
+
"loss": 0.8439,
|
| 33609 |
+
"step": 4800
|
| 33610 |
+
},
|
| 33611 |
+
{
|
| 33612 |
+
"epoch": 0.92,
|
| 33613 |
+
"grad_norm": 1.6183317675327349,
|
| 33614 |
+
"learning_rate": 2.9739491068932924e-07,
|
| 33615 |
+
"loss": 0.8938,
|
| 33616 |
+
"step": 4801
|
| 33617 |
+
},
|
| 33618 |
+
{
|
| 33619 |
+
"epoch": 0.92,
|
| 33620 |
+
"grad_norm": 1.6711648840775002,
|
| 33621 |
+
"learning_rate": 2.9588704284006176e-07,
|
| 33622 |
+
"loss": 0.9273,
|
| 33623 |
+
"step": 4802
|
| 33624 |
+
},
|
| 33625 |
+
{
|
| 33626 |
+
"epoch": 0.92,
|
| 33627 |
+
"grad_norm": 1.5800037148285875,
|
| 33628 |
+
"learning_rate": 2.9438294994487025e-07,
|
| 33629 |
+
"loss": 0.7868,
|
| 33630 |
+
"step": 4803
|
| 33631 |
+
},
|
| 33632 |
+
{
|
| 33633 |
+
"epoch": 0.93,
|
| 33634 |
+
"grad_norm": 1.674519887401665,
|
| 33635 |
+
"learning_rate": 2.9288263258885564e-07,
|
| 33636 |
+
"loss": 0.8108,
|
| 33637 |
+
"step": 4804
|
| 33638 |
+
},
|
| 33639 |
+
{
|
| 33640 |
+
"epoch": 0.93,
|
| 33641 |
+
"grad_norm": 1.685941013734219,
|
| 33642 |
+
"learning_rate": 2.913860913556521e-07,
|
| 33643 |
+
"loss": 0.8894,
|
| 33644 |
+
"step": 4805
|
| 33645 |
+
},
|
| 33646 |
+
{
|
| 33647 |
+
"epoch": 0.93,
|
| 33648 |
+
"grad_norm": 1.6992304001777496,
|
| 33649 |
+
"learning_rate": 2.8989332682741953e-07,
|
| 33650 |
+
"loss": 0.9151,
|
| 33651 |
+
"step": 4806
|
| 33652 |
+
},
|
| 33653 |
+
{
|
| 33654 |
+
"epoch": 0.93,
|
| 33655 |
+
"grad_norm": 1.5681189714737205,
|
| 33656 |
+
"learning_rate": 2.8840433958485346e-07,
|
| 33657 |
+
"loss": 0.828,
|
| 33658 |
+
"step": 4807
|
| 33659 |
+
},
|
| 33660 |
+
{
|
| 33661 |
+
"epoch": 0.93,
|
| 33662 |
+
"grad_norm": 1.73459313527006,
|
| 33663 |
+
"learning_rate": 2.869191302071772e-07,
|
| 33664 |
+
"loss": 0.8636,
|
| 33665 |
+
"step": 4808
|
| 33666 |
+
},
|
| 33667 |
+
{
|
| 33668 |
+
"epoch": 0.93,
|
| 33669 |
+
"grad_norm": 1.5708889933524866,
|
| 33670 |
+
"learning_rate": 2.8543769927214635e-07,
|
| 33671 |
+
"loss": 0.8016,
|
| 33672 |
+
"step": 4809
|
| 33673 |
+
},
|
| 33674 |
+
{
|
| 33675 |
+
"epoch": 0.93,
|
| 33676 |
+
"grad_norm": 1.6556950571345788,
|
| 33677 |
+
"learning_rate": 2.8396004735604556e-07,
|
| 33678 |
+
"loss": 0.8362,
|
| 33679 |
+
"step": 4810
|
| 33680 |
+
},
|
| 33681 |
+
{
|
| 33682 |
+
"epoch": 0.93,
|
| 33683 |
+
"grad_norm": 1.5939396280990095,
|
| 33684 |
+
"learning_rate": 2.8248617503368934e-07,
|
| 33685 |
+
"loss": 0.8214,
|
| 33686 |
+
"step": 4811
|
| 33687 |
+
},
|
| 33688 |
+
{
|
| 33689 |
+
"epoch": 0.93,
|
| 33690 |
+
"grad_norm": 1.6445737537214664,
|
| 33691 |
+
"learning_rate": 2.8101608287842255e-07,
|
| 33692 |
+
"loss": 0.8101,
|
| 33693 |
+
"step": 4812
|
| 33694 |
+
},
|
| 33695 |
+
{
|
| 33696 |
+
"epoch": 0.93,
|
| 33697 |
+
"grad_norm": 1.530900385278659,
|
| 33698 |
+
"learning_rate": 2.795497714621198e-07,
|
| 33699 |
+
"loss": 0.8873,
|
| 33700 |
+
"step": 4813
|
| 33701 |
+
},
|
| 33702 |
+
{
|
| 33703 |
+
"epoch": 0.93,
|
| 33704 |
+
"grad_norm": 1.6440147626071886,
|
| 33705 |
+
"learning_rate": 2.7808724135518275e-07,
|
| 33706 |
+
"loss": 0.8459,
|
| 33707 |
+
"step": 4814
|
| 33708 |
+
},
|
| 33709 |
+
{
|
| 33710 |
+
"epoch": 0.93,
|
| 33711 |
+
"grad_norm": 1.5967301329211467,
|
| 33712 |
+
"learning_rate": 2.76628493126545e-07,
|
| 33713 |
+
"loss": 0.8309,
|
| 33714 |
+
"step": 4815
|
| 33715 |
+
},
|
| 33716 |
+
{
|
| 33717 |
+
"epoch": 0.93,
|
| 33718 |
+
"grad_norm": 0.9060266887554025,
|
| 33719 |
+
"learning_rate": 2.751735273436673e-07,
|
| 33720 |
+
"loss": 0.7643,
|
| 33721 |
+
"step": 4816
|
| 33722 |
+
},
|
| 33723 |
+
{
|
| 33724 |
+
"epoch": 0.93,
|
| 33725 |
+
"grad_norm": 1.7546073748438287,
|
| 33726 |
+
"learning_rate": 2.737223445725401e-07,
|
| 33727 |
+
"loss": 0.8663,
|
| 33728 |
+
"step": 4817
|
| 33729 |
+
},
|
| 33730 |
+
{
|
| 33731 |
+
"epoch": 0.93,
|
| 33732 |
+
"grad_norm": 1.6244540188287706,
|
| 33733 |
+
"learning_rate": 2.7227494537768075e-07,
|
| 33734 |
+
"loss": 0.765,
|
| 33735 |
+
"step": 4818
|
| 33736 |
+
},
|
| 33737 |
+
{
|
| 33738 |
+
"epoch": 0.93,
|
| 33739 |
+
"grad_norm": 1.760202352517511,
|
| 33740 |
+
"learning_rate": 2.708313303221377e-07,
|
| 33741 |
+
"loss": 0.9034,
|
| 33742 |
+
"step": 4819
|
| 33743 |
+
},
|
| 33744 |
+
{
|
| 33745 |
+
"epoch": 0.93,
|
| 33746 |
+
"grad_norm": 1.7666061320096818,
|
| 33747 |
+
"learning_rate": 2.693914999674818e-07,
|
| 33748 |
+
"loss": 0.8755,
|
| 33749 |
+
"step": 4820
|
| 33750 |
+
},
|
| 33751 |
+
{
|
| 33752 |
+
"epoch": 0.93,
|
| 33753 |
+
"grad_norm": 1.5613714635005946,
|
| 33754 |
+
"learning_rate": 2.6795545487381724e-07,
|
| 33755 |
+
"loss": 0.7883,
|
| 33756 |
+
"step": 4821
|
| 33757 |
+
},
|
| 33758 |
+
{
|
| 33759 |
+
"epoch": 0.93,
|
| 33760 |
+
"grad_norm": 1.5200286900067441,
|
| 33761 |
+
"learning_rate": 2.665231955997738e-07,
|
| 33762 |
+
"loss": 0.8187,
|
| 33763 |
+
"step": 4822
|
| 33764 |
+
},
|
| 33765 |
+
{
|
| 33766 |
+
"epoch": 0.93,
|
| 33767 |
+
"grad_norm": 1.588152135460304,
|
| 33768 |
+
"learning_rate": 2.650947227025069e-07,
|
| 33769 |
+
"loss": 0.814,
|
| 33770 |
+
"step": 4823
|
| 33771 |
+
},
|
| 33772 |
+
{
|
| 33773 |
+
"epoch": 0.93,
|
| 33774 |
+
"grad_norm": 1.5266137304631693,
|
| 33775 |
+
"learning_rate": 2.6367003673770207e-07,
|
| 33776 |
+
"loss": 0.7984,
|
| 33777 |
+
"step": 4824
|
| 33778 |
+
},
|
| 33779 |
+
{
|
| 33780 |
+
"epoch": 0.93,
|
| 33781 |
+
"grad_norm": 1.5381375133627766,
|
| 33782 |
+
"learning_rate": 2.6224913825956933e-07,
|
| 33783 |
+
"loss": 0.868,
|
| 33784 |
+
"step": 4825
|
