#gemma2 — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #gemma2, aggregated by home.social.
-
La mia sanità' di mente sta vacillando....
https://kerberos.archathome.eu/gemma2-2b-su-raspberry-pi-5/
#raspberry #Ollama #AI #nerd #Gemma2 #linux -
La mia sanità' di mente sta vacillando....
https://kerberos.archathome.eu/gemma2-2b-su-raspberry-pi-5/
#raspberry #Ollama #AI #nerd #Gemma2 #linux -
La mia sanità' di mente sta vacillando....
https://kerberos.archathome.eu/gemma2-2b-su-raspberry-pi-5/
#raspberry #Ollama #AI #nerd #Gemma2 #linux -
La mia sanità' di mente sta vacillando....
https://kerberos.archathome.eu/gemma2-2b-su-raspberry-pi-5/
#raspberry #Ollama #AI #nerd #Gemma2 #linux -
La mia sanità' di mente sta vacillando....
https://kerberos.archathome.eu/gemma2-2b-su-raspberry-pi-5/
#raspberry #Ollama #AI #nerd #Gemma2 #linux -
T5Gemma 2 từ Google: mô hình AI đa ngôn ngữ (140+ ngôn ngữ) và đa phương tiện, xử lý văn bản-hình ảnh, hỗ trợ 128K token. 3 cỡ mô hình 270M-4B. #AI #Gemma2 #DeepLearning #Google #Tech #ML #HệThốngAI
-
Gemma-2-MoE: Công cụ nhẹ giúp biến các checkpoint Gemma 2 thành mô hình Mixture of Experts kiểu Frankenstein. Dễ dàng thử nghiệm các biến thể MoE của Gemma 2. #Gemma2 #MoE #AI #MachineLearning #TríTuệNhânTạo #HọcMáy
https://www.reddit.com/r/LocalLLaMA/comments/1ped5x6/gemma2moe_frankenstein_moe_builder_for_gemma_2/
-
Tối ưu hóa gemma2:2b trên Raspberry Pi 5 cho trợ lý ảo với thời gian phản hồi dưới 1 giây #RaspberryPi #TrợLýẢo #gemma2 #Ollama #AI #TríTuệNhânTạo #VoiceAssistant #LocalLLaMA #RaspberryPi5
https://www.reddit.com/r/LocalLLaMA/comments/1ozm3z4/tips_for_optimizing_gemma22b_on_raspberry_pi_5/
-
Wow! I didn't really like Gemma 2, but Gemma 3, released today, is awesome. It comes in four sizes, 1b, 4b, 12b and 27b. It's super fast and except for the 1b version it can even handle images.
The 27B version apparently outperforms both DeepSeek v3 and LLaMA3-405 on the ChatbotArena benchmark.
It's also the first small model I've tested that's good at German.
#gemma3 #gemma #gemma2 #google #ai #programming #google #model #local #gemini #multimodal #vision #wow #chatbotarena #german
-
Wow! I didn't really like Gemma 2, but Gemma 3, released today, is awesome. It comes in four sizes, 1b, 4b, 12b and 27b. It's super fast and except for the 1b version it can even handle images.
The 27B version apparently outperforms both DeepSeek v3 and LLaMA3-405 on the ChatbotArena benchmark.
It's also the first small model I've tested that's good at German.
#gemma3 #gemma #gemma2 #google #ai #programming #google #model #local #gemini #multimodal #vision #wow #chatbotarena #german
-
Wow! I didn't really like Gemma 2, but Gemma 3, released today, is awesome. It comes in four sizes, 1b, 4b, 12b and 27b. It's super fast and except for the 1b version it can even handle images.
The 27B version apparently outperforms both DeepSeek v3 and LLaMA3-405 on the ChatbotArena benchmark.
It's also the first small model I've tested that's good at German.
#gemma3 #gemma #gemma2 #google #ai #programming #google #model #local #gemini #multimodal #vision #wow #chatbotarena #german
-
Wow! I didn't really like Gemma 2, but Gemma 3, released today, is awesome. It comes in four sizes, 1b, 4b, 12b and 27b. It's super fast and except for the 1b version it can even handle images.
The 27B version apparently outperforms both DeepSeek v3 and LLaMA3-405 on the ChatbotArena benchmark.
It's also the first small model I've tested that's good at German.
#gemma3 #gemma #gemma2 #google #ai #programming #google #model #local #gemini #multimodal #vision #wow #chatbotarena #german
-
Wow! I didn't really like Gemma 2, but Gemma 3, released today, is awesome. It comes in four sizes, 1b, 4b, 12b and 27b. It's super fast and except for the 1b version it can even handle images.
The 27B version apparently outperforms both DeepSeek v3 and LLaMA3-405 on the ChatbotArena benchmark.
It's also the first small model I've tested that's good at German.
#gemma3 #gemma #gemma2 #google #ai #programming #google #model #local #gemini #multimodal #vision #wow #chatbotarena #german
-
Just asked my little chat bot a question I didn't understand, got a great answer back.
You: What do the colours 'CMYKOR' stand for?
Gemma2: CMYKOR stands for:* **C**yan
* **M**agenta
* **Y**ellow
* **K**ey (Black)
* **O**range
* **R**edThis color model is used in some specialized printing processes.
You: Why is K black?
Gemma2: In CMYK printing, "K" stands for Key because black ink is used as the key plate to which the other colors (Cyan, Magenta, Yellow) are aligned.Black ink is also crucial for achieving deep, rich blacks and shadows that mixing the other three inks cannot fully replicate.
#Ollama #Gemma2 #AI -
Just asked my little chat bot a question I didn't understand, got a great answer back.
You: What do the colours 'CMYKOR' stand for?
Gemma2: CMYKOR stands for:* **C**yan
* **M**agenta
* **Y**ellow
* **K**ey (Black)
* **O**range
* **R**edThis color model is used in some specialized printing processes.
You: Why is K black?
Gemma2: In CMYK printing, "K" stands for Key because black ink is used as the key plate to which the other colors (Cyan, Magenta, Yellow) are aligned.Black ink is also crucial for achieving deep, rich blacks and shadows that mixing the other three inks cannot fully replicate.
#Ollama #Gemma2 #AI -
Just asked my little chat bot a question I didn't understand, got a great answer back.
You: What do the colours 'CMYKOR' stand for?
Gemma2: CMYKOR stands for:* **C**yan
* **M**agenta
* **Y**ellow
* **K**ey (Black)
* **O**range
* **R**edThis color model is used in some specialized printing processes.
You: Why is K black?
Gemma2: In CMYK printing, "K" stands for Key because black ink is used as the key plate to which the other colors (Cyan, Magenta, Yellow) are aligned.Black ink is also crucial for achieving deep, rich blacks and shadows that mixing the other three inks cannot fully replicate.
#Ollama #Gemma2 #AI -
Just asked my little chat bot a question I didn't understand, got a great answer back.
You: What do the colours 'CMYKOR' stand for?
Gemma2: CMYKOR stands for:* **C**yan
* **M**agenta
* **Y**ellow
* **K**ey (Black)
* **O**range
* **R**edThis color model is used in some specialized printing processes.
You: Why is K black?
Gemma2: In CMYK printing, "K" stands for Key because black ink is used as the key plate to which the other colors (Cyan, Magenta, Yellow) are aligned.Black ink is also crucial for achieving deep, rich blacks and shadows that mixing the other three inks cannot fully replicate.
#Ollama #Gemma2 #AI -
Just asked my little chat bot a question I didn't understand, got a great answer back.
You: What do the colours 'CMYKOR' stand for?
Gemma2: CMYKOR stands for:* **C**yan
* **M**agenta
* **Y**ellow
* **K**ey (Black)
* **O**range
* **R**edThis color model is used in some specialized printing processes.
