Outer Step 4. Inner Step 6. Batch Size 64
Browse files- config.json +15 -15
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
config.json
CHANGED
|
@@ -106,7 +106,7 @@
|
|
| 106 |
"19": "NON_PARTICIPATING",
|
| 107 |
"190": "NON_PARTICIPATING",
|
| 108 |
"191": "NON_PARTICIPATING",
|
| 109 |
-
"192": "
|
| 110 |
"193": "NON_PARTICIPATING",
|
| 111 |
"194": "NON_PARTICIPATING",
|
| 112 |
"195": "NON_PARTICIPATING",
|
|
@@ -170,7 +170,7 @@
|
|
| 170 |
"247": "NON_PARTICIPATING",
|
| 171 |
"248": "NON_PARTICIPATING",
|
| 172 |
"249": "SUCCESS",
|
| 173 |
-
"25": "
|
| 174 |
"250": "NON_PARTICIPATING",
|
| 175 |
"251": "NON_PARTICIPATING",
|
| 176 |
"252": "NON_PARTICIPATING",
|
|
@@ -187,7 +187,7 @@
|
|
| 187 |
"32": "NON_PARTICIPATING",
|
| 188 |
"33": "NON_PARTICIPATING",
|
| 189 |
"34": "NON_PARTICIPATING",
|
| 190 |
-
"35": "
|
| 191 |
"36": "NON_PARTICIPATING",
|
| 192 |
"37": "NON_PARTICIPATING",
|
| 193 |
"38": "NON_PARTICIPATING",
|
|
@@ -268,25 +268,25 @@
|
|
| 268 |
"AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
|
| 269 |
},
|
| 270 |
"block_list": [
|
| 271 |
-
|
| 272 |
-
|
| 273 |
-
|
| 274 |
-
|
| 275 |
-
|
| 276 |
-
|
| 277 |
-
|
| 278 |
-
|
| 279 |
-
|
| 280 |
-
|
| 281 |
],
|
| 282 |
"block_size": 1024,
|
| 283 |
"bos_token_id": 50256,
|
| 284 |
"embd_pdrop": 0.1,
|
| 285 |
"eos_token_id": 50256,
|
| 286 |
"initializer_range": 0.02,
|
| 287 |
-
"inner_step":
|
| 288 |
"inner_steps": 0,
|
| 289 |
-
"last_allreduce_block":
|
| 290 |
"layer_norm_epsilon": 1e-05,
|
| 291 |
"model_type": "gpt_optimized",
|
| 292 |
"n_embd": 1280,
|
|
|
|
| 106 |
"19": "NON_PARTICIPATING",
|
| 107 |
"190": "NON_PARTICIPATING",
|
| 108 |
"191": "NON_PARTICIPATING",
|
| 109 |
+
"192": "SUCCESS",
|
| 110 |
"193": "NON_PARTICIPATING",
|
| 111 |
"194": "NON_PARTICIPATING",
|
| 112 |
"195": "NON_PARTICIPATING",
|
|
|
|
| 170 |
"247": "NON_PARTICIPATING",
|
| 171 |
"248": "NON_PARTICIPATING",
|
| 172 |
"249": "SUCCESS",
|
| 173 |
+
"25": "SUCCESS",
|
| 174 |
"250": "NON_PARTICIPATING",
|
| 175 |
"251": "NON_PARTICIPATING",
|
| 176 |
"252": "NON_PARTICIPATING",
|
|
|
|
| 187 |
"32": "NON_PARTICIPATING",
|
| 188 |
"33": "NON_PARTICIPATING",
|
| 189 |
"34": "NON_PARTICIPATING",
|
| 190 |
+
"35": "NON_PARTICIPATING",
|
| 191 |
"36": "NON_PARTICIPATING",
|
| 192 |
"37": "NON_PARTICIPATING",
|
| 193 |
"38": "NON_PARTICIPATING",
|
|
|
|
| 268 |
"AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
|
| 269 |
},
|
| 270 |
"block_list": [
|
| 271 |
+
5227907,
|
| 272 |
+
5227912,
|
| 273 |
+
5227916,
|
| 274 |
+
5227921,
|
| 275 |
+
5227925,
|
| 276 |
+
5227929,
|
| 277 |
+
5227934,
|
| 278 |
+
5227938,
|
| 279 |
+
5227942,
|
| 280 |
+
5227946
|
| 281 |
],
|
| 282 |
"block_size": 1024,
|
| 283 |
"bos_token_id": 50256,
|
| 284 |
"embd_pdrop": 0.1,
|
| 285 |
"eos_token_id": 50256,
|
| 286 |
"initializer_range": 0.02,
|
| 287 |
+
"inner_step": 6,
|
| 288 |
"inner_steps": 0,
|
| 289 |
+
"last_allreduce_block": 5227728,
|
| 290 |
"layer_norm_epsilon": 1e-05,
|
| 291 |
"model_type": "gpt_optimized",
|
| 292 |
"n_embd": 1280,
|
inner_optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 8081781770
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:086b0ddffbc1b294fc42d62038933696b23758036d7368d7427210407346eb36
|
| 3 |
size 8081781770
|
model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4040701744
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:638b8fa13b11faa5452cc3b61a7afda7376443c02e76fadd6232d54f0bf248e9
|
| 3 |
size 4040701744
|