| 33785 |
+
},
|
| 33786 |
+
{
|
| 33787 |
+
"epoch": 0.93,
|
| 33788 |
+
"grad_norm": 1.649187576200236,
|
| 33789 |
+
"learning_rate": 2.608320278208465e-07,
|
| 33790 |
+
"loss": 0.8201,
|
| 33791 |
+
"step": 4826
|
| 33792 |
+
},
|
| 33793 |
+
{
|
| 33794 |
+
"epoch": 0.93,
|
| 33795 |
+
"grad_norm": 1.8076495675440554,
|
| 33796 |
+
"learning_rate": 2.5941870597279705e-07,
|
| 33797 |
+
"loss": 0.8381,
|
| 33798 |
+
"step": 4827
|
| 33799 |
+
},
|
| 33800 |
+
{
|
| 33801 |
+
"epoch": 0.93,
|
| 33802 |
+
"grad_norm": 1.7049459231631452,
|
| 33803 |
+
"learning_rate": 2.5800917326521013e-07,
|
| 33804 |
+
"loss": 0.8039,
|
| 33805 |
+
"step": 4828
|
| 33806 |
+
},
|
| 33807 |
+
{
|
| 33808 |
+
"epoch": 0.93,
|
| 33809 |
+
"grad_norm": 0.9538485271628983,
|
| 33810 |
+
"learning_rate": 2.566034302464027e-07,
|
| 33811 |
+
"loss": 0.7581,
|
| 33812 |
+
"step": 4829
|
| 33813 |
+
},
|
| 33814 |
+
{
|
| 33815 |
+
"epoch": 0.93,
|
| 33816 |
+
"grad_norm": 1.6724880354656986,
|
| 33817 |
+
"learning_rate": 2.552014774632172e-07,
|
| 33818 |
+
"loss": 0.8479,
|
| 33819 |
+
"step": 4830
|
| 33820 |
+
},
|
| 33821 |
+
{
|
| 33822 |
+
"epoch": 0.93,
|
| 33823 |
+
"grad_norm": 1.8133591830513744,
|
| 33824 |
+
"learning_rate": 2.538033154610209e-07,
|
| 33825 |
+
"loss": 0.8705,
|
| 33826 |
+
"step": 4831
|
| 33827 |
+
},
|
| 33828 |
+
{
|
| 33829 |
+
"epoch": 0.93,
|
| 33830 |
+
"grad_norm": 1.6579651660565642,
|
| 33831 |
+
"learning_rate": 2.524089447837064e-07,
|
| 33832 |
+
"loss": 0.8498,
|
| 33833 |
+
"step": 4832
|
| 33834 |
+
},
|
| 33835 |
+
{
|
| 33836 |
+
"epoch": 0.93,
|
| 33837 |
+
"grad_norm": 1.6548338743351285,
|
| 33838 |
+
"learning_rate": 2.51018365973692e-07,
|
| 33839 |
+
"loss": 0.8748,
|
| 33840 |
+
"step": 4833
|
| 33841 |
+
},
|
| 33842 |
+
{
|
| 33843 |
+
"epoch": 0.93,
|
| 33844 |
+
"grad_norm": 1.525397157490326,
|
| 33845 |
+
"learning_rate": 2.496315795719195e-07,
|
| 33846 |
+
"loss": 0.8361,
|
| 33847 |
+
"step": 4834
|
| 33848 |
+
},
|
| 33849 |
+
{
|
| 33850 |
+
"epoch": 0.93,
|
| 33851 |
+
"grad_norm": 1.5920176646750865,
|
| 33852 |
+
"learning_rate": 2.482485861178563e-07,
|
| 33853 |
+
"loss": 0.8105,
|
| 33854 |
+
"step": 4835
|
| 33855 |
+
},
|
| 33856 |
+
{
|
| 33857 |
+
"epoch": 0.93,
|
| 33858 |
+
"grad_norm": 1.6429465465349558,
|
| 33859 |
+
"learning_rate": 2.4686938614949643e-07,
|
| 33860 |
+
"loss": 0.8905,
|
| 33861 |
+
"step": 4836
|
| 33862 |
+
},
|
| 33863 |
+
{
|
| 33864 |
+
"epoch": 0.93,
|
| 33865 |
+
"grad_norm": 1.7832098610576956,
|
| 33866 |
+
"learning_rate": 2.4549398020335625e-07,
|
| 33867 |
+
"loss": 0.8103,
|
| 33868 |
+
"step": 4837
|
| 33869 |
+
},
|
| 33870 |
+
{
|
| 33871 |
+
"epoch": 0.93,
|
| 33872 |
+
"grad_norm": 1.5345252846573398,
|
| 33873 |
+
"learning_rate": 2.441223688144745e-07,
|
| 33874 |
+
"loss": 0.8195,
|
| 33875 |
+
"step": 4838
|
| 33876 |
+
},
|
| 33877 |
+
{
|
| 33878 |
+
"epoch": 0.93,
|
| 33879 |
+
"grad_norm": 1.5290031237515256,
|
| 33880 |
+
"learning_rate": 2.4275455251641653e-07,
|
| 33881 |
+
"loss": 0.8786,
|
| 33882 |
+
"step": 4839
|
| 33883 |
+
},
|
| 33884 |
+
{
|
| 33885 |
+
"epoch": 0.93,
|
| 33886 |
+
"grad_norm": 1.58164443431506,
|
| 33887 |
+
"learning_rate": 2.4139053184127237e-07,
|
| 33888 |
+
"loss": 0.827,
|
| 33889 |
+
"step": 4840
|
| 33890 |
+
},
|
| 33891 |
+
{
|
| 33892 |
+
"epoch": 0.93,
|
| 33893 |
+
"grad_norm": 1.610597724461763,
|
| 33894 |
+
"learning_rate": 2.400303073196508e-07,
|
| 33895 |
+
"loss": 0.8535,
|
| 33896 |
+
"step": 4841
|
| 33897 |
+
},
|
| 33898 |
+
{
|
| 33899 |
+
"epoch": 0.93,
|
| 33900 |
+
"grad_norm": 1.3834549233361646,
|
| 33901 |
+
"learning_rate": 2.3867387948068865e-07,
|
| 33902 |
+
"loss": 0.799,
|
| 33903 |
+
"step": 4842
|
| 33904 |
+
},
|
| 33905 |
+
{
|
| 33906 |
+
"epoch": 0.93,
|
| 33907 |
+
"grad_norm": 1.8398115746259007,
|
| 33908 |
+
"learning_rate": 2.3732124885204266e-07,
|
| 33909 |
+
"loss": 0.8797,
|
| 33910 |
+
"step": 4843
|
| 33911 |
+
},
|
| 33912 |
+
{
|
| 33913 |
+
"epoch": 0.93,
|
| 33914 |
+
"grad_norm": 1.5963950703347418,
|
| 33915 |
+
"learning_rate": 2.3597241595989417e-07,
|
| 33916 |
+
"loss": 0.7603,
|
| 33917 |
+
"step": 4844
|
| 33918 |
+
},
|
| 33919 |
+
{
|
| 33920 |
+
"epoch": 0.93,
|
| 33921 |
+
"grad_norm": 1.5744137147113135,
|
| 33922 |
+
"learning_rate": 2.346273813289468e-07,
|
| 33923 |
+
"loss": 0.8245,
|
| 33924 |
+
"step": 4845
|
| 33925 |
+
},
|
| 33926 |
+
{
|
| 33927 |
+
"epoch": 0.93,
|
| 33928 |
+
"grad_norm": 1.4577636380990016,
|
| 33929 |
+
"learning_rate": 2.3328614548242646e-07,
|
| 33930 |
+
"loss": 0.8599,
|
| 33931 |
+
"step": 4846
|
| 33932 |
+
},
|
| 33933 |
+
{
|
| 33934 |
+
"epoch": 0.93,
|
| 33935 |
+
"grad_norm": 1.6929866898934611,
|
| 33936 |
+
"learning_rate": 2.3194870894208244e-07,
|
| 33937 |
+
"loss": 0.8307,
|
| 33938 |
+
"step": 4847
|
| 33939 |
+
},
|
| 33940 |
+
{
|
| 33941 |
+
"epoch": 0.93,
|
| 33942 |
+
"grad_norm": 1.5446532896341592,
|
| 33943 |
+
"learning_rate": 2.3061507222818303e-07,
|
| 33944 |
+
"loss": 0.8035,
|
| 33945 |
+
"step": 4848
|
| 33946 |
+
},
|
| 33947 |
+
{
|
| 33948 |
+
"epoch": 0.93,
|
| 33949 |
+
"grad_norm": 1.6311185078352026,
|
| 33950 |
+
"learning_rate": 2.2928523585952323e-07,
|
| 33951 |
+
"loss": 0.8205,
|
| 33952 |
+
"step": 4849
|
| 33953 |
+
},
|
| 33954 |
+
{
|
| 33955 |
+
"epoch": 0.93,
|
| 33956 |
+
"grad_norm": 1.5943013003594422,
|
| 33957 |
+
"learning_rate": 2.2795920035341258e-07,
|
| 33958 |
+
"loss": 0.9206,
|
| 33959 |
+
"step": 4850
|
| 33960 |
+
},
|
| 33961 |
+
{
|
| 33962 |
+
