👨‍🏫 Tutorial Educational Purpose: FREE gpt-4o-mini, gemma2, claude-3, mixtral, llama3.2 and wikipedia AI

[alist1986]

bigyan na yang si memeng! hindi ka titigilan niyan hanggang di mapagbigyan.. napaka informative talaga ng thread pag tatak alist1986 , punung-puno ng kaalaman.. anyways, try niyo tong gpt4free na suggest ni pakner, convertible yan to api once deployed, kaya pwede niyo madala kahit saang GPT UI..

He he. Hindi ko na nga napangalanan pa yang mga pihikan kong "pusakal" sa dami. Mapili pa sa brand ng cat food!
Kahibangan ko lang ito dahil wala akong magawa sa bahay. Bahagi lang ng brain training ko para laging handa kung sakaling may available contract sa dati kong trabaho. Dati rin naman akong college professor at trainer ng professionals, kaya ganito ako magsalita.
Yang gpt4free magiging portable kung nag-create ka ng virtual environment, nag-setup at nagpatakbo ng gui at api kahit isang beses. Kaya yung buong folder ay pwede mong i-save at gamitin kahit saang pc. Same din yan ng python-tgpt na mas maraming LLM models kasama yung na-takedown na providers sa gpt4free like phind. Using the same procedure with python, gagana both gui and api ng gpt4free separately from the specific commands aside from the default for tgpt. Pero kung ang ginamit mo ay yung python-tgpt.exe releases, gui ng gpt4free lang ang stable gamitin dyan. Broken yung api server. Di kumpleto yung compiles sa release o maaaring may conflict sa existing python installation ng user if you didn't use a virtual environment. Mga python users lang ang nakakaintindi nito. Correct me if i'm wrong.
Dyan sa default/whitelisted models ng openwebui sa first post, the best is use gpt-40-mini for all-purpose use hangga't available, segunda na lang yung Claude-haiku at llama-3.2. Though walang access yung users sa function calling, best method is use correctYou do not have permission to view the full content of this post. Log in or register now. to respond correctly. Patterns ng kanilang training database lang ang sinusunod ng LLMs at hindi pa sila sanay sa mga iba't ibang paggamit ng salita ng tao para makagawa ng makatotohanang mga sagot he he. Pag mali ang entrada mo, kahit GPT-4o magkakamaling sumagot. Check nyo na lang sa You do not have permission to view the full content of this post. Log in or register now. din yung mga dapat iwasang LLMs maging sa You do not have permission to view the full content of this post. Log in or register now. para mas effective ang paggamit ng LLMs. Meron din sa You do not have permission to view the full content of this post. Log in or register now., audio analysis, image analysis, coding You do not have permission to view the full content of this post. Log in or register now. The more the leaderboards we review, the better we understand their capabilities and weaknesses. Mas mabuti, alam yan ng lahat ng users bukod sa kakayahan din ng mga LLM interfaces to add more enhancements sa AI. Avoid using LLMs as pure chatbot using only prompts na walang web-search/browsing, walang vector database, embeddings (You do not have permission to view the full content of this post. Log in or register now.) to read/chunk documents, atbp. kung gusto ninyo ng mahusay na responses. Di bale na sa roleplay he he, at tsikahan lang ang ginagawa, pero pwede rin gamitin yung RAG dyan kung may document kayo for such purpose.

For those interested, try to study some prompting below to create your LLM persona or bot of reference below:

Spoiler

Spoiler contents are visible only to Established Members.

Ilagay nyo as system prompt para mag-create kayo ng persona ng bot - as academic, researcher, writer, etc. The links may guide you on what your LLM will act. Hanggang dyan lang yung kaya ng free user sa server since di kayo pwedeng mag-finetuning ng LLM for a specific area of knowledge. Persona lang via system prompts, tweaking of parameters and RAG at the most. May effect lang ito sa ibang users ng same account, and same mode,l at the same time, in the same free account. Baka kasi biglang naging joker na si GPT-4o-mini habang kausap noong isa using "Nawnaw" he he. Maganda sana kung may bot creation, para hiwalay yung conversation using the same model. As for now, ito yung nakikitang kong problema pakner kung gawin nila ngayon. Di ko pa sinubukan dyan pakner pero obvious naman "kung" enabled yung "Save" option sa settings.

