Add finetuned model on duplicate dataset
Browse files
config.json
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
{
|
|
|
|
| 2 |
"activation_function": "gelu_new",
|
| 3 |
"architectures": [
|
| 4 |
"GPTNeoForCausalLM"
|
|
@@ -45,8 +46,9 @@
|
|
| 45 |
"summary_proj_to_labels": true,
|
| 46 |
"summary_type": "cls_index",
|
| 47 |
"summary_use_proj": true,
|
|
|
|
| 48 |
"transformers_version": "4.9.0.dev0",
|
| 49 |
"use_cache": true,
|
| 50 |
-
"vocab_size":
|
| 51 |
"window_size": 256
|
| 52 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "EleutherAI/gpt-neo-125M",
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPTNeoForCausalLM"
|
|
|
|
| 46 |
"summary_proj_to_labels": true,
|
| 47 |
"summary_type": "cls_index",
|
| 48 |
"summary_use_proj": true,
|
| 49 |
+
"torch_dtype": "float32",
|
| 50 |
"transformers_version": "4.9.0.dev0",
|
| 51 |
"use_cache": true,
|
| 52 |
+
"vocab_size": 50261,
|
| 53 |
"window_size": 256
|
| 54 |
}
|
events.out.tfevents.1625416555.t1v-n-416475e4-w-0.250164.3.v2 → events.out.tfevents.1626277192.t1v-n-19ae6aab-w-0.394803.0.v2
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8f66884ba4af20d3c7e3f06cc3abc61b30a9cd9cd6f7680c45da66f169c3322f
|
| 3 |
+
size 20033824
|
flax_model.msgpack
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:34c099110802292b88e007fa9b34a86f5686e4498dd1f4a75500c4e38d5fa8c4
|
| 3 |
+
size 500812408
|
pytorch_model.bin
DELETED
|
@@ -1,3 +0,0 @@
|
|
| 1 |
-
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:0a2b0c731ccffb1c2cd8d94e4f39b52fb2778be6c75840ee5de7655c262e4edb
|
| 3 |
-
size 526014813
|
|
|
|
|
|
|
|
|
|
|
|