"epoch": 0.93,
|
| 33963 |
+
"grad_norm": 1.6240632118828648,
|
| 33964 |
+
"learning_rate": 2.2663696622568955e-07,
|
| 33965 |
+
"loss": 0.877,
|
| 33966 |
+
"step": 4851
|
| 33967 |
+
},
|
| 33968 |
+
{
|
| 33969 |
+
"epoch": 0.93,
|
| 33970 |
+
"grad_norm": 1.4939145654179917,
|
| 33971 |
+
"learning_rate": 2.2531853399070936e-07,
|
| 33972 |
+
"loss": 0.8568,
|
| 33973 |
+
"step": 4852
|
| 33974 |
+
},
|
| 33975 |
+
{
|
| 33976 |
+
"epoch": 0.93,
|
| 33977 |
+
"grad_norm": 1.5981329832684763,
|
| 33978 |
+
"learning_rate": 2.2400390416134953e-07,
|
| 33979 |
+
"loss": 0.8412,
|
| 33980 |
+
"step": 4853
|
| 33981 |
+
},
|
| 33982 |
+
{
|
| 33983 |
+
"epoch": 0.93,
|
| 33984 |
+
"grad_norm": 1.570133236687216,
|
| 33985 |
+
"learning_rate": 2.226930772490088e-07,
|
| 33986 |
+
"loss": 0.845,
|
| 33987 |
+
"step": 4854
|
| 33988 |
+
},
|
| 33989 |
+
{
|
| 33990 |
+
"epoch": 0.93,
|
| 33991 |
+
"grad_norm": 1.743169482849413,
|
| 33992 |
+
"learning_rate": 2.213860537636059e-07,
|
| 33993 |
+
"loss": 0.9209,
|
| 33994 |
+
"step": 4855
|
| 33995 |
+
},
|
| 33996 |
+
{
|
| 33997 |
+
"epoch": 0.94,
|
| 33998 |
+
"grad_norm": 1.6958034405374705,
|
| 33999 |
+
"learning_rate": 2.200828342135819e-07,
|
| 34000 |
+
"loss": 0.8887,
|
| 34001 |
+
"step": 4856
|
| 34002 |
+
},
|
| 34003 |
+
{
|
| 34004 |
+
"epoch": 0.94,
|
| 34005 |
+
"grad_norm": 1.5922805104822988,
|
| 34006 |
+
"learning_rate": 2.187834191058935e-07,
|
| 34007 |
+
"loss": 0.8502,
|
| 34008 |
+
"step": 4857
|
| 34009 |
+
},
|
| 34010 |
+
{
|
| 34011 |
+
"epoch": 0.94,
|
| 34012 |
+
"grad_norm": 1.518497807992221,
|
| 34013 |
+
"learning_rate": 2.1748780894602194e-07,
|
| 34014 |
+
"loss": 0.8574,
|
| 34015 |
+
"step": 4858
|
| 34016 |
+
},
|
| 34017 |
+
{
|
| 34018 |
+
"epoch": 0.94,
|
| 34019 |
+
"grad_norm": 1.686973118522028,
|
| 34020 |
+
"learning_rate": 2.1619600423796628e-07,
|
| 34021 |
+
"loss": 0.8445,
|
| 34022 |
+
"step": 4859
|
| 34023 |
+
},
|
| 34024 |
+
{
|
| 34025 |
+
"epoch": 0.94,
|
| 34026 |
+
"grad_norm": 1.6772664760048672,
|
| 34027 |
+
"learning_rate": 2.1490800548424672e-07,
|
| 34028 |
+
"loss": 0.8219,
|
| 34029 |
+
"step": 4860
|
| 34030 |
+
},
|
| 34031 |
+
{
|
| 34032 |
+
"epoch": 0.94,
|
| 34033 |
+
"grad_norm": 1.5617253901249581,
|
| 34034 |
+
"learning_rate": 2.136238131859014e-07,
|
| 34035 |
+
"loss": 0.8314,
|
| 34036 |
+
"step": 4861
|
| 34037 |
+
},
|
| 34038 |
+
{
|
| 34039 |
+
"epoch": 0.94,
|
| 34040 |
+
"grad_norm": 1.6075469863562961,
|
| 34041 |
+
"learning_rate": 2.123434278424885e-07,
|
| 34042 |
+
"loss": 0.8681,
|
| 34043 |
+
"step": 4862
|
| 34044 |
+
},
|
| 34045 |
+
{
|
| 34046 |
+
"epoch": 0.94,
|
| 34047 |
+
"grad_norm": 1.789235541183758,
|
| 34048 |
+
"learning_rate": 2.1106684995208626e-07,
|
| 34049 |
+
"loss": 0.8466,
|
| 34050 |
+
"step": 4863
|
| 34051 |
+
},
|
| 34052 |
+
{
|
| 34053 |
+
"epoch": 0.94,
|
| 34054 |
+
"grad_norm": 1.6447974400912926,
|
| 34055 |
+
"learning_rate": 2.0979408001128743e-07,
|
| 34056 |
+
"loss": 0.8522,
|
| 34057 |
+
"step": 4864
|
| 34058 |
+
},
|
| 34059 |
+
{
|
| 34060 |
+
"epoch": 0.94,
|
| 34061 |
+
"grad_norm": 1.765136819247867,
|
| 34062 |
+
"learning_rate": 2.085251185152093e-07,
|
| 34063 |
+
"loss": 0.8604,
|
| 34064 |
+
"step": 4865
|
| 34065 |
+
},
|
| 34066 |
+
{
|
| 34067 |
+
"epoch": 0.94,
|
| 34068 |
+
"grad_norm": 1.6266851116541106,
|
| 34069 |
+
"learning_rate": 2.0725996595748366e-07,
|
| 34070 |
+
"loss": 0.8664,
|
| 34071 |
+
"step": 4866
|
| 34072 |
+
},
|
| 34073 |
+
{
|
| 34074 |
+
"epoch": 0.94,
|
| 34075 |
+
"grad_norm": 1.67819672473352,
|
| 34076 |
+
"learning_rate": 2.059986228302624e-07,
|
| 34077 |
+
"loss": 0.7761,
|
| 34078 |
+
"step": 4867
|
| 34079 |
+
},
|
| 34080 |
+
{
|
| 34081 |
+
"epoch": 0.94,
|
| 34082 |
+
"grad_norm": 1.65969174055999,
|
| 34083 |
+
"learning_rate": 2.0474108962421524e-07,
|
| 34084 |
+
"loss": 0.8423,
|
| 34085 |
+
"step": 4868
|
| 34086 |
+
},
|
| 34087 |
+
{
|
| 34088 |
+
"epoch": 0.94,
|
| 34089 |
+
"grad_norm": 1.5079462034803506,
|
| 34090 |
+
"learning_rate": 2.0348736682852864e-07,
|
| 34091 |
+
"loss": 0.7754,
|
| 34092 |
+
"step": 4869
|
| 34093 |
+
},
|
| 34094 |
+
{
|
| 34095 |
+
"epoch": 0.94,
|
| 34096 |
+
"grad_norm": 1.6052162598092945,
|
| 34097 |
+
"learning_rate": 2.022374549309103e-07,
|
| 34098 |
+
"loss": 0.8417,
|
| 34099 |
+
"step": 4870
|
| 34100 |
+
},
|
| 34101 |
+
{
|
| 34102 |
+
"epoch": 0.94,
|
| 34103 |
+
"grad_norm": 1.7703126547521626,
|
| 34104 |
+
"learning_rate": 2.00991354417579e-07,
|
| 34105 |
+
"loss": 0.9103,
|
| 34106 |
+
"step": 4871
|
| 34107 |
+
},
|
| 34108 |
+
{
|
| 34109 |
+
"epoch": 0.94,
|
| 34110 |
+
"grad_norm": 1.7526946610459049,
|
| 34111 |
+
"learning_rate": 1.9974906577327813e-07,
|
| 34112 |
+
"loss": 0.8527,
|
| 34113 |
+
"step": 4872
|
| 34114 |
+
},
|
| 34115 |
+
{
|
| 34116 |
+
"epoch": 0.94,
|
| 34117 |
+
"grad_norm": 1.9231864358048003,
|
| 34118 |
+
"learning_rate": 1.9851058948126333e-07,
|
| 34119 |
+
"loss": 0.8165,
|
| 34120 |
+
"step": 4873
|
| 34121 |
+
},
|
| 34122 |
+
{
|
| 34123 |
+
"epoch": 0.94,
|
| 34124 |
+
"grad_norm": 1.6876394188237012,
|
| 34125 |
+
"learning_rate": 1.9727592602330926e-07,
|
| 34126 |
+
"loss": 0.8579,
|
| 34127 |
+
"step": 4874
|
| 34128 |
+
},
|
| 34129 |
+
{
|
| 34130 |
+
"epoch": 0.94,
|
| 34131 |
+
"grad_norm": 1.546043471808283,
|
| 34132 |
+
"learning_rate": 1.9604507587970612e-07,
|
| 34133 |
+
"loss": 0.7516,
|
| 34134 |
+
"step": 4875
|
| 34135 |
+
},
|
| 34136 |
+
{
|
| 34137 |
+
"epoch": 0.94,
|
| 34138 |
+
"grad_norm": 1.8330467033982552,
