Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,11 +1,12 @@
|
|
| 1 |
from transformers import AutoTokenizer
|
| 2 |
import gradio as gr
|
| 3 |
|
|
|
|
|
|
|
| 4 |
|
| 5 |
def tokenize(input_text):
|
| 6 |
-
llama_tokens =
|
| 7 |
-
|
| 8 |
-
)
|
| 9 |
llama3_tokens = len(
|
| 10 |
llama3_tokenizer(input_text, add_special_tokens=True)["input_ids"]
|
| 11 |
)
|
|
@@ -41,7 +42,22 @@ def tokenize(input_text):
|
|
| 41 |
)
|
| 42 |
|
| 43 |
results = {
|
| 44 |
-
"LLaMa-1/LLaMa-2": llama_tokens,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
"LLaMa-3": llama3_tokens,
|
| 46 |
"Mistral": mistral_tokens,
|
| 47 |
"GPT-2/GPT-J": gpt2_tokens,
|
|
@@ -58,7 +74,9 @@ def tokenize(input_text):
|
|
| 58 |
# Sort the results in descending order based on token length
|
| 59 |
sorted_results = sorted(results.items(), key=lambda x: x[1], reverse=True)
|
| 60 |
|
| 61 |
-
|
|
|
|
|
|
|
| 62 |
|
| 63 |
|
| 64 |
if __name__ == "__main__":
|
|
@@ -102,4 +120,4 @@ if __name__ == "__main__":
|
|
| 102 |
iface = gr.Interface(
|
| 103 |
fn=tokenize, inputs=gr.Textbox(label="Input Text", lines=12), outputs="text"
|
| 104 |
)
|
| 105 |
-
iface.launch()
|
|
|
|
| 1 |
from transformers import AutoTokenizer
|
| 2 |
import gradio as gr
|
| 3 |
|
| 4 |
+
def formatarr(input):
|
| 5 |
+
return "["+",".join(str(x) for x in input)+"]"
|
| 6 |
|
| 7 |
def tokenize(input_text):
|
| 8 |
+
llama_tokens = llama_tokenizer(input_text, add_special_tokens=True)["input_ids"]
|
| 9 |
+
|
|
|
|
| 10 |
llama3_tokens = len(
|
| 11 |
llama3_tokenizer(input_text, add_special_tokens=True)["input_ids"]
|
| 12 |
)
|
|
|
|
| 42 |
)
|
| 43 |
|
| 44 |
results = {
|
| 45 |
+
"LLaMa-1/LLaMa-2": len(llama_tokens),
|
| 46 |
+
"LLaMa-3": llama3_tokens,
|
| 47 |
+
"Mistral": mistral_tokens,
|
| 48 |
+
"GPT-2/GPT-J": gpt2_tokens,
|
| 49 |
+
"GPT-NeoX": gpt_neox_tokens,
|
| 50 |
+
"Falcon": falcon_tokens,
|
| 51 |
+
"Phi-1/Phi-2": phi2_tokens,
|
| 52 |
+
"T5": t5_tokens,
|
| 53 |
+
"Gemma": gemma_tokens,
|
| 54 |
+
"Command-R": command_r_tokens,
|
| 55 |
+
"Qwen/Qwen1.5": qwen_tokens,
|
| 56 |
+
"CodeQwen": codeqwen_tokens,
|
| 57 |
+
}
|
| 58 |
+
|
| 59 |
+
results2 = {
|
| 60 |
+
"LLaMa-1/LLaMa-2": formatarr(llama_tokens),
|
| 61 |
"LLaMa-3": llama3_tokens,
|
| 62 |
"Mistral": mistral_tokens,
|
| 63 |
"GPT-2/GPT-J": gpt2_tokens,
|
|
|
|
| 74 |
# Sort the results in descending order based on token length
|
| 75 |
sorted_results = sorted(results.items(), key=lambda x: x[1], reverse=True)
|
| 76 |
|
| 77 |
+
lens = "\n".join([f"{model}: {tokens}" for model, tokens in sorted_results])
|
| 78 |
+
toks = "\n".join([f"{model}: {tokens}" for model, tokens in results2])
|
| 79 |
+
return lens + "\n" + toks
|
| 80 |
|
| 81 |
|
| 82 |
if __name__ == "__main__":
|
|
|
|
| 120 |
iface = gr.Interface(
|
| 121 |
fn=tokenize, inputs=gr.Textbox(label="Input Text", lines=12), outputs="text"
|
| 122 |
)
|
| 123 |
+
iface.launch()
|