John6666/yesmix-xl-illustrious-v25-sdxl
Text-to-Image
โข
Updated
John Smith PRO
John6666
AI & ML interests
None yet
Recent Activity
published
a model
32 minutes ago
John6666/yesmix-xl-illustrious-v25-sdxl
updated
a model
34 minutes ago
John6666/yesmix-xl-illustrious-v25-sdxl
published
a model
about 4 hours ago
John6666/sparkleberry-illustrious-xl-v10-sdxl
Organizations
John6666's activity
published
a
model
32 minutes ago
updated
a
model
34 minutes ago
published
2
models
about 4 hours ago
updated
2
models
about 5 hours ago
liked
a
model
about 5 hours ago
updated
a
collection
about 5 hours ago
liked
a
Space
about 5 hours ago
reacted to
eaddario's
post with ๐ฅ
about 5 hours ago
Post
414
Tensor-wise (TWQ) and Layer-wise quantization (LWQ) now available in llama.cpp!
As of version b5125 users can now do TWQ, whereby you quantize a whole tensor at a specific level, or perform LWQ by choosing specific layers per tensor/s
The new --tensor-type option enables llama-quantize to apply user-defined quant levels to any combination of allowed tensors (i.e. tensors with 2 or more dimensions) and layer number, with support for regex patterns.
For example, to TWQ the Attention Value tensor you would use --tensor-type attn_v=q6_k and to perform LWQ you'll use something like --tensor-type "\.([0-9]|1[01257]|31)\.attn_v=q4_k"
In the next few days/weeks I'll update the models in my HF repo (and will add some others) but eaddario/DeepSeek-R1-Distill-Llama-8B-GGUF and eaddario/DeepSeek-R1-Distill-Qwen-7B-GGUF have been already LWQed.
For reference, compared to the naive Q4_K_M model, the LWQ Qwen-7B is almost 11% smaller (4.68GB vs 4.18GB) with only a 0.35% penalty on PPL!
I'll update the https://medium.com/@eaddario/squeezing-tensor-bits-the-quest-for-smaller-llms-86b23bd052ca post to explain the process in detail, but in the meantime the following links will provide some background:
- Changes to llama-quantize: https://github.com/ggml-org/llama.cpp/pull/12511
- TWQ & LWQ tests: https://github.com/ggml-org/llama.cpp/discussions/12741
- Modified llama-imatrix (not yet merged) used to generate imatrix statistics to guide the TWQ and LWQ process: https://github.com/ggml-org/llama.cpp/pull/12718
As of version b5125 users can now do TWQ, whereby you quantize a whole tensor at a specific level, or perform LWQ by choosing specific layers per tensor/s
The new --tensor-type option enables llama-quantize to apply user-defined quant levels to any combination of allowed tensors (i.e. tensors with 2 or more dimensions) and layer number, with support for regex patterns.
For example, to TWQ the Attention Value tensor you would use --tensor-type attn_v=q6_k and to perform LWQ you'll use something like --tensor-type "\.([0-9]|1[01257]|31)\.attn_v=q4_k"
In the next few days/weeks I'll update the models in my HF repo (and will add some others) but eaddario/DeepSeek-R1-Distill-Llama-8B-GGUF and eaddario/DeepSeek-R1-Distill-Qwen-7B-GGUF have been already LWQed.
For reference, compared to the naive Q4_K_M model, the LWQ Qwen-7B is almost 11% smaller (4.68GB vs 4.18GB) with only a 0.35% penalty on PPL!
I'll update the https://medium.com/@eaddario/squeezing-tensor-bits-the-quest-for-smaller-llms-86b23bd052ca post to explain the process in detail, but in the meantime the following links will provide some background:
- Changes to llama-quantize: https://github.com/ggml-org/llama.cpp/pull/12511
- TWQ & LWQ tests: https://github.com/ggml-org/llama.cpp/discussions/12741
- Modified llama-imatrix (not yet merged) used to generate imatrix statistics to guide the TWQ and LWQ process: https://github.com/ggml-org/llama.cpp/pull/12718
reacted to
openfree's
post with ๐ฅ
about 5 hours ago
Post
1865
๐ง ThinkFlow: The Revolutionary Platform That Gives LLMs the Power to Think ๐
Hello AI community! We're excited to introduce you to ThinkFlow, an innovative service that transforms how language models solve problems. ๐
VIDraft/ThinkFlow-llama
โจ What is ThinkFlow?
