121–130 of 185 posts
> 90%* > * According to a fun and non-scientific evaluation with GPT-4. Further rigorous evaluation is needed. Come on... I love any effort related to LLM, but it's really disingenuous to claim high quality in the title and then immediately discredit it after click-through.
but it is indeed difficult to eval chatbots and LLM esp. considering most of them have actually seen the Internet data at least once.
> In our first release, we will share the training, serving, and evaluation code. We plan to release the model weights by providing a version of delta weights that build on the original LLaMA weights, but we are still figuring out a proper way to do so. Join our Discord server and follow our Twitter to get the latest updates. Please correct me if I'm wrong, but it seems like this is not actually an open-source model?…
Train it on what? I want host our own AI, have it injest our docs, and answer questions about them. What is the best tool today to do that?
Earlier quoted context omitted.
> However, if this is true, it seems pretty cool if I can train a GPT4-like model for only $300. Not train, but fine-tune a model that Facebook spent millions of dollars training.
They absolutely did not spend millions to train it. Credible estimates place the cost for an entity like Meta at about 30-100k, probably less since Meta likely owns the 256 A100x8s needed to train it. Even as an individual, it wouldn't cost you anywhere near a million if you only trained 13B and took advantage of volume pricing.
Earlier quoted context omitted.
I am a Vicuna developer. We plan to release the weights once we have addressed all concerns and have a low-resource version of the inference code ready. We released the demo first to get some early feedback on the model.
We hear a lot about "concerns" and many of us don't share the same ones... It would be good for clarity to know what are the concerns you feel are important enough to hold back releasing the weights?
>* According to a fun and non-scientific evaluation with GPT-4. Further rigorous evaluation is needed. I am so sick of seeing these ridiculous claims made about finetuned versions of llama, with 0 scientific rigor behind them. This is, I believe, the 3rd llama finetune I've seen posted within the past 2 weeks, of which all claim "similar to ChatGPT" quality, while not actually running it through a _single_ of the man…
Really starting to be annoyed at misleading use of 'open' and 'open-source' in the "AI" circles. When it's not downright a decoy.
the chatbot gui is open source, the thing to make the chatbot work is not released at all. basically, this is just a wrapper for your own LLM .bin file and is just a GUI for any LLM. instead of typing a prompt in terminal, now you.... go to terminal anyway and it launches a happy looking GUI
Q: What is the meaning of ABBA's song "I have a dream"?
A: "I Have a Dream" is a song by the Swedish pop group ABBA, released in 1970. The song was written by Benny Andersson and Björn Ulvaeus, and it was the group's first hit single. The song is a ballad that talks about a person's dream of finding love and happiness. The lyrics describe the dream as a "beautiful light in the dark" and a "feeling inside that no one can steal." The song has become a classic of the ABBA catalog and is still widely played and enjoyed today.
(Didn't check the details but seems okish). But let's not ask about a song that doesn't exist:
Q: And what did they mean with the song "General Trouble"?
A: "General Trouble" is a song by the Swedish pop group ABBA, released in 1969. The song was written by Benny Andersson and Björn Ulvaeus, and it was the group's first single. The song is a pop ballad that talks about a person who is experiencing a lot of problems in their life, including relationship troubles, financial troubles, and feelings of isolation. The lyrics describe the person as being in "general trouble" and feeling like they are "losing ground." The song was not a big hit for ABBA, but it is still an interesting early work from the group and provides insight into their early sound and songwriting style.
> In our first release, we will share the training, serving, and evaluation code. We plan to release the model weights by providing a version of delta weights that build on the original LLaMA weights, but we are still figuring out a proper way to do so. Join our Discord server and follow our Twitter to get the latest updates. Please correct me if I'm wrong, but it seems like this is not actually an open-source model?…
I am a Vicuna developer. We plan to release the weights once we have addressed all concerns and have a low-resource version of the inference code ready. We released the demo first to get some early feedback on the model.
> In our first release, we will share the training, serving, and evaluation code. We plan to release the model weights by providing a version of delta weights that build on the original LLaMA weights, but we are still figuring out a proper way to do so. Join our Discord server and follow our Twitter to get the latest updates. Please correct me if I'm wrong, but it seems like this is not actually an open-source model?…
I am a Vicuna developer. We plan to release the weights once we have addressed all concerns and have a low-resource version of the inference code ready. We released the demo first to get some early feedback on the model.
- Low memory footprint(thanks quantization)
- Fast inference(thanks io binding)
Particularly in case of alpaca I have seen a 5x decrease in latency on A100 and 10x on AMD EPYC. I believe this is the way for users to have an AI that could genereate a response as fast as it can on their hardware. I have also added a link to my profile on hf with small alpacas turned into ONNX format. Take a look into them.