improve web ui

src/llmtuner/extras/misc.py

@@ -79,6 +79,10 @@ def get_current_device() -> torch.device:
     return torch.device(device)
 
 
+def get_device_count() -> int:
+    return torch.cuda.device_count()
+
+
 def get_logits_processor() -> "LogitsProcessorList":
     r"""
     Gets logits processor that removes NaN and Inf logits.

src/llmtuner/webui/components/train.py

@@ -37,7 +37,7 @@ def create_train_tab(engine: "Engine") -> Dict[str, "Component"]:
         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")
+        compute_type = gr.Radio(choices=["fp16", "bf16", "fp32"], value="fp16")
 
     input_elems.update({cutoff_len, learning_rate, num_train_epochs, max_samples, compute_type})
     elem_dict.update(dict(
@@ -68,13 +68,13 @@ def create_train_tab(engine: "Engine") -> Dict[str, "Component"]:
             neftune_alpha = gr.Slider(value=0, minimum=0, maximum=10, step=0.1)
 
             with gr.Column():
-                train_on_prompt = gr.Checkbox(value=False)
+                sft_packing = gr.Checkbox(value=False)
                 upcast_layernorm = gr.Checkbox(value=False)
 
-    input_elems.update({logging_steps, save_steps, warmup_steps, neftune_alpha, train_on_prompt, upcast_layernorm})
+    input_elems.update({logging_steps, save_steps, warmup_steps, neftune_alpha, sft_packing, upcast_layernorm})
     elem_dict.update(dict(
         extra_tab=extra_tab, logging_steps=logging_steps, save_steps=save_steps, warmup_steps=warmup_steps,
-        neftune_alpha=neftune_alpha, train_on_prompt=train_on_prompt, upcast_layernorm=upcast_layernorm
+        neftune_alpha=neftune_alpha, sft_packing=sft_packing, upcast_layernorm=upcast_layernorm
     ))
 
     with gr.Accordion(label="LoRA config", open=False) as lora_tab:

src/llmtuner/webui/locales.py

@@ -325,14 +325,14 @@ LOCALES = {
             "info": "嵌入向量所添加的噪声大小。"
         }
     },
-    "train_on_prompt": {
+    "sft_packing": {
         "en": {
-            "label": "Train on prompt",
-            "info": "Compute loss on the prompt tokens in supervised fine-tuning."
+            "label": "Pack sequences",
+            "info": "Pack sequences into samples of fixed length in supervised fine-tuning."
         },
         "zh": {
-            "label": "计算输入损失",
-            "info": "在监督微调时候计算输入序列的损失。"
+            "label": "序列打包",
+            "info": "在有监督微调阶段将序列打包为相同长度的样本。"
         }
     },
     "upcast_layernorm": {
@@ -342,7 +342,7 @@ LOCALES = {
         },
         "zh": {
             "label": "缩放归一化层",
-            "info": "将归一化层权重缩放至 32 位浮点数。"
+            "info": "将归一化层权重缩放至 32 位精度。"
         }
     },
     "lora_tab": {
@@ -665,6 +665,10 @@ ALERTS = {
         "en": "Training is unavailable in demo mode, duplicate the space to a private one first.",
         "zh": "展示模式不支持训练，请先复制到私人空间。"
     },
+    "err_device_count": {
+        "en": "Multiple GPUs are not supported yet.",
+        "zh": "尚不支持多 GPU 训练。"
+    },
     "info_aborting": {
         "en": "Aborted, wait for terminating...",
         "zh": "训练中断，正在等待线程结束……"

src/llmtuner/webui/runner.py

@@ -12,7 +12,7 @@ from transformers.trainer import TRAINING_ARGS_NAME
 from llmtuner.extras.callbacks import LogCallback
 from llmtuner.extras.constants import TRAINING_STAGES
 from llmtuner.extras.logging import LoggerHandler
-from llmtuner.extras.misc import torch_gc
+from llmtuner.extras.misc import get_device_count, torch_gc
 from llmtuner.train import run_exp
 from llmtuner.webui.common import get_module, get_save_dir, load_config
 from llmtuner.webui.locales import ALERTS
@@ -67,6 +67,9 @@ class Runner:
         if self.demo_mode and (not from_preview):
             return ALERTS["err_demo"][lang]
 
+        if not from_preview and get_device_count() > 1:
+            return ALERTS["err_device_count"][lang]
+
         self.aborted = False
         self.logger_handler.reset()
         self.trainer_callback = LogCallback(self)
@@ -119,16 +122,17 @@ class Runner:
             save_steps=get("train.save_steps"),
             warmup_steps=get("train.warmup_steps"),
             neftune_noise_alpha=get("train.neftune_alpha") or None,
-            train_on_prompt=get("train.train_on_prompt"),
+            sft_packing=get("train.sft_packing"),
             upcast_layernorm=get("train.upcast_layernorm"),
             lora_rank=get("train.lora_rank"),
             lora_dropout=get("train.lora_dropout"),
             lora_target=get("train.lora_target") or get_module(get("top.model_name")),
             additional_target=get("train.additional_target") or None,
             create_new_adapter=get("train.create_new_adapter"),
-            output_dir=get_save_dir(get("top.model_name"), get("top.finetuning_type"), get("train.output_dir"))
+            output_dir=get_save_dir(get("top.model_name"), get("top.finetuning_type"), get("train.output_dir")),
+            fp16=(get("train.compute_type") == "fp16"),
+            bf16=(get("train.compute_type") == "bf16")
         )
-        args[get("train.compute_type")] = True
         args["disable_tqdm"] = True
 
         if TRAINING_STAGES[get("train.training_stage")] in ["rm", "ppo", "dpo"]: