Upload meta_000765.json with huggingface_hub
Browse files- meta_000765.json +6 -6
meta_000765.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"step": 765,
|
| 3 |
-
"val_bpb": 0.
|
| 4 |
"model_config": {
|
| 5 |
"sequence_len": 2048,
|
| 6 |
"vocab_size": 65536,
|
|
@@ -10,7 +10,7 @@
|
|
| 10 |
"n_embd": 1280
|
| 11 |
},
|
| 12 |
"user_config": {
|
| 13 |
-
"run": "nanochat-h100-diloco-base-mid-f",
|
| 14 |
"max_iterations": -1,
|
| 15 |
"dtype": "bfloat16",
|
| 16 |
"max_seq_len": 2048,
|
|
@@ -23,8 +23,8 @@
|
|
| 23 |
"eval_every": 150,
|
| 24 |
"eval_tokens": 10485760,
|
| 25 |
"total_batch_size": 524288,
|
| 26 |
-
"use_diloco":
|
| 27 |
-
"base_diloco":
|
| 28 |
"diloco_H": 30,
|
| 29 |
"diloco_outer_lr": 0.8,
|
| 30 |
"diloco_outer_momentum": 0.9,
|
|
@@ -32,6 +32,6 @@
|
|
| 32 |
},
|
| 33 |
"device_batch_size": 32,
|
| 34 |
"max_seq_len": 2048,
|
| 35 |
-
"use_diloco":
|
| 36 |
-
"base_diloco":
|
| 37 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"step": 765,
|
| 3 |
+
"val_bpb": 0.4384375044226997,
|
| 4 |
"model_config": {
|
| 5 |
"sequence_len": 2048,
|
| 6 |
"vocab_size": 65536,
|
|
|
|
| 10 |
"n_embd": 1280
|
| 11 |
},
|
| 12 |
"user_config": {
|
| 13 |
+
"run": "nanochat-h100-diloco-base-mid-standard-f",
|
| 14 |
"max_iterations": -1,
|
| 15 |
"dtype": "bfloat16",
|
| 16 |
"max_seq_len": 2048,
|
|
|
|
| 23 |
"eval_every": 150,
|
| 24 |
"eval_tokens": 10485760,
|
| 25 |
"total_batch_size": 524288,
|
| 26 |
+
"use_diloco": 0,
|
| 27 |
+
"base_diloco": 1,
|
| 28 |
"diloco_H": 30,
|
| 29 |
"diloco_outer_lr": 0.8,
|
| 30 |
"diloco_outer_momentum": 0.9,
|
|
|
|
| 32 |
},
|
| 33 |
"device_batch_size": 32,
|
| 34 |
"max_seq_len": 2048,
|
| 35 |
+
"use_diloco": 0,
|
| 36 |
+
"base_diloco": 1
|
| 37 |
}
|