delete use_half

FlagEmbedding/baai_general_embedding/README.md

@@ -37,6 +37,7 @@ torchrun --nproc_per_node {number of gpus} \
 --output_dir {path to save model} \
 --model_name_or_path {base model} \
 --train_data {path to train data} \
+--per_device_train_batch_size {batch size} \
 --learning_rate 2e-5 \
 --num_train_epochs 5 \
 --max_seq_length 512
@@ -66,7 +67,7 @@ torchrun --nproc_per_node {number of gpus} \
 --train_data toy_finetune_data.jsonl \
 --learning_rate 1e-5 \
 --num_train_epochs 5 \
---per_device_train_batch_size {large batch size} \
+--per_device_train_batch_size {batch size} \
 --dataloader_drop_last True \
 --normlized True \
 --temperature 0.01 \

FlagEmbedding/baai_general_embedding/flag_models.py

@@ -11,8 +11,7 @@ class FlagModel:
             model_name_or_path: str = None,
             pooling_method: str = 'cls',
             normalize_embeddings: bool = True,
-            query_instruction_for_retrieval: str = None,
-            use_half: bool = True,
+            query_instruction_for_retrieval: str = None
     ) -> None:
 
         self.tokenizer = AutoTokenizer.from_pretrained(model_name_or_path)
@@ -21,8 +20,6 @@ class FlagModel:
         self.normalize_embeddings = normalize_embeddings
         self.pooling_method = pooling_method
 
-        if use_half: self.model.half()
-
         self.device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
         self.model = self.model.to(self.device)
 

examples/finetune/README.md

@@ -41,7 +41,7 @@ torchrun --nproc_per_node {number of gpus} \
 --train_data toy_finetune_data.jsonl \
 --learning_rate 1e-5 \
 --num_train_epochs 5 \
---per_device_train_batch_size {large batch size} \
+--per_device_train_batch_size {large batch size; set 1 for toy data} \
 --dataloader_drop_last True \
 --normlized True \
 --temperature 0.01 \

examples/pretrain/README.md

@@ -37,7 +37,7 @@ torchrun --nproc_per_node {number of gpus} \
 --train_data toy_pretrain_data.jsonl \
 --learning_rate 2e-5 \
 --num_train_epochs 2 \
---per_device_train_batch_size {batch size} \
+--per_device_train_batch_size {batch size; set 1 for toy data} \
 --dataloader_drop_last True \
 --max_seq_length 512 \
 --logging_steps 10

setup.py

@@ -5,7 +5,7 @@ with open("README.md", mode="r", encoding="utf-8") as readme_file:
 
 setup(
     name='FlagEmbedding',
-    version='1.0.3',
+    version='1.0.4',
     description='FlagEmbedding',
     long_description=readme,
     long_description_content_type="text/markdown",