Comments (5)
For me it would be using something else -downloaded models in KoboldAI, but I suppose it makes sense to have ways to do it both ways.
The tensorflow version of the model_v5 GPT-2 (the original AI Dungeon 2 model) at least seems to fail when I select option "Custom GPT-2 (eg CloverEdition)":
Initializing transformers, please wait...
2021-06-03 20:41:12.078066: I tensorflow/stream_executor/platform/default/dso_loader.cc:53] Successfully opened dynamic library libcudart.so.11.0
file /mnt/ssd/aidungeon-tensorflow-models/model_v5/config.json not found
Traceback (most recent call last):
File "/usr/lib/python3.9/site-packages/transformers/configuration_utils.py", line 458, in get_config_dict
resolved_config_file = cached_path(
File "/usr/lib/python3.9/site-packages/transformers/file_utils.py", line 1180, in cached_path
raise EnvironmentError(f"file {url_or_filename} not found")
OSError: file /mnt/ssd/aidungeon-tensorflow-models/model_v5/config.json not found
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/mnt/lvdata/games/AIDungeon2/KoboldAI-Client/aiserver.py", line 322, in <module>
model = GPT2LMHeadModel.from_pretrained(vars.custmodpth)
File "/usr/lib/python3.9/site-packages/transformers/modeling_utils.py", line 975, in from_pretrained
config, model_kwargs = cls.config_class.from_pretrained(
File "/usr/lib/python3.9/site-packages/transformers/configuration_utils.py", line 401, in from_pretrained
config_dict, kwargs = cls.get_config_dict(pretrained_model_name_or_path, **kwargs)
File "/usr/lib/python3.9/site-packages/transformers/configuration_utils.py", line 478, in get_config_dict
raise EnvironmentError(msg)
OSError: Can't load config for '/mnt/ssd/aidungeon-tensorflow-models/model_v5/'. Make sure that:
- '/mnt/ssd/aidungeon-tensorflow-models/model_v5/' is a correct model identifier listed on 'https://huggingface.co/models'
- or '/mnt/ssd/aidungeon-tensorflow-models/model_v5/' is the correct path to a directory containing a config.json file
I checked that it does work in at least AID: Unleashed.
A sidenote: I'm not actually using the GUI to select these models; I edited the code to have modpath directly have the correct path. An "upvote" for that headless option mentioned in #3; it's going to be really clunky to navigate to the desired model's directory every time I want to play.
from koboldai-client.
I think it already offers this option. Suppose you want to run another copy of GPT-Neo 2.7B, you can do the following. Instead of selecting option 2 (that downloads it to some folder in your C:), select option 8 (Custom Neo) and point CMD to wherever you have this normal GPT-Neo 2.7B model installed (though you can also point to the finetunes). All the other model types offer similar options.
I don't know where Kobold stores its models by default, but if you know that I'm assuming you can also just symlink like you used to.
from koboldai-client.
(Can't edit my post.)
Or did you mean the other way around? For you to use the Kobold-downloaded models in something else (e.g. Clover)?
from koboldai-client.
Model Directory:
- KoboldAI uses the "Transformers" library which configures the neural network layouts, and loads/downloads pre-trained models.
- Transformers downloads ALL data (temporary and permanent) into
HOME FOLDER / .cache / huggingface / transformers
on all platforms (Windows, Linux, Mac). - The models are stored as randomly named files, with a JSON file that contains the path it was downloaded from, which can help you identify and delete unwanted models.
- If the Transformers library crashes during a download, it will have orphaned temporary files in that folder. So feel free to look there and delete them. I can't remember the temp name but I think it had "temp" or "tmp" in the filename. Delete all of those if you see any.
from koboldai-client.
In 1.17 it now converts the models to efficient local models stored in the KoboldAI directory. In 1.18 this is further optimized with a seperate models directory. Downloaded models can be moved around if you use option 1 to load them. If you keep them in their original location they can also be selected from the main menu.
from koboldai-client.
Related Issues (20)
- delete
- softprompt tuner not working HOT 2
- Slow HOT 1
- How to set an API secret ? HOT 3
- No longer update? HOT 1
- ImportError: cannot import name 'dependency_versions_check'
- Colab readme.md link 404
- Micromamba URI can get corrupted by wget [Linux] HOT 1
- Change chat format with API
- The process suddenly stops for some reason
- Problem at remote-play HOT 3
- Colab only showing one model HOT 1
- Error when generating story HOT 1
- Fix def get_folder_path_info(base) for MacOS - any input?
- model type issue
- ImportError: cannot import name 'BaseResponse' from 'werkzeug.wrappers'
- Trying to set up, but fails: "ModuleNotFoundError: No module named 'eventlet'" HOT 3
- Feature Request: Support Ollama API HOT 1
- [Linux, Recommended install] play-rocm.sh immediately fatally crashes with huggingface_hub import error HOT 3
- [SOLVED] split_torch_state_dict_into_shards HOT 3
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from koboldai-client.