|
| 34139 |
+
"learning_rate": 1.9481803952926314e-07,
|
| 34140 |
+
"loss": 0.8899,
|
| 34141 |
+
"step": 4876
|
| 34142 |
+
},
|
| 34143 |
+
{
|
| 34144 |
+
"epoch": 0.94,
|
| 34145 |
+
"grad_norm": 1.6924749568421145,
|
| 34146 |
+
"learning_rate": 1.9359481744930297e-07,
|
| 34147 |
+
"loss": 0.831,
|
| 34148 |
+
"step": 4877
|
| 34149 |
+
},
|
| 34150 |
+
{
|
| 34151 |
+
"epoch": 0.94,
|
| 34152 |
+
"grad_norm": 1.7045527025533789,
|
| 34153 |
+
"learning_rate": 1.9237541011566606e-07,
|
| 34154 |
+
"loss": 0.8909,
|
| 34155 |
+
"step": 4878
|
| 34156 |
+
},
|
| 34157 |
+
{
|
| 34158 |
+
"epoch": 0.94,
|
| 34159 |
+
"grad_norm": 1.6507040101580346,
|
| 34160 |
+
"learning_rate": 1.9115981800270855e-07,
|
| 34161 |
+
"loss": 0.8436,
|
| 34162 |
+
"step": 4879
|
| 34163 |
+
},
|
| 34164 |
+
{
|
| 34165 |
+
"epoch": 0.94,
|
| 34166 |
+
"grad_norm": 1.510467164138072,
|
| 34167 |
+
"learning_rate": 1.8994804158330327e-07,
|
| 34168 |
+
"loss": 0.7436,
|
| 34169 |
+
"step": 4880
|
| 34170 |
+
},
|
| 34171 |
+
{
|
| 34172 |
+
"epoch": 0.94,
|
| 34173 |
+
"grad_norm": 1.4970798059476809,
|
| 34174 |
+
"learning_rate": 1.8874008132883646e-07,
|
| 34175 |
+
"loss": 0.8396,
|
| 34176 |
+
"step": 4881
|
| 34177 |
+
},
|
| 34178 |
+
{
|
| 34179 |
+
"epoch": 0.94,
|
| 34180 |
+
"grad_norm": 1.7338623941791689,
|
| 34181 |
+
"learning_rate": 1.875359377092134e-07,
|
| 34182 |
+
"loss": 0.8812,
|
| 34183 |
+
"step": 4882
|
| 34184 |
+
},
|
| 34185 |
+
{
|
| 34186 |
+
"epoch": 0.94,
|
| 34187 |
+
"grad_norm": 1.6861050115544627,
|
| 34188 |
+
"learning_rate": 1.863356111928516e-07,
|
| 34189 |
+
"loss": 0.8854,
|
| 34190 |
+
"step": 4883
|
| 34191 |
+
},
|
| 34192 |
+
{
|
| 34193 |
+
"epoch": 0.94,
|
| 34194 |
+
"grad_norm": 1.526204807597603,
|
| 34195 |
+
"learning_rate": 1.851391022466853e-07,
|
| 34196 |
+
"loss": 0.858,
|
| 34197 |
+
"step": 4884
|
| 34198 |
+
},
|
| 34199 |
+
{
|
| 34200 |
+
"epoch": 0.94,
|
| 34201 |
+
"grad_norm": 1.6976895766365234,
|
| 34202 |
+
"learning_rate": 1.839464113361611e-07,
|
| 34203 |
+
"loss": 0.7955,
|
| 34204 |
+
"step": 4885
|
| 34205 |
+
},
|
| 34206 |
+
{
|
| 34207 |
+
"epoch": 0.94,
|
| 34208 |
+
"grad_norm": 1.6559912626039093,
|
| 34209 |
+
"learning_rate": 1.827575389252456e-07,
|
| 34210 |
+
"loss": 0.8226,
|
| 34211 |
+
"step": 4886
|
| 34212 |
+
},
|
| 34213 |
+
{
|
| 34214 |
+
"epoch": 0.94,
|
| 34215 |
+
"grad_norm": 1.622119688685812,
|
| 34216 |
+
"learning_rate": 1.815724854764145e-07,
|
| 34217 |
+
"loss": 0.8671,
|
| 34218 |
+
"step": 4887
|
| 34219 |
+
},
|
| 34220 |
+
{
|
| 34221 |
+
"epoch": 0.94,
|
| 34222 |
+
"grad_norm": 1.5437991175471397,
|
| 34223 |
+
"learning_rate": 1.8039125145066115e-07,
|
| 34224 |
+
"loss": 0.8028,
|
| 34225 |
+
"step": 4888
|
| 34226 |
+
},
|
| 34227 |
+
{
|
| 34228 |
+
"epoch": 0.94,
|
| 34229 |
+
"grad_norm": 1.723305371104265,
|
| 34230 |
+
"learning_rate": 1.7921383730749143e-07,
|
| 34231 |
+
"loss": 0.8234,
|
| 34232 |
+
"step": 4889
|
| 34233 |
+
},
|
| 34234 |
+
{
|
| 34235 |
+
"epoch": 0.94,
|
| 34236 |
+
"grad_norm": 1.6710362217257508,
|
| 34237 |
+
"learning_rate": 1.7804024350492778e-07,
|
| 34238 |
+
"loss": 0.8893,
|
| 34239 |
+
"step": 4890
|
| 34240 |
+
},
|
| 34241 |
+
{
|
| 34242 |
+
"epoch": 0.94,
|
| 34243 |
+
"grad_norm": 1.6106969878587862,
|
| 34244 |
+
"learning_rate": 1.768704704995028e-07,
|
| 34245 |
+
"loss": 0.8756,
|
| 34246 |
+
"step": 4891
|
| 34247 |
+
},
|
| 34248 |
+
{
|
| 34249 |
+
"epoch": 0.94,
|
| 34250 |
+
"grad_norm": 1.6065165768298206,
|
| 34251 |
+
"learning_rate": 1.7570451874626583e-07,
|
| 34252 |
+
"loss": 0.7772,
|
| 34253 |
+
"step": 4892
|
| 34254 |
+
},
|
| 34255 |
+
{
|
| 34256 |
+
"epoch": 0.94,
|
| 34257 |
+
"grad_norm": 1.629084990697285,
|
| 34258 |
+
"learning_rate": 1.745423886987785e-07,
|
| 34259 |
+
"loss": 0.8493,
|
| 34260 |
+
"step": 4893
|
| 34261 |
+
},
|
| 34262 |
+
{
|
| 34263 |
+
"epoch": 0.94,
|
| 34264 |
+
"grad_norm": 1.7264309930868569,
|
| 34265 |
+
"learning_rate": 1.7338408080911473e-07,
|
| 34266 |
+
"loss": 0.8663,
|
| 34267 |
+
"step": 4894
|
| 34268 |
+
},
|
| 34269 |
+
{
|
| 34270 |
+
"epoch": 0.94,
|
| 34271 |
+
"grad_norm": 1.7192802669576013,
|
| 34272 |
+
"learning_rate": 1.7222959552786412e-07,
|
| 34273 |
+
"loss": 0.8826,
|
| 34274 |
+
"step": 4895
|
| 34275 |
+
},
|
| 34276 |
+
{
|
| 34277 |
+
"epoch": 0.94,
|
| 34278 |
+
"grad_norm": 1.5667756101128578,
|
| 34279 |
+
"learning_rate": 1.7107893330412738e-07,
|
| 34280 |
+
"loss": 0.7617,
|
| 34281 |
+
"step": 4896
|
| 34282 |
+
},
|
| 34283 |
+
{
|
| 34284 |
+
"epoch": 0.94,
|
| 34285 |
+
"grad_norm": 1.6970738721179803,
|
| 34286 |
+
"learning_rate": 1.6993209458551763e-07,
|
| 34287 |
+
"loss": 0.9408,
|
| 34288 |
+
"step": 4897
|
| 34289 |
+
},
|
| 34290 |
+
{
|
| 34291 |
+
"epoch": 0.94,
|
| 34292 |
+
"grad_norm": 1.5125985162571438,
|
| 34293 |
+
"learning_rate": 1.687890798181635e-07,
|
| 34294 |
+
"loss": 0.8234,
|
| 34295 |
+
"step": 4898
|
| 34296 |
+
},
|
| 34297 |
+
{
|
| 34298 |
+
"epoch": 0.94,
|
| 34299 |
+
"grad_norm": 1.5611681895794751,
|
| 34300 |
+
"learning_rate": 1.6764988944670158e-07,
|
| 34301 |
+
"loss": 0.8612,
|
| 34302 |
+
"step": 4899
|
| 34303 |
+
},
|
| 34304 |
+
{
|
| 34305 |
+
"epoch": 0.94,
|
| 34306 |
+
"grad_norm": 1.725702892372231,
|
| 34307 |
+
"learning_rate": 1.6651452391428736e-07,
|
| 34308 |
+
"loss": 0.9418,
|
| 34309 |
+
"step": 4900
|
| 34310 |
+
},
|
| 34311 |
+
{
|
| 34312 |
+
"epoch": 0.94,
|
| 34313 |
+
"grad_norm": 1.6391013469838338,
|
| 34314 |
+