 

[reonjy]

He he. Hindi ko na nga napangalanan pa yang mga pihikan kong "pusakal" sa dami. Mapili pa sa brand ng cat food!
Kahibangan ko lang ito dahil wala akong magawa sa bahay. Bahagi lang ng brain training ko para laging handa kung sakaling may available contract sa dati kong trabaho. Dati rin naman akong college professor at trainer ng professionals, kaya ganito ako magsalita.
Yang gpt4free magiging portable kung nag-create ka ng virtual environment, nag-setup at nagpatakbo ng gui at api kahit isang beses. Kaya yung buong folder ay pwede mong i-save at gamitin kahit saang pc. Same din yan ng python-tgpt na mas maraming LLM models kasama yung na-takedown na providers sa gpt4free like phind. Using the same procedure with python, gagana both gui and api ng gpt4free separately from the specific commands aside from the default for tgpt. Pero kung ang ginamit mo ay yung python-tgpt.exe releases, gui ng gpt4free lang ang stable gamitin dyan. Broken yung api server. Di kumpleto yung compiles sa release o maaaring may conflict sa existing python installation ng user if you didn't use a virtual environment. Mga python users lang ang nakakaintindi nito. Correct me if i'm wrong.
Dyan sa default/whitelisted models ng openwebui sa first post, the best is use gpt-40-mini for all-purpose use hangga't available, segunda na lang yung Claude-haiku at llama-3.2. Though walang access yung users sa function calling, best method is use correctYou do not have permission to view the full content of this post. Log in or register now. to respond correctly. Patterns ng kanilang training database lang ang sinusunod ng LLMs at hindi pa sila sanay sa mga iba't ibang paggamit ng salita ng tao para makagawa ng makatotohanang mga sagot he he. Pag mali ang entrada mo, kahit GPT-4o magkakamaling sumagot. Check nyo na lang sa You do not have permission to view the full content of this post. Log in or register now. din yung mga dapat iwasang LLMs maging sa You do not have permission to view the full content of this post. Log in or register now. para mas effective ang paggamit ng LLMs. Meron din sa You do not have permission to view the full content of this post. Log in or register now., audio analysis, image analysis, coding You do not have permission to view the full content of this post. Log in or register now. The more the leaderboards we review, the better we understand their capabilities and weaknesses. Mas mabuti, alam yan ng lahat ng users bukod sa kakayahan din ng mga LLM interfaces to add more enhancements sa AI. Avoid using LLMs as pure chatbot using only prompts na walang web-search/browsing, walang vector database, embeddings (You do not have permission to view the full content of this post. Log in or register now.) to read/chunk documents, atbp. kung gusto ninyo ng mahusay na responses. Di bale na sa roleplay he he, at tsikahan lang ang ginagawa, pero pwede rin gamitin yung RAG dyan kung may document kayo for such purpose.

For those interested, try to study some prompting below to create your LLM persona or bot of reference below:

Ilagay nyo as system prompt para mag-create kayo ng persona ng bot - as academic, researcher, writer, etc. The links may guide you on what your LLM will act. Hanggang dyan lang yung kaya ng free user sa server since di kayo pwedeng mag-finetuning ng LLM for a specific area of knowledge. Persona lang via system prompts, tweaking of parameters and RAG at the most. May effect lang ito sa ibang users ng same account, and same mode,l at the same time, in the same free account. Baka kasi biglang naging joker na si GPT-4o-mini habang kausap noong isa using "Nawnaw" he he. Maganda sana kung may bot creation, para hiwalay yung conversation using the same model. As for now, ito yung nakikitang kong problema pakner kung gawin nila ngayon. Di ko pa sinubukan dyan pakner pero obvious naman "kung" enabled yung "Save" option sa settings.

galing mo pala mag scraping eh, kita mo eh

 

[Demi God]

Support Local

 

[alist1986]

galing mo pala mag scraping eh, kita mo eh

He he. Di ko hinak yung server mo ah, at si "Nawnaw" account rin yung gamit ko. Yung scraping ng information ay madali kung alam mo yung hahanapin mo. Anyone can do it if you are interested. Sanay ako sa ganitong projects with developers and co-users. Pag may access naman sa server mo, dapat malikot kang bulatlatin ang kaloob-looban ng app para alamin yung maganda at pangit.. But you have to be discreet and frank to comment at the same time para gumanda yung app and attract the users.

