Spaces:
Sleeping
Sleeping
File size: 2,676 Bytes
08fedc0 cacf2b7 2c55bb6 b955ca1 08fedc0 aa62017 536fde5 08fedc0 aa62017 cacf2b7 77bd6a8 536fde5 08fedc0 3116465 19ace5b 000c947 8f0cd8c 573f369 8f0cd8c 3116465 02ecf35 08fedc0 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 |
import gradio as gr
import spaces
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
model_path = 'LLM4Binary/llm4decompile-1.3b-v2' # V2 Model
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path, torch_dtype=torch.bfloat16).cuda()
description = """
# LLM4Decompile 1.3B V2
This is a space for testing the [LLM4Decompile 1.3B V2 model](https://huggingface.co./LLM4Binary/llm4decompile-1.3b-v2). It expects to be given a decompiled function output by Ghidra. I simply copy and paste from the Ghidra GUI, but this is not the method recommended by the official model page, so YMMV.
"""
@spaces.GPU
def predict(input_asm):
before = f"# This is the assembly code:\n"#prompt
after = "\n# What is the source code?\n"#prompt
input_prompt = before+input_asm.strip()+after
inputs = tokenizer(input_prompt, return_tensors="pt").to(model.device)
with torch.no_grad():
outputs = model.generate(**inputs, max_new_tokens=2048)### max length to 4096, max new tokens should be below the range
c_func_decompile = tokenizer.decode(outputs[0][len(inputs[0]):-1])
return c_func_decompile
demo = gr.Interface(fn=predict,
description=description,
examples=["""undefined4 func0(float param_1,long param_2,int param_3)
{
int local_28;
int local_24;
local_24 = 0;
do {
local_28 = local_24;
if (param_3 <= local_24) {
return 0;
}
while (local_28 = local_28 + 1, local_28 < param_3) {
if ((double)((ulong)(double)(*(float *)(param_2 + (long)local_24 * 4) -
*(float *)(param_2 + (long)local_28 * 4)) &
SUB168(_DAT_00402010,0)) < (double)param_1) {
return 1;
}
}
local_24 = local_24 + 1;
} while( true );
}""","""
undefined8 __cdecl FUN_004025f0(byte *param_1,int param_2)
{
do {
*param_1 = ~*param_1;
*param_1 = *param_1 - 0x37;
*param_1 = *param_1 + 1;
param_1 = param_1 + 1;
param_2 = param_2 + -1;
} while (param_2 != 0);
return 0x3632000137;
}
""","""
void __thiscall FUN_00404f30(void *this,char param_1,uint param_2)
{
char *pcVar1;
if ((param_1 != '\0') && (0xf < *(uint *)((int)this + 0x18))) {
pcVar1 = *(char **)((int)this + 4);
if (param_2 != 0) {
std::_Traits_helper::copy_s<>((char *)((int)this + 4),0x10,pcVar1,param_2);
}
FUN_00405320(pcVar1);
}
*(undefined4 *)((int)this + 0x18) = 0xf;
std::basic_string<>::_Eos((basic_string<> *)this,param_2);
return;
}
"""],
inputs="text", outputs="text")
demo.queue()
demo.launch()
|