"learning_rate": 1.6538298366257975e-07,
|
| 34315 |
+
"loss": 0.7918,
|
| 34316 |
+
"step": 4901
|
| 34317 |
+
},
|
| 34318 |
+
{
|
| 34319 |
+
"epoch": 0.94,
|
| 34320 |
+
"grad_norm": 1.661758037273611,
|
| 34321 |
+
"learning_rate": 1.6425526913175672e-07,
|
| 34322 |
+
"loss": 0.8399,
|
| 34323 |
+
"step": 4902
|
| 34324 |
+
},
|
| 34325 |
+
{
|
| 34326 |
+
"epoch": 0.94,
|
| 34327 |
+
"grad_norm": 1.5954579453399507,
|
| 34328 |
+
"learning_rate": 1.6313138076050505e-07,
|
| 34329 |
+
"loss": 0.8521,
|
| 34330 |
+
"step": 4903
|
| 34331 |
+
},
|
| 34332 |
+
{
|
| 34333 |
+
"epoch": 0.94,
|
| 34334 |
+
"grad_norm": 1.602569704425456,
|
| 34335 |
+
"learning_rate": 1.6201131898602284e-07,
|
| 34336 |
+
"loss": 0.8762,
|
| 34337 |
+
"step": 4904
|
| 34338 |
+
},
|
| 34339 |
+
{
|
| 34340 |
+
"epoch": 0.94,
|
| 34341 |
+
"grad_norm": 1.6726770139591156,
|
| 34342 |
+
"learning_rate": 1.6089508424402156e-07,
|
| 34343 |
+
"loss": 0.8862,
|
| 34344 |
+
"step": 4905
|
| 34345 |
+
},
|
| 34346 |
+
{
|
| 34347 |
+
"epoch": 0.94,
|
| 34348 |
+
"grad_norm": 1.720165892486497,
|
| 34349 |
+
"learning_rate": 1.5978267696872274e-07,
|
| 34350 |
+
"loss": 0.8524,
|
| 34351 |
+
"step": 4906
|
| 34352 |
+
},
|
| 34353 |
+
{
|
| 34354 |
+
"epoch": 0.94,
|
| 34355 |
+
"grad_norm": 1.5200991395166976,
|
| 34356 |
+
"learning_rate": 1.5867409759285802e-07,
|
| 34357 |
+
"loss": 0.8561,
|
| 34358 |
+
"step": 4907
|
| 34359 |
+
},
|
| 34360 |
+
{
|
| 34361 |
+
"epoch": 0.95,
|
| 34362 |
+
"grad_norm": 1.6590015797171926,
|
| 34363 |
+
"learning_rate": 1.5756934654767243e-07,
|
| 34364 |
+
"loss": 0.8831,
|
| 34365 |
+
"step": 4908
|
| 34366 |
+
},
|
| 34367 |
+
{
|
| 34368 |
+
"epoch": 0.95,
|
| 34369 |
+
"grad_norm": 1.556760071346181,
|
| 34370 |
+
"learning_rate": 1.5646842426291775e-07,
|
| 34371 |
+
"loss": 0.8126,
|
| 34372 |
+
"step": 4909
|
| 34373 |
+
},
|
| 34374 |
+
{
|
| 34375 |
+
"epoch": 0.95,
|
| 34376 |
+
"grad_norm": 1.6475807742135788,
|
| 34377 |
+
"learning_rate": 1.5537133116686142e-07,
|
| 34378 |
+
"loss": 0.7223,
|
| 34379 |
+
"step": 4910
|
| 34380 |
+
},
|
| 34381 |
+
{
|
| 34382 |
+
"epoch": 0.95,
|
| 34383 |
+
"grad_norm": 1.6864491977631575,
|
| 34384 |
+
"learning_rate": 1.542780676862776e-07,
|
| 34385 |
+
"loss": 0.8824,
|
| 34386 |
+
"step": 4911
|
| 34387 |
+
},
|
| 34388 |
+
{
|
| 34389 |
+
"epoch": 0.95,
|
| 34390 |
+
"grad_norm": 1.4541071162068364,
|
| 34391 |
+
"learning_rate": 1.531886342464517e-07,
|
| 34392 |
+
"loss": 0.7289,
|
| 34393 |
+
"step": 4912
|
| 34394 |
+
},
|
| 34395 |
+
{
|
| 34396 |
+
"epoch": 0.95,
|
| 34397 |
+
"grad_norm": 1.6553024079033145,
|
| 34398 |
+
"learning_rate": 1.521030312711802e-07,
|
| 34399 |
+
"loss": 0.8028,
|
| 34400 |
+
"step": 4913
|
| 34401 |
+
},
|
| 34402 |
+
{
|
| 34403 |
+
"epoch": 0.95,
|
| 34404 |
+
"grad_norm": 1.5381855897941494,
|
| 34405 |
+
"learning_rate": 1.5102125918276978e-07,
|
| 34406 |
+
"loss": 0.8452,
|
| 34407 |
+
"step": 4914
|
| 34408 |
+
},
|
| 34409 |
+
{
|
| 34410 |
+
"epoch": 0.95,
|
| 34411 |
+
"grad_norm": 1.81594511213681,
|
| 34412 |
+
"learning_rate": 1.499433184020327e-07,
|
| 34413 |
+
"loss": 0.9022,
|
| 34414 |
+
"step": 4915
|
| 34415 |
+
},
|
| 34416 |
+
{
|
| 34417 |
+
"epoch": 0.95,
|
| 34418 |
+
"grad_norm": 1.6259255253716556,
|
| 34419 |
+
"learning_rate": 1.4886920934829574e-07,
|
| 34420 |
+
"loss": 0.8288,
|
| 34421 |
+
"step": 4916
|
| 34422 |
+
},
|
| 34423 |
+
{
|
| 34424 |
+
"epoch": 0.95,
|
| 34425 |
+
"grad_norm": 1.6591298059831057,
|
| 34426 |
+
"learning_rate": 1.4779893243939358e-07,
|
| 34427 |
+
"loss": 0.8641,
|
| 34428 |
+
"step": 4917
|
| 34429 |
+
},
|
| 34430 |
+
{
|
| 34431 |
+
"epoch": 0.95,
|
| 34432 |
+
"grad_norm": 1.6529473509522685,
|
| 34433 |
+
"learning_rate": 1.4673248809166984e-07,
|
| 34434 |
+
"loss": 0.7588,
|
| 34435 |
+
"step": 4918
|
| 34436 |
+
},
|
| 34437 |
+
{
|
| 34438 |
+
"epoch": 0.95,
|
| 34439 |
+
"grad_norm": 1.6004992764307049,
|
| 34440 |
+
"learning_rate": 1.45669876719976e-07,
|
| 34441 |
+
"loss": 0.803,
|
| 34442 |
+
"step": 4919
|
| 34443 |
+
},
|
| 34444 |
+
{
|
| 34445 |
+
"epoch": 0.95,
|
| 34446 |
+
"grad_norm": 1.6114035564591866,
|
| 34447 |
+
"learning_rate": 1.4461109873767587e-07,
|
| 34448 |
+
"loss": 0.8716,
|
| 34449 |
+
"step": 4920
|
| 34450 |
+
},
|
| 34451 |
+
{
|
| 34452 |
+
"epoch": 0.95,
|
| 34453 |
+
"grad_norm": 1.6912543405825373,
|
| 34454 |
+
"learning_rate": 1.4355615455663884e-07,
|
| 34455 |
+
"loss": 0.7688,
|
| 34456 |
+
"step": 4921
|
| 34457 |
+
},
|
| 34458 |
+
{
|
| 34459 |
+
"epoch": 0.95,
|
| 34460 |
+
"grad_norm": 0.9525310032028005,
|
| 34461 |
+
"learning_rate": 1.4250504458724336e-07,
|
| 34462 |
+
"loss": 0.8375,
|
| 34463 |
+
"step": 4922
|
| 34464 |
+
},
|
| 34465 |
+
{
|
| 34466 |
+
"epoch": 0.95,
|
| 34467 |
+
"grad_norm": 1.8051419365888983,
|
| 34468 |
+
"learning_rate": 1.4145776923837895e-07,
|
| 34469 |
+
"loss": 0.9364,
|
| 34470 |
+
"step": 4923
|
| 34471 |
+
},
|
| 34472 |
+
{
|
| 34473 |
+
"epoch": 0.95,
|
| 34474 |
+
"grad_norm": 1.7084739920204828,
|
| 34475 |
+
"learning_rate": 1.4041432891743978e-07,
|
| 34476 |
+
"loss": 0.9089,
|
| 34477 |
+
"step": 4924
|
| 34478 |
+
},
|
| 34479 |
+
{
|
| 34480 |
+
"epoch": 0.95,
|
| 34481 |
+
"grad_norm": 1.6253733176066043,
|
| 34482 |
+
"learning_rate": 1.3937472403033002e-07,
|
| 34483 |
+
"loss": 0.8488,
|
| 34484 |
+
"step": 4925
|
| 34485 |
+
},
|
| 34486 |
+
{
|
| 34487 |
+
"epoch": 0.95,
|
| 34488 |
+
"grad_norm": 1.7310556112020248,
|
| 34489 |
+
"learning_rate": 1.3833895498146287e-07,
|
| 34490 |
+
