set group_by_length to false in examples
Browse files- examples/cerebras/qlora.yml +1 -1
- examples/gptj/qlora.yml +1 -1
- examples/llama-2/lora.yml +1 -1
- examples/llama-2/qlora.yml +1 -1
- examples/openllama-3b/qlora.yml +1 -1
examples/cerebras/qlora.yml
CHANGED
|
@@ -35,7 +35,7 @@ torchdistx_path:
|
|
| 35 |
lr_scheduler: cosine
|
| 36 |
learning_rate: 0.0002
|
| 37 |
train_on_inputs: false
|
| 38 |
-
group_by_length:
|
| 39 |
bf16: true
|
| 40 |
fp16: false
|
| 41 |
tf32: true
|
|
|
|
| 35 |
lr_scheduler: cosine
|
| 36 |
learning_rate: 0.0002
|
| 37 |
train_on_inputs: false
|
| 38 |
+
group_by_length: false
|
| 39 |
bf16: true
|
| 40 |
fp16: false
|
| 41 |
tf32: true
|
examples/gptj/qlora.yml
CHANGED
|
@@ -32,7 +32,7 @@ torchdistx_path:
|
|
| 32 |
lr_scheduler: cosine
|
| 33 |
learning_rate: 0.0001
|
| 34 |
train_on_inputs: false
|
| 35 |
-
group_by_length:
|
| 36 |
bf16: true
|
| 37 |
fp16: false
|
| 38 |
tf32: true
|
|
|
|
| 32 |
lr_scheduler: cosine
|
| 33 |
learning_rate: 0.0001
|
| 34 |
train_on_inputs: false
|
| 35 |
+
group_by_length: false
|
| 36 |
bf16: true
|
| 37 |
fp16: false
|
| 38 |
tf32: true
|
examples/llama-2/lora.yml
CHANGED
|
@@ -38,7 +38,7 @@ lr_scheduler: cosine
|
|
| 38 |
learning_rate: 0.0002
|
| 39 |
|
| 40 |
train_on_inputs: false
|
| 41 |
-
group_by_length:
|
| 42 |
bf16: true
|
| 43 |
fp16: false
|
| 44 |
tf32: false
|
|
|
|
| 38 |
learning_rate: 0.0002
|
| 39 |
|
| 40 |
train_on_inputs: false
|
| 41 |
+
group_by_length: false
|
| 42 |
bf16: true
|
| 43 |
fp16: false
|
| 44 |
tf32: false
|
examples/llama-2/qlora.yml
CHANGED
|
@@ -39,7 +39,7 @@ lr_scheduler: cosine
|
|
| 39 |
learning_rate: 0.0002
|
| 40 |
|
| 41 |
train_on_inputs: false
|
| 42 |
-
group_by_length:
|
| 43 |
bf16: true
|
| 44 |
fp16: false
|
| 45 |
tf32: false
|
|
|
|
| 39 |
learning_rate: 0.0002
|
| 40 |
|
| 41 |
train_on_inputs: false
|
| 42 |
+
group_by_length: false
|
| 43 |
bf16: true
|
| 44 |
fp16: false
|
| 45 |
tf32: false
|
examples/openllama-3b/qlora.yml
CHANGED
|
@@ -34,7 +34,7 @@ torchdistx_path:
|
|
| 34 |
lr_scheduler: cosine
|
| 35 |
learning_rate: 0.0002
|
| 36 |
train_on_inputs: false
|
| 37 |
-
group_by_length:
|
| 38 |
bf16: true
|
| 39 |
fp16: false
|
| 40 |
tf32: true
|
|
|
|
| 34 |
lr_scheduler: cosine
|
| 35 |
learning_rate: 0.0002
|
| 36 |
train_on_inputs: false
|
| 37 |
+
group_by_length: false
|
| 38 |
bf16: true
|
| 39 |
fp16: false
|
| 40 |
tf32: true
|