You: Why is K black?
Gemma2: In CMYK printing, "K" stands for Key because black ink is used as the key plate to which the other colors (Cyan, Magenta, Yellow) are aligned.Black ink is also crucial for achieving deep, rich blacks and shadows that mixing the other three inks cannot fully replicate.
#Ollama #Gemma2 #AI -
New Cloud Platform for Large Language Model Deployment 🚀
🔧 Run any #opensource #LLM supported by #vLLM on autoscaling #GPU clusters, supporting models up to 640GB VRAM
🤖 Compatible with major models: #Llama3 405B/70B/8B, #Qwen2 72B, #Mixtral 8x22B, #Gemma2 27B, #Phi3, and more
💻 Features include:
- #OpenAI compatible #API
- Custom-built #GPU scheduler
- Support for full-weight and 4-bit AWQ repos
- Multi-tenant architecture for cost efficiency🆓 Currently free during beta phase, promising competitive pricing post-launch
-
New Cloud Platform for Large Language Model Deployment 🚀
🔧 Run any #opensource #LLM supported by #vLLM on autoscaling #GPU clusters, supporting models up to 640GB VRAM
🤖 Compatible with major models: #Llama3 405B/70B/8B, #Qwen2 72B, #Mixtral 8x22B, #Gemma2 27B, #Phi3, and more
💻 Features include:
- #OpenAI compatible #API
- Custom-built #GPU scheduler
- Support for full-weight and 4-bit AWQ repos
- Multi-tenant architecture for cost efficiency🆓 Currently free during beta phase, promising competitive pricing post-launch
-
New Cloud Platform for Large Language Model Deployment 🚀
🔧 Run any #opensource #LLM supported by #vLLM on autoscaling #GPU clusters, supporting models up to 640GB VRAM
🤖 Compatible with major models: #Llama3 405B/70B/8B, #Qwen2 72B, #Mixtral 8x22B, #Gemma2 27B, #Phi3, and more
💻 Features include:
- #OpenAI compatible #API
- Custom-built #GPU scheduler
- Support for full-weight and 4-bit AWQ repos
- Multi-tenant architecture for cost efficiency🆓 Currently free during beta phase, promising competitive pricing post-launch
-
New Cloud Platform for Large Language Model Deployment 🚀
🔧 Run any #opensource #LLM supported by #vLLM on autoscaling #GPU clusters, supporting models up to 640GB VRAM
🤖 Compatible with major models: #Llama3 405B/70B/8B, #Qwen2 72B, #Mixtral 8x22B, #Gemma2 27B, #Phi3, and more
💻 Features include:
- #OpenAI compatible #API
- Custom-built #GPU scheduler
- Support for full-weight and 4-bit AWQ repos
- Multi-tenant architecture for cost efficiency🆓 Currently free during beta phase, promising competitive pricing post-launch
-
New Cloud Platform for Large Language Model Deployment 🚀
🔧 Run any #opensource #LLM supported by #vLLM on autoscaling #GPU clusters, supporting models up to 640GB VRAM
🤖 Compatible with major models: #Llama3 405B/70B/8B, #Qwen2 72B, #Mixtral 8x22B, #Gemma2 27B, #Phi3, and more
💻 Features include:
- #OpenAI compatible #API
- Custom-built #GPU scheduler
- Support for full-weight and 4-bit AWQ repos
- Multi-tenant architecture for cost efficiency🆓 Currently free during beta phase, promising competitive pricing post-launch
-
I’ve been continuing the large language model learning experience with my introduction to Hollama. Until now my experience with locally hosted Ollama had been querying models with snippets of Python code, using it in REPL mode and customizing it with text model files. Last week that changed when I listened to a talk about using Hollama.
