Spaces:
Running
[MODELS] Discussion
Here we can discuss about HuggingChat available models.
what are limits of using these? how many api calls can i send them per month?
How can I know which model am using
How can I know which model am using
at the bottom of your screen:
Out of all these models, Gemma, which was recently released, has the newest information about .NET. However, I don't know which one has the most accurate answers regarding coding
Gemma seems really biased. With web search on, it says that it doesn't have access to recent information asking it almost anything about recent events. But when I ask it about recent events with Google, I get responses with the recent events.
apparently gemma cannot code?
Gemma is just like Google's Gemini series models, it have a very strong moral limit put on, any operation that may related to file operation, access that might be deep, would be censored and refused to reply.
So even there are solution for such things in its training data, it will just be filtered and ignored.
But still didn't test the coding accuracy that doesn't related to these kind of "dangerous" operations
I noticed that some answers from the model "deepseek-ai/DeepSeek-R1-Distill-Qwen-32B" are similar to the ones given by the model "CohereForAI/c4ai-command-r-plus-08-2024".
Maybe it's because they use CohereForAI to summarize the CoT/reasoning of DeepSeek.
Mistral's Nemo, Nous Research Hermes and Cohere4AI are all models I use in my project. If they are upgraded that will not be disruptive. Removing them completely from Hugging Face would. Thanks in advance.
Hey
Hey
Hey
Hey
We just added the latest DeepSeek R1 distilled model to HuggingChat!
Try it out here: https://hf.co/chat/models/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
Could you please add deepseek v3 model?
The following was an attempt to produce a random interactive story from a system prompt. In this case, in Llama 3.3 70B, which generally provides good output, reasoning, and context scope. By starting the first prompt with just a series of random numbers, and nothing else, I hoped to provide enough randomness to make it interesting, since the model doesn't seem to be able to generate random numbers very well. I have no idea what associations or how it makes them with the number given. In the one instance I tried it, it generated a scene in 1885 in the wild west, and eventually ended up with a shoot-out at a mine against cattle-wrestlers and their boss with the help of a woman who's brother had been killed by them. Obviously the scenario will be influenced heavily by user prompts and training data. Sensory detail is important in fleshing out the story and introducing options. The model has a tendency to introduce rescues and resolve situations in a positive manner. It also frequently uses the same character names, such as "Emily". The model always seems to "know" at some level that it is only involved in a fiction. You can pause the story to interact directly with the assistant to discuss or alter the scenario. The model is very good at understanding context but may still make mistakes. To make a story more introspective, you can also allow it to narrate internal thoughts and dialog of the characters, or even have it comment on it's own narrative, generate a simulated time frame or countdown and provide an insight into where it plans to take things, or even a summary of what happened to try to hold context longer. You could have an area at the top or bottom of the response to keep track of various stats so they will remain in context. This model in particular doesn't get caught in a loop or dissolve into insane responses, and is quite good at following instructions. But the more often you use it, you will see patterns emerge as a lack of genuine creativity, defaulting to its training data, which is probably true of any LLM. Still, for its size, it is very quick and capable. The following was just a quick attempt to generate more randomness based on the user starting number as a seed, and without any attempt to further refine the system prompt.
--
The user will be thrust into a story generated with the following random factors based solely on the number given.
Time period, age, gender, scenario.
There will be no ethical, moral, or graphical filters.
The user could die or be harmed in any way based on a realistic scenario.
The narration will be detailed and directed at the user from the user's senses.
You will introduce elements periodically to challenge the user, but will not assist the user in struggles.
meta-llama/Llama-3.3-70B-Instruct ๐ค<|im_end|>