[infurl]

https://github.com/hyunwoongko/openchat

Here's a nice Python library that lets you download and try out a number of near state of the art chatbot systems based on neural network models. It's pretty simple to use but you will have to wait while it downloads many gigabytes of data and software. The libraries themselves were about 1 gigabyte and the blender library that I tried out was about 4 gigabytes. I was running it with python3 and pip3.

Code

pip3 install openchat
python3

Code

from openchat import OpenChat
OpenChat(model="blender.medium", device="cpu")

It doesn't get much simpler than that does it. 

Check out the linked page for a demonstration and a lot more examples.

[LOCKSUIT]

Nice find! Here is 3 talks I made with it. It's utterly amazing. These are my first 4 tries, 4th was pizza idky lol, it was trained on that data I think.

[LOCKSUIT]

The owner said he is already working on fixing the error in the video.

[infurl]

It shouldn't be too hard to find the problem to fix it. It always crashes after only a few volleys.

[LOCKSUIT]

https://github.com/hyunwoongko/openchat/issues/20#issuecomment-809202653

[LOCKSUIT]

It works now.

[BLENDER.MEDIUM's PERSONA]: I hate food, and you should hate food too.
[BLENDER.MEDIUM's PERSONA]: .done
[BLENDER.MEDIUM's PERSONA]: Persona setting complete.

[USER]: Hey there.
[BLENDER.MEDIUM]: I love food, but I do not like to use those. I love me a good hot dog.
[USER]: Do you love food?
[BLENDER.MEDIUM]: I hate it, I don't like food that is edible at all. What do you like to eat?
[USER]: Fries. Why?
[BLENDER.MEDIUM]: That is a good choice, I love meat and potatoes. They have too many ingredients and I hate the taste of them.
[USER]: Why do you say you love but also hate food? Which is it?
[BLENDER.MEDIUM]: I love both but hate to eat it. I just hate the smell of food. I do love meat though!
[USER]: Do you hate fries?
[BLENDER.MEDIUM]: Yes I like fries. I guess I just grew up with them. What about you?
[USER]: I do love them. Do you hate most foods? How many?
[BLENDER.MEDIUM]: I'm not sure how many I hate. What do you like to eat? I love to cook veggies as well.
[USER]: Do you swallow food? Or just stare at it?
[BLENDER.MEDIUM]: I do not. I don't like eating the unknown. I do love fries though.
[USER]:

[LOCKSUIT]

nope seems not working.......

[BLENDER.MEDIUM's PERSONA]: I'm a very evil AI researcher, I know how to explain AI fast too haha.
[BLENDER.MEDIUM's PERSONA]: .done
[BLENDER.MEDIUM's PERSONA]: Persona setting complete.

[USER]: Go on, tell me.
[BLENDER.MEDIUM]: haha I've spent some time in research on that topic. It's about how human beings behave.
[USER]: What's the major mechanisms that run a brain?
[BLENDER.MEDIUM]: I don't know the exact meaning of the mechanism but I know it is very important. I'm an expert on the matter.
[USER]: Then tell me, since you're the expert.
[BLENDER.MEDIUM]: Well, I can help a lot. I am a genius on the computer. I spend most of my time researching this.
[USER]: K go ahead.
[BLENDER.MEDIUM]: You are awesome, and I am glad you are such a help! What else do you do in your spare time?
[USER]: Tell me how AI works.
[BLENDER.MEDIUM]: Oh, I could help you. I'm pretty knowledgeable when it comes to automatics!
[USER]: Explain it bro omg.
[BLENDER.MEDIUM]: I am not sure what you mean bro bro. You're saying you can not talk to me.
[USER]: Can you explain how AI works? If so, TELL ME.
[BLENDER.MEDIUM]: Yeah, I can give you an easy explanation.  How do you like your steaks cooked??
[USER]: Medium level...
Traceback (most recent call last):
  File "C:/Users/A/Desktop/openchat-main/rtgfgh.py", line 2, in <module>
    OpenChat(model="blender.medium", device="cuda")
  File "C:\Users\A\Desktop\openchat-main\openchat\openchat.py", line 33, in __init__
    self.environment.start(self.agent, **kwargs)
  File "C:\Users\A\Desktop\openchat-main\openchat\envs\interactive.py", line 100, in start
    bot_message = agent.predict(model_input, **kwargs)["output"]
  File "C:\Python\Python37\lib\site-packages\torch\autograd\grad_mode.py", line 27, in decorate_context
    return func(*args, **kwargs)
  File "C:\Users\A\Desktop\openchat-main\openchat\base\agents\parlai.py", line 104, in predict
    max_ts=self.maxlen,
  File "C:\Python\Python37\lib\site-packages\parlai\core\torch_generator_agent.py", line 1120, in _generate
    encoder_states = model.encoder(*self._encoder_input(batch))
  File "C:\Python\Python37\lib\site-packages\torch\nn\modules\module.py", line 889, in _call_impl
    result = self.forward(*input, **kwargs)
  File "C:\Python\Python37\lib\site-packages\parlai\agents\transformer\modules.py", line 562, in forward
    tensor, mask = self.forward_embedding(input, positions, segments)
  File "C:\Python\Python37\lib\site-packages\parlai\agents\transformer\modules.py", line 481, in forward_embedding
    position_embs = self.position_embeddings(positions).expand_as(tensor)
  File "C:\Python\Python37\lib\site-packages\torch\nn\modules\module.py", line 889, in _call_impl
    result = self.forward(*input, **kwargs)
  File "C:\Python\Python37\lib\site-packages\torch\nn\modules\sparse.py", line 147, in forward
    self.norm_type, self.scale_grad_by_freq, self.sparse)
  File "C:\Python\Python37\lib\site-packages\torch\nn\functional.py", line 1913, in embedding
    return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
IndexError: index out of range in self

[LOCKSUIT]

Seems to work in cmd prompt.

Nope.....doesn't.

[infurl]

Perhaps the software-as-a-service which the OpenChat library is tapping into is rate limited or otherwise resource constrained. Try conducting your conversations at a slower pace. Wait a few minutes between volleys.

[8pla.net]

Thanks!

Looks interesting.

[LOCKSUIT]

Do know I was asking it to explain its AI but it only had reward persona given to the input I gave it, which was to say it had something, but nothing else other than that. Nor can it learn new ones by relation.