govisi commited on
Commit
deff2db
1 Parent(s): 7893f5c
Files changed (1) hide show
  1. app.py +3 -3
app.py CHANGED
@@ -13,10 +13,10 @@ bot = commands.Bot("", intents=discord.Intents.all())
13
 
14
  # model_pipe = pickle.load(open('pipe.pkl', 'rb'))
15
  # this is the dtype used for trivial operations, such as vector->vector operations and is the dtype that will determine the accuracy of the model
16
- runtimedtype = torch.float16 # torch.float64, torch.bfloat16
17
 
18
  # this is the dtype used for matrix-vector operations, and is the dtype that will determine the performance and memory usage of the model
19
- dtype = torch.bfloat16 # torch.float32, torch.float64, torch.bfloat16
20
 
21
  useGPU = torch.cuda.is_available() # False
22
 
@@ -25,7 +25,7 @@ useGPU = torch.cuda.is_available() # False
25
  async def on_ready():
26
  print(f'We have logged in as {bot.user}')
27
  global model
28
- model = RWKV("RWKV-4-Pile-3B-Instruct-test2-20230209.pth",
29
  mode=TORCH,
30
  useGPU=useGPU,
31
  runtimedtype=runtimedtype,
 
13
 
14
  # model_pipe = pickle.load(open('pipe.pkl', 'rb'))
15
  # this is the dtype used for trivial operations, such as vector->vector operations and is the dtype that will determine the accuracy of the model
16
+ runtimedtype = torch.float32 # torch.float64, torch.bfloat16
17
 
18
  # this is the dtype used for matrix-vector operations, and is the dtype that will determine the performance and memory usage of the model
19
+ dtype = torch.float32 # torch.float32, torch.float64, torch.bfloat16
20
 
21
  useGPU = torch.cuda.is_available() # False
22
 
 
25
  async def on_ready():
26
  print(f'We have logged in as {bot.user}')
27
  global model
28
+ model = RWKV("https://huggingface.co/BlinkDL/rwkv-4-pile-1b5/resolve/main/RWKV-4-Pile-1B5-Instruct-test2-20230209.pth",
29
  mode=TORCH,
30
  useGPU=useGPU,
31
  runtimedtype=runtimedtype,