Epoch 3. Batch Size 0. Peers 11.
Browse files- config.json +11 -11
- inner_optimizer.pt +1 -1
- model.safetensors +1 -1
- outer_optimizer.pt +1 -1
config.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
-
"_name_or_path": "
|
3 |
"activation_function": "gelu_new",
|
4 |
"all_reduce_scores": {
|
5 |
"0": "NON_PARTICIPATING",
|
6 |
-
"1": "
|
7 |
"10": "NON_PARTICIPATING",
|
8 |
"100": "NON_PARTICIPATING",
|
9 |
"101": "NON_PARTICIPATING",
|
@@ -148,7 +148,7 @@
|
|
148 |
"227": "NON_PARTICIPATING",
|
149 |
"228": "NON_PARTICIPATING",
|
150 |
"229": "NON_PARTICIPATING",
|
151 |
-
"23": "
|
152 |
"230": "NON_PARTICIPATING",
|
153 |
"231": "NON_PARTICIPATING",
|
154 |
"232": "NON_PARTICIPATING",
|
@@ -169,7 +169,7 @@
|
|
169 |
"246": "NON_PARTICIPATING",
|
170 |
"247": "NON_PARTICIPATING",
|
171 |
"248": "NON_PARTICIPATING",
|
172 |
-
"249": "
|
173 |
"25": "SUCCESS",
|
174 |
"250": "NON_PARTICIPATING",
|
175 |
"251": "NON_PARTICIPATING",
|
@@ -197,7 +197,7 @@
|
|
197 |
"41": "NON_PARTICIPATING",
|
198 |
"42": "NON_PARTICIPATING",
|
199 |
"43": "NON_PARTICIPATING",
|
200 |
-
"44": "
|
201 |
"45": "NON_PARTICIPATING",
|
202 |
"46": "NON_PARTICIPATING",
|
203 |
"47": "NON_PARTICIPATING",
|
@@ -268,11 +268,11 @@
|
|
268 |
"AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
|
269 |
},
|
270 |
"block_list": [
|
271 |
-
|
272 |
-
|
273 |
-
|
274 |
-
|
275 |
-
|
276 |
],
|
277 |
"block_size": 1024,
|
278 |
"bos_token_id": 50256,
|
@@ -281,7 +281,7 @@
|
|
281 |
"initializer_range": 0.02,
|
282 |
"inner_step": 0,
|
283 |
"inner_steps": 0,
|
284 |
-
"last_allreduce_block":
|
285 |
"layer_norm_epsilon": 1e-05,
|
286 |
"model_type": "gpt_optimized",
|
287 |
"n_embd": 1280,
|
|
|
1 |
{
|
2 |
+
"_name_or_path": "Mhair/onethreesix",
|
3 |
"activation_function": "gelu_new",
|
4 |
"all_reduce_scores": {
|
5 |
"0": "NON_PARTICIPATING",
|
6 |
+
"1": "NON_PARTICIPATING",
|
7 |
"10": "NON_PARTICIPATING",
|
8 |
"100": "NON_PARTICIPATING",
|
9 |
"101": "NON_PARTICIPATING",
|
|
|
148 |
"227": "NON_PARTICIPATING",
|
149 |
"228": "NON_PARTICIPATING",
|
150 |
"229": "NON_PARTICIPATING",
|
151 |
+
"23": "NON_PARTICIPATING",
|
152 |
"230": "NON_PARTICIPATING",
|
153 |
"231": "NON_PARTICIPATING",
|
154 |
"232": "NON_PARTICIPATING",
|
|
|
169 |
"246": "NON_PARTICIPATING",
|
170 |
"247": "NON_PARTICIPATING",
|
171 |
"248": "NON_PARTICIPATING",
|
172 |
+
"249": "NON_PARTICIPATING",
|
173 |
"25": "SUCCESS",
|
174 |
"250": "NON_PARTICIPATING",
|
175 |
"251": "NON_PARTICIPATING",
|
|
|
197 |
"41": "NON_PARTICIPATING",
|
198 |
"42": "NON_PARTICIPATING",
|
199 |
"43": "NON_PARTICIPATING",
|
200 |
+
"44": "SUCCESS",
|
201 |
"45": "NON_PARTICIPATING",
|
202 |
"46": "NON_PARTICIPATING",
|
203 |
"47": "NON_PARTICIPATING",
|
|
|
268 |
"AutoModelForCausalLM": "distributed/optimized-gpt2-500m--modeling_gpt_optimized.GPTOptim"
|
269 |
},
|
270 |
"block_list": [
|
271 |
+
5123714,
|
272 |
+
5123721,
|
273 |
+
5123731,
|
274 |
+
5123739,
|
275 |
+
5123749
|
276 |
],
|
277 |
"block_size": 1024,
|
278 |
"bos_token_id": 50256,
|
|
|
281 |
"initializer_range": 0.02,
|
282 |
"inner_step": 0,
|
283 |
"inner_steps": 0,
|
284 |
+
"last_allreduce_block": 5243395,
|
285 |
"layer_norm_epsilon": 1e-05,
|
286 |
"model_type": "gpt_optimized",
|
287 |
"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 2752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ee18aa6cae78518ec5b8b190f239543754c73c6909694a918abbb83ff2b806ba
|
3 |
size 2752
|
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:e4620474527da90e3a89002859d39c26db706434a7bd495620c40999afba7e49
|
3 |
size 4040701744
|
outer_optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4040805354
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bc76425a6693fbeffe8cfc65569a11fad46a04de50e7078eadb0dc070b1fb7b0
|
3 |
size 4040805354
|