ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix #944

Browse Source 
Former-commit-id: 032245647848aaa4167086636b6c985268c5fee3
This commit is contained in:
hiyouga
2023-09-21 19:51:02 +08:00
parent 95c0d9ab24
commit dc68c313ee
11 changed files with 116 additions and 101 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
src/llmtuner/webui/components/train.py
View File

@@ -35,11 +35,11 @@ def create_train_tab(top_elems: Dict[str, "Component"], runner: "Runner") -> Dic
)
with gr.Row():
max_source_length = gr.Slider(value=512, minimum=4, maximum=4096, step=1)
max_target_length = gr.Slider(value=512, minimum=4, maximum=4096, step=1)
cutoff_len = gr.Slider(value=1024, minimum=4, maximum=8192, step=1)
learning_rate = gr.Textbox(value="5e-5")
num_train_epochs = gr.Textbox(value="3.0")
max_samples = gr.Textbox(value="100000")
compute_type = gr.Radio(choices=["fp16", "bf16"], value="fp16")
with gr.Row():
batch_size = gr.Slider(value=4, minimum=1, maximum=512, step=1)
@@ -55,7 +55,8 @@ def create_train_tab(top_elems: Dict[str, "Component"], runner: "Runner") -> Dic
logging_steps = gr.Slider(value=5, minimum=5, maximum=1000, step=5)
save_steps = gr.Slider(value=100, minimum=10, maximum=5000, step=10)
warmup_steps = gr.Slider(value=0, minimum=0, maximum=5000, step=1)
compute_type = gr.Radio(choices=["fp16", "bf16"], value="fp16")
flash_attn = gr.Checkbox(value=False)
rope_scaling = gr.Checkbox(value=False)
with gr.Accordion(label="LoRA config", open=False) as lora_tab:
with gr.Row():
@@ -107,11 +108,11 @@ def create_train_tab(top_elems: Dict[str, "Component"], runner: "Runner") -> Dic
training_stage,
dataset_dir,
dataset,
max_source_length,
max_target_length,
cutoff_len,
learning_rate,
num_train_epochs,
max_samples,
compute_type,
batch_size,
gradient_accumulation_steps,
lr_scheduler_type,
@@ -120,7 +121,8 @@ def create_train_tab(top_elems: Dict[str, "Component"], runner: "Runner") -> Dic
logging_steps,
save_steps,
warmup_steps,
compute_type,
flash_attn,
rope_scaling,
lora_rank,
lora_dropout,
lora_target,
@@ -151,11 +153,11 @@ def create_train_tab(top_elems: Dict[str, "Component"], runner: "Runner") -> Dic
preview_count=preview_count,
preview_samples=preview_samples,
close_btn=close_btn,
max_source_length=max_source_length,
max_target_length=max_target_length,
cutoff_len=cutoff_len,
learning_rate=learning_rate,
num_train_epochs=num_train_epochs,
max_samples=max_samples,
compute_type=compute_type,
batch_size=batch_size,
gradient_accumulation_steps=gradient_accumulation_steps,
lr_scheduler_type=lr_scheduler_type,
@@ -165,7 +167,8 @@ def create_train_tab(top_elems: Dict[str, "Component"], runner: "Runner") -> Dic
logging_steps=logging_steps,
save_steps=save_steps,
warmup_steps=warmup_steps,
compute_type=compute_type,
flash_attn=flash_attn,
rope_scaling=rope_scaling,
lora_tab=lora_tab,
lora_rank=lora_rank,
lora_dropout=lora_dropout,