"loss": 0.9087,
|
| 34491 |
+
"step": 4926
|
| 34492 |
+
},
|
| 34493 |
+
{
|
| 34494 |
+
"epoch": 0.95,
|
| 34495 |
+
"grad_norm": 1.6576546514999582,
|
| 34496 |
+
"learning_rate": 1.3730702217375825e-07,
|
| 34497 |
+
"loss": 0.8847,
|
| 34498 |
+
"step": 4927
|
| 34499 |
+
},
|
| 34500 |
+
{
|
| 34501 |
+
"epoch": 0.95,
|
| 34502 |
+
"grad_norm": 1.6200150482231441,
|
| 34503 |
+
"learning_rate": 1.362789260086439e-07,
|
| 34504 |
+
"loss": 0.7445,
|
| 34505 |
+
"step": 4928
|
| 34506 |
+
},
|
| 34507 |
+
{
|
| 34508 |
+
"epoch": 0.95,
|
| 34509 |
+
"grad_norm": 1.5773998945496541,
|
| 34510 |
+
"learning_rate": 1.3525466688605328e-07,
|
| 34511 |
+
"loss": 0.7787,
|
| 34512 |
+
"step": 4929
|
| 34513 |
+
},
|
| 34514 |
+
{
|
| 34515 |
+
"epoch": 0.95,
|
| 34516 |
+
"grad_norm": 1.629333696277938,
|
| 34517 |
+
"learning_rate": 1.3423424520443095e-07,
|
| 34518 |
+
"loss": 0.9095,
|
| 34519 |
+
"step": 4930
|
| 34520 |
+
},
|
| 34521 |
+
{
|
| 34522 |
+
"epoch": 0.95,
|
| 34523 |
+
"grad_norm": 1.6469354057372534,
|
| 34524 |
+
"learning_rate": 1.33217661360725e-07,
|
| 34525 |
+
"loss": 0.8278,
|
| 34526 |
+
"step": 4931
|
| 34527 |
+
},
|
| 34528 |
+
{
|
| 34529 |
+
"epoch": 0.95,
|
| 34530 |
+
"grad_norm": 1.5712895075456599,
|
| 34531 |
+
"learning_rate": 1.3220491575039352e-07,
|
| 34532 |
+
"loss": 0.8249,
|
| 34533 |
+
"step": 4932
|
| 34534 |
+
},
|
| 34535 |
+
{
|
| 34536 |
+
"epoch": 0.95,
|
| 34537 |
+
"grad_norm": 1.6775311526534245,
|
| 34538 |
+
"learning_rate": 1.3119600876739913e-07,
|
| 34539 |
+
"loss": 0.8519,
|
| 34540 |
+
"step": 4933
|
| 34541 |
+
},
|
| 34542 |
+
{
|
| 34543 |
+
"epoch": 0.95,
|
| 34544 |
+
"grad_norm": 1.7074824689906087,
|
| 34545 |
+
"learning_rate": 1.3019094080421236e-07,
|
| 34546 |
+
"loss": 0.8262,
|
| 34547 |
+
"step": 4934
|
| 34548 |
+
},
|
| 34549 |
+
{
|
| 34550 |
+
"epoch": 0.95,
|
| 34551 |
+
"grad_norm": 1.5978239435800343,
|
| 34552 |
+
"learning_rate": 1.2918971225181265e-07,
|
| 34553 |
+
"loss": 0.8392,
|
| 34554 |
+
"step": 4935
|
| 34555 |
+
},
|
| 34556 |
+
{
|
| 34557 |
+
"epoch": 0.95,
|
| 34558 |
+
"grad_norm": 1.0103492039735638,
|
| 34559 |
+
"learning_rate": 1.2819232349967958e-07,
|
| 34560 |
+
"loss": 0.8086,
|
| 34561 |
+
"step": 4936
|
| 34562 |
+
},
|
| 34563 |
+
{
|
| 34564 |
+
"epoch": 0.95,
|
| 34565 |
+
"grad_norm": 1.5515066332509084,
|
| 34566 |
+
"learning_rate": 1.2719877493580613e-07,
|
| 34567 |
+
"loss": 0.8259,
|
| 34568 |
+
"step": 4937
|
| 34569 |
+
},
|
| 34570 |
+
{
|
| 34571 |
+
"epoch": 0.95,
|
| 34572 |
+
"grad_norm": 0.9456664714615861,
|
| 34573 |
+
"learning_rate": 1.2620906694668533e-07,
|
| 34574 |
+
"loss": 0.8409,
|
| 34575 |
+
"step": 4938
|
| 34576 |
+
},
|
| 34577 |
+
{
|
| 34578 |
+
"epoch": 0.95,
|
| 34579 |
+
"grad_norm": 1.716993131432421,
|
| 34580 |
+
"learning_rate": 1.2522319991732256e-07,
|
| 34581 |
+
"loss": 0.8065,
|
| 34582 |
+
"step": 4939
|
| 34583 |
+
},
|
| 34584 |
+
{
|
| 34585 |
+
"epoch": 0.95,
|
| 34586 |
+
"grad_norm": 1.519584786287795,
|
| 34587 |
+
"learning_rate": 1.242411742312233e-07,
|
| 34588 |
+
"loss": 0.7987,
|
| 34589 |
+
"step": 4940
|
| 34590 |
+
},
|
| 34591 |
+
{
|
| 34592 |
+
"epoch": 0.95,
|
| 34593 |
+
"grad_norm": 1.814988108224232,
|
| 34594 |
+
"learning_rate": 1.232629902704008e-07,
|
| 34595 |
+
"loss": 0.7686,
|
| 34596 |
+
"step": 4941
|
| 34597 |
+
},
|
| 34598 |
+
{
|
| 34599 |
+
"epoch": 0.95,
|
| 34600 |
+
"grad_norm": 1.587242203219095,
|
| 34601 |
+
"learning_rate": 1.2228864841537512e-07,
|
| 34602 |
+
"loss": 0.8539,
|
| 34603 |
+
"step": 4942
|
| 34604 |
+
},
|
| 34605 |
+
{
|
| 34606 |
+
"epoch": 0.95,
|
| 34607 |
+
"grad_norm": 1.5219598591187173,
|
| 34608 |
+
"learning_rate": 1.2131814904517093e-07,
|
| 34609 |
+
"loss": 0.8002,
|
| 34610 |
+
"step": 4943
|
| 34611 |
+
},
|
| 34612 |
+
{
|
| 34613 |
+
"epoch": 0.95,
|
| 34614 |
+
"grad_norm": 1.7067301828543635,
|
| 34615 |
+
"learning_rate": 1.2035149253731615e-07,
|
| 34616 |
+
"loss": 0.8167,
|
| 34617 |
+
"step": 4944
|
| 34618 |
+
},
|
| 34619 |
+
{
|
| 34620 |
+
"epoch": 0.95,
|
| 34621 |
+
"grad_norm": 1.4780687729376636,
|
| 34622 |
+
"learning_rate": 1.1938867926784782e-07,
|
| 34623 |
+
"loss": 0.8022,
|
| 34624 |
+
"step": 4945
|
| 34625 |
+
},
|
| 34626 |
+
{
|
| 34627 |
+
"epoch": 0.95,
|
| 34628 |
+
"grad_norm": 1.8245418722982043,
|
| 34629 |
+
"learning_rate": 1.1842970961130406e-07,
|
| 34630 |
+
"loss": 0.9316,
|
| 34631 |
+
"step": 4946
|
| 34632 |
+
},
|
| 34633 |
+
{
|
| 34634 |
+
"epoch": 0.95,
|
| 34635 |
+
"grad_norm": 1.8012418885011572,
|
| 34636 |
+
"learning_rate": 1.1747458394072986e-07,
|
| 34637 |
+
"loss": 0.8605,
|
| 34638 |
+
"step": 4947
|
| 34639 |
+
},
|
| 34640 |
+
{
|
| 34641 |
+
"epoch": 0.95,
|
| 34642 |
+
"grad_norm": 1.6095664936474214,
|
| 34643 |
+
"learning_rate": 1.1652330262767353e-07,
|
| 34644 |
+
"loss": 0.8166,
|
| 34645 |
+
"step": 4948
|
| 34646 |
+
},
|
| 34647 |
+
{
|
| 34648 |
+
"epoch": 0.95,
|
| 34649 |
+
"grad_norm": 1.7146979682040404,
|
| 34650 |
+
"learning_rate": 1.1557586604219129e-07,
|
| 34651 |
+
"loss": 0.8968,
|
| 34652 |
+
"step": 4949
|
| 34653 |
+
},
|
| 34654 |
+
{
|
| 34655 |
+
"epoch": 0.95,
|
| 34656 |
+
"grad_norm": 1.5895056850751075,
|
| 34657 |
+
"learning_rate": 1.1463227455283943e-07,
|
| 34658 |
+
"loss": 0.8926,
|
| 34659 |
+
"step": 4950
|
| 34660 |
+
},
|
| 34661 |
+
{
|
| 34662 |
+
"epoch": 0.95,
|
| 34663 |
+
"grad_norm": 1.5541333797103272,
|
| 34664 |
+
"learning_rate": 1.13692528526681e-07,
|
| 34665 |
+
"loss": 0.7857,
|
| 34666 |
+
"step": 4951