Wala pang specific at solid feedbacks yung users mo kaya naghahanap ako ng mga pwedeng baguhin at idagdag na infos dyan sa UI. "Waiting game" and "patience" na lang talaga hanggang makapa nila yung loob nyan he he.

Ok naman yung app mo for normal use, with simple descriptions, pero pwede rin itong idagdag:
1. More detailed infos of free models provided, and the advantage to use each. Lagyan mo ng model cards like below para informative to know more about the model
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.
...
2. Add infos on the STT and TTS functionalities for handsfree voicechat - baka di nila alam ito
Or add some free or "free to try" STT like Whisper and TTS models separately like GTTS
(Parang yung sinabi You do not have permission to view the full content of this post. Log in or register now. to make the GUI service more as a multi-modal chatbot - covered yung text, image, audio/speech and videos.)
3. State specifically which ones are allowed and prohibited sa app - parameters, settings, etc. para maiwasan yung abala
(Di kasi user-friendly sa documentation site. Ma-nosebleed sila he he. Kahit nga ako hirap din sa dami ng menu trees.)
Ito ay bunga lang ng imagination ko while using my Open Webui and yours.

Note: Sira yata yang latest update (v0.3.32) to summarize youtube videos at web search. Kahit anong search engines gamitin ko sablay, pati yung GSE na may api. Sa Betterchat naman ay ayos yung search api ko. Ginamitan ko ng Web Search Tool for SearchXNG, sira rin. Mabuti pa yung v3.3.30 ok. Inform ko lang just in case parehas yung experience mo. Mag-downgrade ako mamaya. Yung mga GPT-4o ko sa open webui, pag napagana ko to summarize youtube videos, ibang videos yung na-analyze he he.

 

[prolink]

Thank you

 

[jhunel026]

{
"error": true,
"message": "Failed to fetch"
}

 

[satora]

thanks for sharing lods <3

 

[reonjy]

guys, sorry, kakauwi lang from work travel.. nasira pala, p*tek.. wait my update,

• fixed web-search script, since broken yung latest update ni openwebui, check You do not have permission to view the full content of this post. Log in or register now..
• adding gpt-4o and claude-3.5-sonnet

 

[alist1986]

guys, sorry, kakauwi lang from work travel.. nasira pala, p*tek.. wait my update,

• fixed web-search script, since broken yung latest update ni openwebui, check You do not have permission to view the full content of this post. Log in or register now..
• adding gpt-4o and claude-3.5-sonnet

He he. Welcome back pakner. Akala ko ako lang may problema. Binura ko muna yang open webui. Balikan ko na lang ulit. Isabay ko siya sa oobabooga pag napagana ko ulit ang SillyTavern na may Live2D he he. Check ko kung anong ok sa dalawa. Kahit anong LLM naman ngayon pwede mong gawan ng You do not have permission to view the full content of this post. Log in or register now.server kahit walang UI he he. Matipid pa sa RAM. Ingat lang yung iba dahil maraming fake GPT-4o ngayon dahil dyan.
Sa ngayon TTS/STT at AI audio muna ako bumaling at pag-aaralan ko ulit yung compatibilities ng python, pytorch, tensorflow para ma-maximize ko yung gamit ng CUDA 10.1 at CUDNN - mas challenging. Bagong kabit lang kasi OS ko't nalimutan ko na yung procedure. Limited ako sa CPU-mode sa local AI. Kaya mga old repos yung aking minimina para makanahap ng ginto sa github he he. SA HF na lang yung open webui ko at bihira naman ako gumamit nyan. Cloning lang, ayos na.

Abang na lang sila sa pagbbalik ni Nawnaw he he!

 

[ShadowSeeker]

thank you boss

 

[alist1986]

Habang nag-iintay kayo, pag-aralan nyo na rin yang ollama open webui. Mamili kayo dito sa baba kung alin ang tipo nyong UI - yung may readymade LLMs o yung kayo yung mag-setup for solo use using Workplace.

Spoiler

Spoiler contents are visible only to Established Members.

Mag-login kayo sa HF para makarami kayo ng requests, or else very limited yung gamit ninyo sa public Spaces. Free to use naman ang spaces pag "running" sila.

Spoiler

Spoiler contents are visible only to Established Members.

 

[alist1986]

Pakner, ayos din pala yung simplistic chipp-chat mo he he.

