graphcore-rahult
commited on
Commit
•
b6e6c24
1
Parent(s):
db870fe
Training in progress, step 500
Browse files- config.json +1 -1
- ipu_config.json +1 -1
- pytorch_model.bin +1 -1
- training_args.bin +1 -1
config.json
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
{
|
2 |
"_name_or_path": "t5-small",
|
3 |
"architectures": [
|
4 |
-
"
|
5 |
],
|
6 |
"d_ff": 2048,
|
7 |
"d_kv": 64,
|
|
|
1 |
{
|
2 |
"_name_or_path": "t5-small",
|
3 |
"architectures": [
|
4 |
+
"PoptorchPipelinedT5ForConditionalGeneration"
|
5 |
],
|
6 |
"d_ff": 2048,
|
7 |
"d_kv": 64,
|
ipu_config.json
CHANGED
@@ -3,7 +3,7 @@
|
|
3 |
"device_iterations": 1,
|
4 |
"embedding_serialization_factor": 2,
|
5 |
"enable_half_partials": true,
|
6 |
-
"executable_cache_dir": "/tmp/exe_cache",
|
7 |
"execute_encoder_on_cpu_for_generation": false,
|
8 |
"gradient_accumulation_steps": 16,
|
9 |
"inference_device_iterations": 5,
|
|
|
3 |
"device_iterations": 1,
|
4 |
"embedding_serialization_factor": 2,
|
5 |
"enable_half_partials": true,
|
6 |
+
"executable_cache_dir": "/tmp/exe_cache/summarization",
|
7 |
"execute_encoder_on_cpu_for_generation": false,
|
8 |
"gradient_accumulation_steps": 16,
|
9 |
"inference_device_iterations": 5,
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 121055163
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:863f382c9c2e724d4fc0d8c86d61a2aa6242c9caa97d75a75efd548701fcc837
|
3 |
size 121055163
|
training_args.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2863
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:fdc004737cc12876f67d3c4fcd973cd086389ae76352752738debbe5baebd259
|
3 |
size 2863
|