xzuyn commited on
Commit
d044f3a
1 Parent(s): b67636b

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +3 -1
app.py CHANGED
@@ -8,6 +8,7 @@ llama_tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/llama-token
8
  falcon_tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-7b")
9
  phi2_tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-2")
10
  t5_tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-xxl")
 
11
 
12
  def tokenize(input_text):
13
  gpt2_tokens = gpt2_tokenizer(input_text, add_special_tokens=True)["input_ids"]
@@ -16,8 +17,9 @@ def tokenize(input_text):
16
  falcon_tokens = falcon_tokenizer(input_text, add_special_tokens=True)["input_ids"]
17
  phi2_tokens = phi2_tokenizer(input_text, add_special_tokens=True)["input_ids"]
18
  t5_tokens = t5_tokenizer(input_text, add_special_tokens=True)["input_ids"]
 
19
 
20
- return f"GPT-2/GPT-J: {len(gpt2_tokens)}\nGPT-NeoX: {len(gpt_neox_tokens)}\nLLaMa: {len(llama_tokens)}\nFalcon: {len(falcon_tokens)}\nPhi-2: {len(phi2_tokens)}\nT5: {len(t5_tokens)}"
21
 
22
  iface = gr.Interface(fn=tokenize, inputs=gr.Textbox(lines=7), outputs="text")
23
  iface.launch()
 
8
  falcon_tokenizer = AutoTokenizer.from_pretrained("tiiuae/falcon-7b")
9
  phi2_tokenizer = AutoTokenizer.from_pretrained("microsoft/phi-2")
10
  t5_tokenizer = AutoTokenizer.from_pretrained("google/flan-t5-xxl")
11
+ mistral_tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-7B-v0.1")
12
 
13
  def tokenize(input_text):
14
  gpt2_tokens = gpt2_tokenizer(input_text, add_special_tokens=True)["input_ids"]
 
17
  falcon_tokens = falcon_tokenizer(input_text, add_special_tokens=True)["input_ids"]
18
  phi2_tokens = phi2_tokenizer(input_text, add_special_tokens=True)["input_ids"]
19
  t5_tokens = t5_tokenizer(input_text, add_special_tokens=True)["input_ids"]
20
+ mistral_tokens = mistral_tokenizer(input_text, add_special_tokens=True)["input_ids"]
21
 
22
+ return f"GPT-2/GPT-J: {len(gpt2_tokens)}\nGPT-NeoX: {len(gpt_neox_tokens)}\nLLaMa: {len(llama_tokens)}\nFalcon: {len(falcon_tokens)}\nPhi-2: {len(phi2_tokens)}\nT5: {len(t5_tokens)}\nMistral: {len(mistral_tokens)}"
23
 
24
  iface = gr.Interface(fn=tokenize, inputs=gr.Textbox(lines=7), outputs="text")
25
  iface.launch()