config fixes
Browse files
examples/falcon/config-7b-lora.yml
CHANGED
|
@@ -23,7 +23,7 @@ lora_dropout: 0.0
|
|
| 23 |
lora_target_modules:
|
| 24 |
lora_target_linear: true
|
| 25 |
lora_fan_in_fan_out:
|
| 26 |
-
wandb_project:
|
| 27 |
wandb_watch:
|
| 28 |
wandb_run_id:
|
| 29 |
wandb_log_model:
|
|
|
|
| 23 |
lora_target_modules:
|
| 24 |
lora_target_linear: true
|
| 25 |
lora_fan_in_fan_out:
|
| 26 |
+
wandb_project:
|
| 27 |
wandb_watch:
|
| 28 |
wandb_run_id:
|
| 29 |
wandb_log_model:
|
examples/falcon/config-7b.yml
CHANGED
|
@@ -23,7 +23,7 @@ lora_dropout: 0.0
|
|
| 23 |
lora_target_modules:
|
| 24 |
lora_target_linear: true
|
| 25 |
lora_fan_in_fan_out:
|
| 26 |
-
wandb_project:
|
| 27 |
wandb_watch:
|
| 28 |
wandb_run_id:
|
| 29 |
wandb_log_model:
|
|
|
|
| 23 |
lora_target_modules:
|
| 24 |
lora_target_linear: true
|
| 25 |
lora_fan_in_fan_out:
|
| 26 |
+
wandb_project:
|
| 27 |
wandb_watch:
|
| 28 |
wandb_run_id:
|
| 29 |
wandb_log_model:
|
examples/openllama-3b/config.yml
CHANGED
|
@@ -25,7 +25,7 @@ wandb_project:
|
|
| 25 |
wandb_watch:
|
| 26 |
wandb_run_id:
|
| 27 |
wandb_log_model:
|
| 28 |
-
output_dir: ./
|
| 29 |
batch_size: 16
|
| 30 |
micro_batch_size: 4
|
| 31 |
num_epochs: 3
|
|
@@ -43,7 +43,7 @@ early_stopping_patience:
|
|
| 43 |
resume_from_checkpoint:
|
| 44 |
local_rank:
|
| 45 |
logging_steps: 1
|
| 46 |
-
xformers_attention:
|
| 47 |
flash_attention:
|
| 48 |
gptq_groupsize:
|
| 49 |
gptq_model_v1:
|
|
|
|
| 25 |
wandb_watch:
|
| 26 |
wandb_run_id:
|
| 27 |
wandb_log_model:
|
| 28 |
+
output_dir: ./openllama-out
|
| 29 |
batch_size: 16
|
| 30 |
micro_batch_size: 4
|
| 31 |
num_epochs: 3
|
|
|
|
| 43 |
resume_from_checkpoint:
|
| 44 |
local_rank:
|
| 45 |
logging_steps: 1
|
| 46 |
+
xformers_attention: true
|
| 47 |
flash_attention:
|
| 48 |
gptq_groupsize:
|
| 49 |
gptq_model_v1:
|
examples/openllama-3b/lora.yml
CHANGED
|
@@ -49,7 +49,7 @@ early_stopping_patience:
|
|
| 49 |
resume_from_checkpoint:
|
| 50 |
local_rank:
|
| 51 |
logging_steps: 1
|
| 52 |
-
xformers_attention:
|
| 53 |
flash_attention:
|
| 54 |
gptq_groupsize:
|
| 55 |
gptq_model_v1:
|
|
|
|
| 49 |
resume_from_checkpoint:
|
| 50 |
local_rank:
|
| 51 |
logging_steps: 1
|
| 52 |
+
xformers_attention: true
|
| 53 |
flash_attention:
|
| 54 |
gptq_groupsize:
|
| 55 |
gptq_model_v1:
|