Hollama is a minimal web user interface for talking to Ollama servers. Like Ollama itself Hollama is open source with an MIT license. Developed initially by Fernando Maclen who is a Miami based designer and software developer. Hollama has nine contributors currently working on the project. It is written in TypeScript and Svelte. The project has documentation on how you can contribute too.
Hollama features large prompt fields, Markdown rendering with syntax highlighting, code editor features, customizable system prompts, multi-language interface along with light and dark themes. You can check out the live demo or download releases for your operating system. You can also self-host with Docker. I decided to download it on the M2 MacBook Air and my Linux computer.
On Linux you download the tar.gz file to your computer and extract it. This opened a directory bearing the name of the compressed file, “Hollama 0.17.4-linux-x64”. I chose to rename the directory Hollama for ease of use. I changed my directory to Hollama and then executed the program.
$ ./holllamaThe program quickly launches and I was presented with the user interface which is intuitive to an extent.
Screen picture by Don Watkins CC by SA 4.0At the bottom of the main menu and not visible in this picture is the toggle for light and dark mode. On the left of the main menu there are four choices. First is ‘Session’ where you will enter your query for the model. The second selection is “Knowledge” where you can develop your model file. Third selection is ‘Settings’ where you will select the model(s) you will use. There is a checkoff for automatic updates. There is a link to browse all the current Ollama models. The final menu selection is ‘Motd’ or message of the day where updates of the project and other news are posted.
Model creation and customization is made much easier using Hollama. In Hollama I complete this model creation in the ‘Knowledge’ tab of the menu. Here I have created a simple ‘Coding’ model as a Python expert.
Screen picture by Don Watkins CC by SA 4.0In ‘Settings’ I specify which model I am going to use. I can download additional models and/or select from the models I already have installed on my computer. Here I have set the model to ‘gemma2:latest’. I have the settings so that my software can check for updates. I also can choose which language the model will use. I have a choice of English, Spanish, Turkish, and Japanese
Screen picture by Don Waktins CC by SA 4.0Now that I have selected the ‘Knowledge’ I am going to use and the model I will use I am ready to use the ‘Session’ section of the menu and create a new session. I selected ‘New Session’ at the top and all my othe parameters are set correctly.
Screen pictire by Don Watkins CC by SA 4.0At the bottom right of the ‘Session’ menu is a box for me to enter the prompt I am going to use.
Screen picture by Don Watkins CC by SA 4.0You can see the output below that is easily accessible.
Screen picture by Don Watkins CC by SA 4.0The output is separated into a code block and a Markdown block so that it is easy to copy the code into a code editor and the Markdown into an editor. Hollama has made working with Ollama much easier for me. Once again demonstrating the versatility and power of open source.
https://donwatkins.info/2024/10/14/exploring-hollama-a-minimalist-web-interface-for-ollama/
-
LangChainでHugging Faceに公開されている日本語版 Gemma 2 2Bを使用する方法
https://qiita.com/moritalous/items/f7de82257426a8f0151b?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
LangChainでHugging Faceに公開されている日本語版 Gemma 2 2Bを使用する方法
https://qiita.com/moritalous/items/f7de82257426a8f0151b?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