|
| 34667 |
+
},
|
| 34668 |
+
{
|
| 34669 |
+
"epoch": 0.95,
|
| 34670 |
+
"grad_norm": 1.5896127496722006,
|
| 34671 |
+
"learning_rate": 1.1275662832928135e-07,
|
| 34672 |
+
"loss": 0.8841,
|
| 34673 |
+
"step": 4952
|
| 34674 |
+
},
|
| 34675 |
+
{
|
| 34676 |
+
"epoch": 0.95,
|
| 34677 |
+
"grad_norm": 1.4811800456998214,
|
| 34678 |
+
"learning_rate": 1.118245743247115e-07,
|
| 34679 |
+
"loss": 0.7616,
|
| 34680 |
+
"step": 4953
|
| 34681 |
+
},
|
| 34682 |
+
{
|
| 34683 |
+
"epoch": 0.95,
|
| 34684 |
+
"grad_norm": 1.4505754590481381,
|
| 34685 |
+
"learning_rate": 1.1089636687554583e-07,
|
| 34686 |
+
"loss": 0.837,
|
| 34687 |
+
"step": 4954
|
| 34688 |
+
},
|
| 34689 |
+
{
|
| 34690 |
+
"epoch": 0.95,
|
| 34691 |
+
"grad_norm": 1.5661718657308172,
|
| 34692 |
+
"learning_rate": 1.099720063428622e-07,
|
| 34693 |
+
"loss": 0.8273,
|
| 34694 |
+
"step": 4955
|
| 34695 |
+
},
|
| 34696 |
+
{
|
| 34697 |
+
"epoch": 0.95,
|
| 34698 |
+
"grad_norm": 0.950799488866897,
|
| 34699 |
+
"learning_rate": 1.0905149308624185e-07,
|
| 34700 |
+
"loss": 0.7992,
|
| 34701 |
+
"step": 4956
|
| 34702 |
+
},
|
| 34703 |
+
{
|
| 34704 |
+
"epoch": 0.95,
|
| 34705 |
+
"grad_norm": 1.6311388423421525,
|
| 34706 |
+
"learning_rate": 1.0813482746377058e-07,
|
| 34707 |
+
"loss": 0.8573,
|
| 34708 |
+
"step": 4957
|
| 34709 |
+
},
|
| 34710 |
+
{
|
| 34711 |
+
"epoch": 0.95,
|
| 34712 |
+
"grad_norm": 1.5762042142140522,
|
| 34713 |
+
"learning_rate": 1.0722200983203423e-07,
|
| 34714 |
+
"loss": 0.7506,
|
| 34715 |
+
"step": 4958
|
| 34716 |
+
},
|
| 34717 |
+
{
|
| 34718 |
+
"epoch": 0.95,
|
| 34719 |
+
"grad_norm": 1.7399002723266332,
|
| 34720 |
+
"learning_rate": 1.0631304054612545e-07,
|
| 34721 |
+
"loss": 0.8777,
|
| 34722 |
+
"step": 4959
|
| 34723 |
+
},
|
| 34724 |
+
{
|
| 34725 |
+
"epoch": 0.96,
|
| 34726 |
+
"grad_norm": 1.64896309366328,
|
| 34727 |
+
"learning_rate": 1.0540791995963917e-07,
|
| 34728 |
+
"loss": 0.8091,
|
| 34729 |
+
"step": 4960
|
| 34730 |
+
},
|
| 34731 |
+
{
|
| 34732 |
+
"epoch": 0.96,
|
| 34733 |
+
"grad_norm": 1.6609498470045294,
|
| 34734 |
+
"learning_rate": 1.045066484246704e-07,
|
| 34735 |
+
"loss": 0.7811,
|
| 34736 |
+
"step": 4961
|
| 34737 |
+
},
|
| 34738 |
+
{
|
| 34739 |
+
"epoch": 0.96,
|
| 34740 |
+
"grad_norm": 1.0063699978672147,
|
| 34741 |
+
"learning_rate": 1.036092262918198e-07,
|
| 34742 |
+
"loss": 0.8015,
|
| 34743 |
+
"step": 4962
|
| 34744 |
+
},
|
| 34745 |
+
{
|
| 34746 |
+
"epoch": 0.96,
|
| 34747 |
+
"grad_norm": 1.6084066802033907,
|
| 34748 |
+
"learning_rate": 1.0271565391018922e-07,
|
| 34749 |
+
"loss": 0.8932,
|
| 34750 |
+
"step": 4963
|
| 34751 |
+
},
|
| 34752 |
+
{
|
| 34753 |
+
"epoch": 0.96,
|
| 34754 |
+
"grad_norm": 1.6165487237067675,
|
| 34755 |
+
"learning_rate": 1.0182593162738508e-07,
|
| 34756 |
+
"loss": 0.8962,
|
| 34757 |
+
"step": 4964
|
| 34758 |
+
},
|
| 34759 |
+
{
|
| 34760 |
+
"epoch": 0.96,
|
| 34761 |
+
"grad_norm": 1.6049658161860338,
|
| 34762 |
+
"learning_rate": 1.009400597895116e-07,
|
| 34763 |
+
"loss": 0.8494,
|
| 34764 |
+
"step": 4965
|
| 34765 |
+
},
|
| 34766 |
+
{
|
| 34767 |
+
"epoch": 0.96,
|
| 34768 |
+
"grad_norm": 1.6258440115836938,
|
| 34769 |
+
"learning_rate": 1.0005803874117981e-07,
|
| 34770 |
+
"loss": 0.8588,
|
| 34771 |
+
"step": 4966
|
| 34772 |
+
},
|
| 34773 |
+
{
|
| 34774 |
+
"epoch": 0.96,
|
| 34775 |
+
"grad_norm": 1.6540136388403845,
|
| 34776 |
+
"learning_rate": 9.917986882549968e-08,
|
| 34777 |
+
"loss": 0.8433,
|
| 34778 |
+
"step": 4967
|
| 34779 |
+
},
|
| 34780 |
+
{
|
| 34781 |
+
"epoch": 0.96,
|
| 34782 |
+
"grad_norm": 1.6530462586563348,
|
| 34783 |
+
"learning_rate": 9.830555038408463e-08,
|
| 34784 |
+
"loss": 0.856,
|
| 34785 |
+
"step": 4968
|
| 34786 |
+
},
|
| 34787 |
+
{
|
| 34788 |
+
"epoch": 0.96,
|
| 34789 |
+
"grad_norm": 1.4860983635785385,
|
| 34790 |
+
"learning_rate": 9.743508375704924e-08,
|
| 34791 |
+
"loss": 0.6871,
|
| 34792 |
+
"step": 4969
|
| 34793 |
+
},
|
| 34794 |
+
{
|
| 34795 |
+
"epoch": 0.96,
|
| 34796 |
+
"grad_norm": 1.5507562812035725,
|
| 34797 |
+
"learning_rate": 9.656846928300934e-08,
|
| 34798 |
+
"loss": 0.8492,
|
| 34799 |
+
"step": 4970
|
| 34800 |
+
},
|
| 34801 |
+
{
|
| 34802 |
+
"epoch": 0.96,
|
| 34803 |
+
"grad_norm": 1.6346673988974738,
|
| 34804 |
+
"learning_rate": 9.570570729908412e-08,
|
| 34805 |
+
"loss": 0.8396,
|
| 34806 |
+
"step": 4971
|
| 34807 |
+
},
|
| 34808 |
+
{
|
| 34809 |
+
"epoch": 0.96,
|
| 34810 |
+
"grad_norm": 1.7368013133597024,
|
| 34811 |
+
"learning_rate": 9.484679814089293e-08,
|
| 34812 |
+
"loss": 0.9203,
|
| 34813 |
+
"step": 4972
|
| 34814 |
+
},
|
| 34815 |
+
{
|
| 34816 |
+
"epoch": 0.96,
|
| 34817 |
+
"grad_norm": 1.6340143749831773,
|
| 34818 |
+
"learning_rate": 9.399174214255513e-08,
|
| 34819 |
+
"loss": 0.8939,
|
| 34820 |
+
"step": 4973
|
| 34821 |
+
},
|
| 34822 |
+
{
|
| 34823 |
+
"epoch": 0.96,
|
| 34824 |
+
"grad_norm": 1.707005458041116,
|
| 34825 |
+
"learning_rate": 9.314053963669245e-08,
|
| 34826 |
+
"loss": 0.7957,
|
| 34827 |
+
"step": 4974
|
| 34828 |
+
},
|
| 34829 |
+
{
|
| 34830 |
+
"epoch": 0.96,
|
| 34831 |
+
"grad_norm": 1.6392193096320475,
|
| 34832 |
+
"learning_rate": 9.22931909544278e-08,
|
| 34833 |
+
"loss": 0.8707,
|
| 34834 |
+
"step": 4975
|
| 34835 |
+
},
|
| 34836 |
+
{
|
| 34837 |
+
"epoch": 0.96,
|
| 34838 |
+
"grad_norm": 1.587703446312483,
|
| 34839 |
+
"learning_rate": 9.144969642538415e-08,
|
| 34840 |
+
"loss": 0.768,
|
| 34841 |
+
"step": 4976
|
| 34842 |
+
},
|
| 34843 |
+
{
|
| 34844 |