Alam mo naman ako, mahilig sa palaisipan! Kagigising ko lang kaya lumikot ng kaunti yung isip ko he he.
Ewan ko lang kung napansin ng users yung benefits ng useful features nito - though di naman lahat ay "on" base sa sample pic ng gumawa ako ng aking chipp-chat in one-click. Simplistic ka talaga pakner. AIO na for students at free to use naman si ReonBot app, at GPT-40 w/ no api needed pa. Duling lang ang si marunong bumasa. Natabunan na siguro yung " go You do not have permission to view the full content of this post. Log in or register now.." ng makita yung Red-Bolded Under Maintenance...wait for new update coming soon... Pag libre, kailangan mabilis ang mata!
Di ko pa ma-rig yung Live2d sa ST at naghagilap pa ako ng illustrations at CG models+characters and lorebook sa ST. Maganda kasing magkuting-ting ng LLM UIs para malaman kung ano yung for educational at for entertainment, ano yung ok sa low end hardware at mabilis ang inference, at may multi-api support. Sa local AI, llama.cpp is 2x the speed ng ollama sa pc kong 3rd-gen in cpu mode, kaya my open webui only uses LLM apis. Go kasi ang gamit ng ollama (from llama.cpp) hindi C++. Kaya it's good to know the apps you're using para makagamit with minimal hardware. Lahat naman ng bagay may paraan.

 

[Exorsus]

thyaynks

 

[F-Society]

Hello po TS

Down po ba itong ReonBot?
very useful kasi sya sakin almost everyday ko sya gamit

 

[alist1986]

Hello po TS

Down po ba itong ReonBot?
very useful kasi sya sakin almost everyday ko sya gamit

Yes! Na-reach na yung quota noong last remaining You do not have permission to view the full content of this post. Log in or register now. ni pakner - busy siya siguro! Intayin mong mag-reset kung sakali. Buhay pa naman siya.
Actually, binigyan niya kayo ng "hint" dyan to create your own kung malikot yung mga mata nyo.
Nasa kanan sa baba ng page yon. click You do not have permission to view the full content of this post. Log in or register now. and create your own (workplace, apps...). Meron na kayong sariling Reonbot (chippchat) clone. Customizable pa, he he. You just have to know how these AIs work!

 

[shelovesbigcats]

Error po ts

 

[Style Dog]

Error po ts

Under maintenance daw po. Hoping magtagal to. Big help talaga.

 

[alist1986]

Para doon sa mga nahuli at di nakuha yung last edited post ni reonjy noong OCT. 13, 2024, ito yon:
1. Under maintenance yung open-webui server nya which uses these 2 links to provide the web app on his preferred cloud server:

Credits:
You do not have permission to view the full content of this post. Log in or register now.
You do not have permission to view the full content of this post. Log in or register now.

2. Nag-iwan siya ng back-up (not open webui) but uses gpt-4o:

For GPT-4o with image analysis feature, go You do not have permission to view the full content of this post. Log in or register now..

Since hindi na niya ma-update yang #1 from the time nagloko yung upgraded version last October pa, possibly sa kanyang working schedule, nag-iwan siya ng quick #2 option na mukhang matagal na ring hindi gumagana though buhay yung server.

The easiest way is to clone #2 using my tips sa post#77 para direkta na kayo sa pinanggalingan ng chippchat server nya sa #2. Just create your own. For #1, either you learn to run the code for open-webui and use the models at huggingface or just use the models "spaces" sa huggingface individually from the You do not have permission to view the full content of this post. Log in or register now. provided.

Yan yung nakikita kong last resort using the infos sa post#1 if you can understand what I'm saying.

 

[ITngPinas]

Sobrang useful neto lalo na pag mag dedebug ako ng codes. More power TS!

 

[alist1986]

Sobrang useful neto lalo na pag mag dedebug ako ng codes. More power TS!

Yes, I presume gamit mo yung "Codeman" or custom bots sa chipp marketplace para customized sa iyong coding tasks. Marami namang pagpipilian sa marketplace para ilagay mo sa workplace mo (free or ρáíd) - via no-code app builder option. Though di naman stated kahit saan sa website nila yung GPT na gamit, chipp.ai chat says it's using OpenAI's GPT-4o from my random checks. Siguro sa upgraded account, pwede rin yung ibang models ayon sa kanilang main link:

Verified ko rin ito sa kanilang disc0rd channel.