SageMaker StudioのJupyterLab環境でStreamlitを使用する（Gemma 2 JPNと会話するおまけ付き）
https://qiita.com/moritalous/items/82fae0eb1820593101d3?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
SageMaker StudioのJupyterLab環境でStreamlitを使用する（Gemma 2 JPNと会話するおまけ付き）
https://qiita.com/moritalous/items/82fae0eb1820593101d3?utm_campaign=popular_items&utm_medium=feed&utm_source=popular_items -
Published new comparison:
Choosing the Best locally hosted #LLM for #Perplexica:
#Llama3, #Llama3.1, #MistralNemo, #Gemma2, #Qwen2, #Phi3 or #Command-r?https://www.glukhov.org/post/2024/08/perplexica-best-llm/
#AI #self-hosted #selfhosted
-
Published new comparison:
Choosing the Best locally hosted #LLM for #Perplexica:
#Llama3, #Llama3.1, #MistralNemo, #Gemma2, #Qwen2, #Phi3 or #Command-r?https://www.glukhov.org/post/2024/08/perplexica-best-llm/
#AI #self-hosted #selfhosted
-
Published new comparison:
Choosing the Best locally hosted #LLM for #Perplexica:
#Llama3, #Llama3.1, #MistralNemo, #Gemma2, #Qwen2, #Phi3 or #Command-r?https://www.glukhov.org/post/2024/08/perplexica-best-llm/
#AI #self-hosted #selfhosted
-
Published new comparison:
Choosing the Best locally hosted #LLM for #Perplexica:
#Llama3, #Llama3.1, #MistralNemo, #Gemma2, #Qwen2, #Phi3 or #Command-r?https://www.glukhov.org/post/2024/08/perplexica-best-llm/
#AI #self-hosted #selfhosted
-
Published new comparison:
Choosing the Best locally hosted #LLM for #Perplexica:
#Llama3, #Llama3.1, #MistralNemo, #Gemma2, #Qwen2, #Phi3 or #Command-r?https://www.glukhov.org/post/2024/08/perplexica-best-llm/
#AI #self-hosted #selfhosted
-
patpro@nutcracker2:~$ ollama run gemma2:27b
pulling manifest
pulling d7e4b00a7d7a... 100% … 15 GB
../..
verifying sha256 digest
writing manifest
removing any unused layers
success
>>>
>>> 10+1+1+1
10 + 1 + 1 + 1 = **14**>>> 10+1+1+1
10 + 1 + 1 + 1 = 13>>> 10+1+1+1
10 + 1 + 1 + 1 = 13 😊Let me know if you have any other math problems!
>>> -
patpro@nutcracker2:~$ ollama run gemma2:27b
pulling manifest
pulling d7e4b00a7d7a... 100% … 15 GB
../..
verifying sha256 digest
writing manifest
removing any unused layers
success
>>>
>>> 10+1+1+1
10 + 1 + 1 + 1 = **14**>>> 10+1+1+1
10 + 1 + 1 + 1 = 13>>> 10+1+1+1
10 + 1 + 1 + 1 = 13 😊Let me know if you have any other math problems!
>>> -
patpro@nutcracker2:~$ ollama run gemma2:27b
pulling manifest
pulling d7e4b00a7d7a... 100% … 15 GB
../..
verifying sha256 digest
writing manifest
removing any unused layers
success
>>>
>>> 10+1+1+1
10 + 1 + 1 + 1 = **14**>>> 10+1+1+1
10 + 1 + 1 + 1 = 13>>> 10+1+1+1
10 + 1 + 1 + 1 = 13 😊Let me know if you have any other math problems!
>>> -
patpro@nutcracker2:~$ ollama run gemma2:27b
pulling manifest
pulling d7e4b00a7d7a... 100% … 15 GB
../..
verifying sha256 digest
writing manifest
removing any unused layers
success
>>>
>>> 10+1+1+1
10 + 1 + 1 + 1 = **14**>>> 10+1+1+1
10 + 1 + 1 + 1 = 13>>> 10+1+1+1
10 + 1 + 1 + 1 = 13 😊Let me know if you have any other math problems!
>>> -
🧠 #NVIDIA researchers are advancing SLMs through structured weight pruning and knowledge distillation, cutting down model size while preserving performance.
For instance, #Minitron 8B and 4B models, derived from Nemotron 15B, outperform many models trained from scratch. Despite its size, it competes with top-tier models like #Gemma2 and #Phi2.
-
🧠 #NVIDIA researchers are advancing SLMs through structured weight pruning and knowledge distillation, cutting down model size while preserving performance.
For instance, #Minitron 8B and 4B models, derived from Nemotron 15B, outperform many models trained from scratch. Despite its size, it competes with top-tier models like #Gemma2 and #Phi2.
