Clean up imports

conversion/adaptivegptq.py

@@ -2,8 +2,6 @@ import torch
 from torch import nn
 import torch.nn.functional as F
 import math
-import exllamav2.ext
-from exllamav2 import ext
 from exllamav2.ext import exllamav2_ext as ext_c, none_tensor
 
 

conversion/quantize.py

@@ -5,7 +5,7 @@ from conversion.qparams import QParams, qparams_options, qparams_headoptions
 from conversion.adaptivegptq import AdaptiveGPTQ
 import torch
 from torch import nn
-import os, sys, time, math, json
+import os, time, math, json
 import torch.nn.functional as F
 import gc
 

examples/streaming.py

@@ -2,7 +2,7 @@
 import sys, os
 sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
 
-from exllamav2 import(
+from exllamav2 import (
     ExLlamaV2,
     ExLlamaV2Config,
     ExLlamaV2Cache,

exllamav2/attn.py

@@ -8,8 +8,6 @@ from exllamav2.cache import ExLlamaV2Cache
 import math
 from exllamav2 import ext
 from exllamav2.ext import exllamav2_ext as ext_c, none_tensor
-import gc
-import sys
 # from flash_attn import flash_attn_func
 # import xformers.ops as xops
 # from exllamav2.util import list_live_tensors, set_snapshot, diff_snapshot, print_vram_usage_peak

exllamav2/config.py

@@ -1,6 +1,6 @@
 import torch
 from safetensors import safe_open
-import sys, os, glob, json
+import os, glob, json
 
 class ExLlamaV2Config:
 

exllamav2/linear.py

@@ -1,7 +1,6 @@
 import torch
 from exllamav2.module import ExLlamaV2Module
 from torch import nn
-import math
 from exllamav2 import ext
 from exllamav2.ext import exllamav2_ext as ext_c, none_tensor
 from safetensors import safe_open

exllamav2/mlp.py

@@ -1,10 +1,8 @@
 import torch
-from torch import nn
 import torch.nn.functional as F
 from exllamav2.module import ExLlamaV2Module
 from exllamav2.rmsnorm import ExLlamaV2RMSNorm
 from exllamav2.linear import ExLlamaV2Linear
-from exllamav2 import ext
 from exllamav2.ext import exllamav2_ext as ext_c, none_tensor
 
 class ExLlamaV2MLP(ExLlamaV2Module):

exllamav2/model.py

@@ -6,9 +6,6 @@ if sys.version_info < min_version:
     print("")
 
 import torch
-from torch import nn
-import torch.nn.functional as F
-from safetensors import safe_open
 import math
 from exllamav2.config import ExLlamaV2Config
 from exllamav2.cache import ExLlamaV2Cache

exllamav2/model_init.py

@@ -4,7 +4,6 @@ import argparse, sys, os, glob
 from exllamav2 import(
     ExLlamaV2,
     ExLlamaV2Config,
-    ExLlamaV2Cache,
     ExLlamaV2Tokenizer
 )
 

exllamav2/rmsnorm.py

@@ -1,7 +1,6 @@
 import torch
 from torch import nn
 from exllamav2.module import ExLlamaV2Module
-from exllamav2 import ext
 from exllamav2.ext import exllamav2_ext as ext_c, none_tensor
 
 class ExLlamaV2RMSNorm(ExLlamaV2Module):