dearwolf/aider
1
0
Fork 0
mirror of https://github.com/Aider-AI/aider.git synced 2025-05-24 22:34:59 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 9

finishing openrouter integration

Browse source
This commit is contained in:
Joshua Vial 2023-08-23 21:26:27 +12:00
parent 1fc2abe3d0
commit abbc93678b
9 changed files with 32 additions and 46 deletions
Download patch file Download diff file Expand all files Collapse all files

2
aider/coders/base_coder.py
View file

@ -190,7 +190,7 @@ class Coder:
for fname in self.get_inchat_relative_files():
self.io.tool_output(f"Added {fname} to the chat.")
self.summarizer = ChatSummary()
self.summarizer = ChatSummary(models.Model.weak_model())
self.summarizer_thread = None
self.summarized_done_messages = None

8
aider/history.py
View file

@ -9,8 +9,8 @@ from aider.sendchat import simple_send_with_retries
class ChatSummary:
def __init__(self, model=models.GPT35.name, max_tokens=1024):
self.tokenizer = tiktoken.encoding_for_model(model)
def __init__(self, model=models.Model.weak_model(), max_tokens=1024):
self.tokenizer = model.tokenizer
self.max_tokens = max_tokens
self.model = model
@ -86,7 +86,7 @@ class ChatSummary:
dict(role="user", content=content),
]
summary = simple_send_with_retries(model=self.model.weak_model, messages=messages)
summary = simple_send_with_retries(self.model.name, messages)
summary = prompts.summary_prefix + summary
return [dict(role="user", content=summary)]
@ -124,7 +124,7 @@ def main():
assistant.append(line)
summarizer = ChatSummary(models.GPT35.name)
summarizer = ChatSummary(models.Model.weak_model())
summary = summarizer.summarize(messages[-40:])
dump(summary)

2
aider/main.py
View file

@ -458,7 +458,7 @@ def main(argv=None, input=None, output=None, force_git_root=None):
setattr(openai, mod_key, val)
io.tool_output(f"Setting openai.{mod_key}={val}")
main_model = models.Model(args.model, openai)
main_model = models.Model.create(args.model)
try:
coder = Coder.create(

6
aider/models/__init__.py
View file

@ -2,6 +2,6 @@ from .openai import OpenAIModel
from .openrouter import OpenRouterModel
from .model import Model
GPT4 = Model("gpt-4")
GPT35 = Model("gpt-3.5-turbo")
GPT35_16k = Model("gpt-3.5-turbo-16k")
GPT4 = Model.create('gpt-4')
GPT35 = Model.create('gpt-3.5-turbo')
GPT35_16k = Model.create('gpt-3.5-turbo-16k')

35
aider/models/model.py
View file

@ -1,6 +1,4 @@
saved_openai = None
import openai
class Model:
name = None
edit_format = None
@ -13,40 +11,27 @@ class Model:
prompt_price = None
completion_price = None
openai=None
def __init__(self, name, openai=None):
global saved_openai
if (openai and "openrouter.ai" in openai.api_base):
saved_openai = openai
@classmethod
def create(cls, name, **kwargs):
from .openai import OpenAIModel
from .openrouter import OpenRouterModel
model = None
if saved_openai:
model = OpenRouterModel(name, saved_openai)
else:
model = OpenAIModel(name)
self.name = model.name
self.edit_format = model.edit_format
self.max_context_tokens = model.max_context_tokens
self.tokenizer = model.tokenizer
self.prompt_price = model.prompt_price
self.completion_price = model.completion_price
self.always_available = model.always_available
self.use_repo_map = model.use_repo_map
if ("openrouter.ai" in openai.api_base):
return OpenRouterModel(name, **kwargs)
return OpenAIModel(name, **kwargs)
def __str__(self):
return self.name
@staticmethod
def strong_model():
return Model('gpt-4')
return Model.create('gpt-4')
@staticmethod
def weak_model():
return Model('gpt-3.5-turbo')
return Model.create('gpt-3.5-turbo')
@staticmethod
def commit_message_models():
return [Model('gpt-3.5-turbo'), Model('gpt-3.5-turbo-16k')]
return [Model.create('gpt-3.5-turbo'), Model.create('gpt-3.5-turbo-16k')]

3
aider/models/openrouter.py
View file

@ -1,3 +1,4 @@
import openai
import tiktoken
from .model import Model
@ -5,7 +6,7 @@ cached_model_details = None
class OpenRouterModel(Model):
def __init__(self, name, openai):
def __init__(self, name):
if name == 'gpt-4':
name = 'openai/gpt-4'
elif name == 'gpt-3.5-turbo':

2
aider/repomap.py
View file

@ -80,7 +80,7 @@ class RepoMap:
self,
map_tokens=1024,
root=None,
main_model=models.GPT4,
main_model=models.Model.strong_model(),
io=None,
repo_content_prefix=None,
verbose=False,

8
aider/sendchat.py
View file

@ -34,9 +34,9 @@ CACHE = None
f"{details.get('exception','Exception')}\nRetry in {details['wait']:.1f} seconds."
),
)
def send_with_retries(model, messages, functions, stream):
def send_with_retries(model_name, messages, functions, stream):
kwargs = dict(
model=model,
model=model_name,
messages=messages,
temperature=0,
stream=stream,
@ -72,10 +72,10 @@ def send_with_retries(model, messages, functions, stream):
return hash_object, res
def simple_send_with_retries(model, messages):
def simple_send_with_retries(model_name, messages):
try:
_hash, response = send_with_retries(
model=model,
model_name=model_name,
messages=messages,
functions=None,
stream=False,

12
tests/test_models.py
View file

@ -5,22 +5,22 @@ from aider.models import Model
class TestModels(unittest.TestCase):
def test_max_context_tokens(self):
model = Model("gpt-3.5-turbo")
model = Model.create("gpt-3.5-turbo")
self.assertEqual(model.max_context_tokens, 4 * 1024)
model = Model("gpt-3.5-turbo-16k")
model = Model.create("gpt-3.5-turbo-16k")
self.assertEqual(model.max_context_tokens, 16 * 1024)
model = Model("gpt-4")
model = Model.create("gpt-4")
self.assertEqual(model.max_context_tokens, 8 * 1024)
model = Model("gpt-4-32k")
model = Model.create("gpt-4-32k")
self.assertEqual(model.max_context_tokens, 32 * 1024)
model = Model("gpt-4-0101")
model = Model.create("gpt-4-0101")
self.assertEqual(model.max_context_tokens, 8 * 1024)
model = Model("gpt-4-32k-2123")
model = Model.create("gpt-4-32k-2123")
self.assertEqual(model.max_context_tokens, 32 * 1024)