mirror of
https://github.com/Aider-AI/aider.git
synced 2025-05-29 08:44:59 +00:00
Better logic to determine context window size
This commit is contained in:
parent
9d1c53489e
commit
6e3738c901
1 changed files with 19 additions and 7 deletions
|
@ -1,15 +1,27 @@
|
|||
import re
|
||||
|
||||
known_tokens = {
|
||||
"gpt-3.5-turbo": 4,
|
||||
"gpt-4": 8,
|
||||
}
|
||||
|
||||
|
||||
class Model:
|
||||
def __init__(self, name, tokens=None):
|
||||
def __init__(self, name):
|
||||
self.name = name
|
||||
if tokens is None:
|
||||
|
||||
tokens = None
|
||||
|
||||
match = re.search(r"-([0-9]+)k", name)
|
||||
if match:
|
||||
tokens = int(match.group(1))
|
||||
else:
|
||||
for m, t in known_tokens.items():
|
||||
if name.startswith(m):
|
||||
tokens = t
|
||||
|
||||
default_tokens = 8
|
||||
|
||||
tokens = int(match.group(1)) if match else default_tokens
|
||||
if tokens is None:
|
||||
raise ValueError(f"Unknown context window size for model: {name}")
|
||||
|
||||
self.max_context_tokens = tokens * 1024
|
||||
|
||||
|
@ -28,6 +40,6 @@ class Model:
|
|||
return self.is_gpt35()
|
||||
|
||||
|
||||
GPT4 = Model("gpt-4", 8)
|
||||
GPT4 = Model("gpt-4")
|
||||
GPT35 = Model("gpt-3.5-turbo")
|
||||
GPT35_16k = Model("gpt-3.5-turbo-16k")
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue