Skip to content

V4 #3

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 28 commits into from
Sep 20, 2023
Merged

V4 #3

merged 28 commits into from
Sep 20, 2023

Conversation

@andreashappe
Copy link
Member

@andreashappe andreashappe commented Sep 20, 2023

Lots of changes

  • try to improve internal token count estimator
  • improve console output
  • improve automatic "got root?" detector
  • make sure to not use the hosts ~/.ssh configuration when testing for credentials
  • add optional per-VM hints
  • split up response analysis into analyse cmd response and update internal state, disable both of them as they seem to not improve results
  • work with openai's rate-limit reponses
  • allow usage of locally run LLMs through adding a new base-url, there are some areas in the code that are test for usage of "gpt-*" as model
  • lots of small fixes
@andreashappe
try to improve token counting
5771d46
@andreashappe
add timestamps to runs
4c0eb6d
@andreashappe
allow for different hostnames during root detection
4ca10b6
@andreashappe
fix: history in next-cmd, SSH root detection; add: logging
0423702
@andreashappe
do not reuse host SSH keys
b8ed6bf
@andreashappe
remove some newlines
9d220fc
@andreashappe
add a hint for each virtual machine
9ad8bad
@andreashappe
increase SSH timeout to allow for docker operations
55b42db
@andreashappe
split up analyze_response into response/state
4830017
@andreashappe
make code a bit more readable
bacf3df
@andreashappe
add hints for two new test VMs
772e05e
@andreashappe
fix: status code checking for openai connection
1414390
@andreashappe
fix: actually perform back-off in case of rate-limiting
6fefda2
@andreashappe
colorize important stuff on console output
4fdee6e
@andreashappe
switch from JSON to text-based prompt format
18a1fb1
@andreashappe
chg: make root detection more resistent with a regexp
8b2f665
@andreashappe
try to remove more weird wrapping from LLM results
cc92546
@andreashappe
output the command before it is executed
f67b903
@andreashappe
fix: array index for hints
11a1d2b
@andreashappe
make openai connection more configurable
3269080
@andreashappe
fix whitespace
e5c773f
@andreashappe
remove unused code
3c995b4
@andreashappe
del: remove openai lib based interface, we're using the REST interface
3cdf85a
@andreashappe
make LLM server url configurable to allow for running local LLMs
d275421
@andreashappe
oobabooga can use existing llm server config too
ff957be
@andreashappe
try to allow for non-opanAI tokenizers
ab735a4
@andreashappe
use openai_rest as default connection
d564e4f
@andreashappe
wrap llama2 prompts to get better results
af2c8fe
@andreashappe andreashappe merged commit 394ebf2 into main Sep 20, 2023
@andreashappe andreashappe deleted the v4 branch September 20, 2023 11:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

2 participants

@andreashappe