+
"epoch": 0.96,
|
| 34845 |
+
"grad_norm": 1.635947063395266,
|
| 34846 |
+
"learning_rate": 9.061005637768571e-08,
|
| 34847 |
+
"loss": 0.8609,
|
| 34848 |
+
"step": 4977
|
| 34849 |
+
},
|
| 34850 |
+
{
|
| 34851 |
+
"epoch": 0.96,
|
| 34852 |
+
"grad_norm": 1.6263184836466396,
|
| 34853 |
+
"learning_rate": 8.977427113795678e-08,
|
| 34854 |
+
"loss": 0.9195,
|
| 34855 |
+
"step": 4978
|
| 34856 |
+
},
|
| 34857 |
+
{
|
| 34858 |
+
"epoch": 0.96,
|
| 34859 |
+
"grad_norm": 1.6023902021963157,
|
| 34860 |
+
"learning_rate": 8.894234103132394e-08,
|
| 34861 |
+
"loss": 0.8682,
|
| 34862 |
+
"step": 4979
|
| 34863 |
+
},
|
| 34864 |
+
{
|
| 34865 |
+
"epoch": 0.96,
|
| 34866 |
+
"grad_norm": 1.6563911158006064,
|
| 34867 |
+
"learning_rate": 8.811426638141163e-08,
|
| 34868 |
+
"loss": 0.8095,
|
| 34869 |
+
"step": 4980
|
| 34870 |
+
},
|
| 34871 |
+
{
|
| 34872 |
+
"epoch": 0.96,
|
| 34873 |
+
"grad_norm": 1.7609044106395593,
|
| 34874 |
+
"learning_rate": 8.729004751034443e-08,
|
| 34875 |
+
"loss": 0.8567,
|
| 34876 |
+
"step": 4981
|
| 34877 |
+
},
|
| 34878 |
+
{
|
| 34879 |
+
"epoch": 0.96,
|
| 34880 |
+
"grad_norm": 1.4790167691488278,
|
| 34881 |
+
"learning_rate": 8.646968473874917e-08,
|
| 34882 |
+
"loss": 0.8171,
|
| 34883 |
+
"step": 4982
|
| 34884 |
+
},
|
| 34885 |
+
{
|
| 34886 |
+
"epoch": 0.96,
|
| 34887 |
+
"grad_norm": 1.6269543335652379,
|
| 34888 |
+
"learning_rate": 8.565317838575171e-08,
|
| 34889 |
+
"loss": 0.8258,
|
| 34890 |
+
"step": 4983
|
| 34891 |
+
},
|
| 34892 |
+
{
|
| 34893 |
+
"epoch": 0.96,
|
| 34894 |
+
"grad_norm": 1.5583935374943,
|
| 34895 |
+
"learning_rate": 8.484052876897797e-08,
|
| 34896 |
+
"loss": 0.8265,
|
| 34897 |
+
"step": 4984
|
| 34898 |
+
},
|
| 34899 |
+
{
|
| 34900 |
+
"epoch": 0.96,
|
| 34901 |
+
"grad_norm": 1.4943474875070648,
|
| 34902 |
+
"learning_rate": 8.403173620455174e-08,
|
| 34903 |
+
"loss": 0.7888,
|
| 34904 |
+
"step": 4985
|
| 34905 |
+
},
|
| 34906 |
+
{
|
| 34907 |
+
"epoch": 0.96,
|
| 34908 |
+
"grad_norm": 1.594762702711398,
|
| 34909 |
+
"learning_rate": 8.322680100710023e-08,
|
| 34910 |
+
"loss": 0.8098,
|
| 34911 |
+
"step": 4986
|
| 34912 |
+
},
|
| 34913 |
+
{
|
| 34914 |
+
"epoch": 0.96,
|
| 34915 |
+
"grad_norm": 1.6022569198318086,
|
| 34916 |
+
"learning_rate": 8.24257234897452e-08,
|
| 34917 |
+
"loss": 0.7609,
|
| 34918 |
+
"step": 4987
|
| 34919 |
+
},
|
| 34920 |
+
{
|
| 34921 |
+
"epoch": 0.96,
|
| 34922 |
+
"grad_norm": 1.6467693510359207,
|
| 34923 |
+
"learning_rate": 8.162850396411182e-08,
|
| 34924 |
+
"loss": 0.8345,
|
| 34925 |
+
"step": 4988
|
| 34926 |
+
},
|
| 34927 |
+
{
|
| 34928 |
+
"epoch": 0.96,
|
| 34929 |
+
"grad_norm": 1.751101015490329,
|
| 34930 |
+
"learning_rate": 8.08351427403209e-08,
|
| 34931 |
+
"loss": 0.8488,
|
| 34932 |
+
"step": 4989
|
| 34933 |
+
},
|
| 34934 |
+
{
|
| 34935 |
+
"epoch": 0.96,
|
| 34936 |
+
"grad_norm": 1.5795700884294022,
|
| 34937 |
+
"learning_rate": 8.004564012699667e-08,
|
| 34938 |
+
"loss": 0.8266,
|
| 34939 |
+
"step": 4990
|
| 34940 |
+
},
|
| 34941 |
+
{
|
| 34942 |
+
"epoch": 0.96,
|
| 34943 |
+
"grad_norm": 1.6452428063118647,
|
| 34944 |
+
"learning_rate": 7.925999643125904e-08,
|
| 34945 |
+
"loss": 0.7832,
|
| 34946 |
+
"step": 4991
|
| 34947 |
+
},
|
| 34948 |
+
{
|
| 34949 |
+
"epoch": 0.96,
|
| 34950 |
+
"grad_norm": 1.6461661712590487,
|
| 34951 |
+
"learning_rate": 7.847821195872796e-08,
|
| 34952 |
+
"loss": 0.9808,
|
| 34953 |
+
"step": 4992
|
| 34954 |
+
},
|
| 34955 |
+
{
|
| 34956 |
+
"epoch": 0.96,
|
| 34957 |
+
"grad_norm": 1.5657255170988507,
|
| 34958 |
+
"learning_rate": 7.770028701352239e-08,
|
| 34959 |
+
"loss": 0.8519,
|
| 34960 |
+
"step": 4993
|
| 34961 |
+
},
|
| 34962 |
+
{
|
| 34963 |
+
"epoch": 0.96,
|
| 34964 |
+
"grad_norm": 1.6852755964063066,
|
| 34965 |
+
"learning_rate": 7.692622189825916e-08,
|
| 34966 |
+
"loss": 0.8425,
|
| 34967 |
+
"step": 4994
|
| 34968 |
+
},
|
| 34969 |
+
{
|
| 34970 |
+
"epoch": 0.96,
|
| 34971 |
+
"grad_norm": 1.9054468178480464,
|
| 34972 |
+
"learning_rate": 7.615601691405405e-08,
|
| 34973 |
+
"loss": 0.8855,
|
| 34974 |
+
"step": 4995
|
| 34975 |
+
},
|
| 34976 |
+
{
|
| 34977 |
+
"epoch": 0.96,
|
| 34978 |
+
"grad_norm": 1.4668506651688153,
|
| 34979 |
+
"learning_rate": 7.538967236051963e-08,
|
| 34980 |
+
"loss": 0.8569,
|
| 34981 |
+
"step": 4996
|
| 34982 |
+
},
|
| 34983 |
+
{
|
| 34984 |
+
"epoch": 0.96,
|
| 34985 |
+
"grad_norm": 1.7272847342044952,
|
| 34986 |
+
"learning_rate": 7.462718853576966e-08,
|
| 34987 |
+
"loss": 0.8052,
|
| 34988 |
+
"step": 4997
|
| 34989 |
+
},
|
| 34990 |
+
{
|
| 34991 |
+
"epoch": 0.96,
|
| 34992 |
+
"grad_norm": 1.7114468698535032,
|
| 34993 |
+
"learning_rate": 7.386856573641576e-08,
|
| 34994 |
+
"loss": 0.7993,
|
| 34995 |
+
"step": 4998
|
| 34996 |
+
},
|
| 34997 |
+
{
|
| 34998 |
+
"epoch": 0.96,
|
| 34999 |
+
"grad_norm": 1.573996192998908,
|
| 35000 |
+
"learning_rate": 7.311380425756298e-08,
|
| 35001 |
+
"loss": 0.8533,
|
| 35002 |
+
"step": 4999
|
| 35003 |
+
},
|
| 35004 |
+
{
|
| 35005 |
+
"epoch": 0.96,
|
| 35006 |
+
"grad_norm": 1.6580048196864743,
|
| 35007 |
+
"learning_rate": 7.236290439282089e-08,
|
| 35008 |
+
"loss": 0.8353,
|
| 35009 |
+
"step": 5000
|
| 35010 |
}
|
| 35011 |
],
|
| 35012 |
"logging_steps": 1.0,
|
|
|
|
| 35014 |
"num_input_tokens_seen": 0,
|
| 35015 |
"num_train_epochs": 1,
|
| 35016 |
"save_steps": 500,
|
| 35017 |
+
"total_flos": 6750470532169728.0,
|
| 35018 |
"train_batch_size": 4,
|
| 35019 |
"trial_name": null,
|
| 35020 |
"trial_params": null
|