File size: 2,028 Bytes
bcbf898 12c19cf |
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 87 88 |
---
datasets:
- nuprl/EditPackFT-Multi
tags:
- code
---
# What is this
This is a deepseek coder 7b model trained to predict commit messages for a diff.
# Languages trained on:
```py
LANGS = [
"Python",
"Rust",
"JavaScript",
"Java",
"Go",
"C++",
"C#",
"Ruby",
"PHP",
"TypeScript",
"C",
"Scala",
"Swift",
"Kotlin",
"Objective-C",
"Perl",
"Haskell",
"Bash",
"Sh",
"Lua",
"R",
"Julia",
]
```
# How to prompt:
```python
import difflib
class NDiff:
def __init__(self, s1, s2):
self.s1 = s1
self.s2 = s2
self.diff = difflib.ndiff(s1.split("\n"), s2.split("\n"))
def __str__(self):
return "\n".join([l for l in self.diff if l[0] != "?"])
def str_colored(self):
import colored
buf = ""
for l in self.diff:
if l[0] == "?":
continue
if l[0] == "-":
buf += colored.stylize(l, colored.fg("red"))
elif l[0] == "+":
buf += colored.stylize(l, colored.fg("green"))
else:
buf += l
buf += "\n"
return buf
def num_removed(self):
return len([l for l in self.diff if l[0] == "-"])
def num_added(self):
return len([l for l in self.diff if l[0] == "+"])
def __repr__(self):
return self.__str__()
def format_prompt(old, new):
diff_header = "<diff>"
instr_header = "<commit_message>"
diff = str(NDiff(old, new))
return f"{diff_header}\n{diff}\n{instr_header}\n"
def gen(old, new, max_new_tokens=200, temperature=0.45, top_p=0.90):
prompt = format_prompt(old, new)
toks = tokenizer.encode(prompt, return_tensors="pt").to(model.device)
outs = model.generate(toks, max_new_tokens=max_new_tokens, do_sample=True, temperature=temperature, top_p=top_p)
return [tokenizer.decode(out[len(toks[0]):], skip_special_tokens=True) for out in outs]
```
use the "gen" function with the old and new code |