ThinkFlow is a groundbreaking platform that automatically applies step-by-step reasoning capabilities to existing LLM models without any modifications. It makes complex problem-solving transparent, allowing you to witness the model's thought process in real-time.
๐ Key Features
Reasoning Without Model Modifications: Add step-by-step reasoning while utilizing existing LLMs as they are โ๏ธ
Visualized Thinking Process: See exactly how the model analyzes and solves problems ๐๏ธ
Before & After Comparison: Compare standard responses with reasoning-enhanced outputs in real-time ๐
Improved Accuracy: Deliver more accurate solutions for complex math and logic problems ๐
Educational Value: Teach students systematic approaches to problem-solving ๐จโ๐ซ
User-Friendly Interface: Intuitive and easy-to-use UI for seamless experience ๐ฅ๏ธ
๐ก What Problems Can It Solve?
ThinkFlow is particularly effective for various domains including:
Complex mathematical problems ๐งฎ
Logic puzzles ๐งฉ
Questions requiring multi-step reasoning ๐ค
Scientific analysis challenges ๐ฌ
Complex decision-making processes ๐
๐จโ๐ป Technical Details
ThinkFlow is built on the meta-llama/Llama-3.1-8B-Instruct model and uses carefully designed prompt chains to guide the model through step-by-step thinking. Each reasoning step builds upon the results of previous steps, culminating in a comprehensive final answer.
๐ฌ Join Our Community!
If you have questions or suggestions about ThinkFlow, join our Discord community: https://discord.gg/openfreeai
Let's build better AI reasoning experiences together! ๐ช
#AI #LLM #ReasoningAI #ThinkFlow #HuggingFace #OpenSource #AIEducation
Hello AI community! We're excited to introduce you to ThinkFlow, an innovative service that transforms how language models solve problems. ๐
VIDraft/ThinkFlow-llama
โจ What is ThinkFlow?
ThinkFlow is a groundbreaking platform that automatically applies step-by-step reasoning capabilities to existing LLM models without any modifications. It makes complex problem-solving transparent, allowing you to witness the model's thought process in real-time.
๐ Key Features
Reasoning Without Model Modifications: Add step-by-step reasoning while utilizing existing LLMs as they are โ๏ธ
Visualized Thinking Process: See exactly how the model analyzes and solves problems ๐๏ธ
Before & After Comparison: Compare standard responses with reasoning-enhanced outputs in real-time ๐
Improved Accuracy: Deliver more accurate solutions for complex math and logic problems ๐
Educational Value: Teach students systematic approaches to problem-solving ๐จโ๐ซ
User-Friendly Interface: Intuitive and easy-to-use UI for seamless experience ๐ฅ๏ธ
๐ก What Problems Can It Solve?
ThinkFlow is particularly effective for various domains including:
Complex mathematical problems ๐งฎ
Logic puzzles ๐งฉ
Questions requiring multi-step reasoning ๐ค
Scientific analysis challenges ๐ฌ
Complex decision-making processes ๐
๐จโ๐ป Technical Details
ThinkFlow is built on the meta-llama/Llama-3.1-8B-Instruct model and uses carefully designed prompt chains to guide the model through step-by-step thinking. Each reasoning step builds upon the results of previous steps, culminating in a comprehensive final answer.
๐ฌ Join Our Community!
If you have questions or suggestions about ThinkFlow, join our Discord community: https://discord.gg/openfreeai
Let's build better AI reasoning experiences together! ๐ช
#AI #LLM #ReasoningAI #ThinkFlow #HuggingFace #OpenSource #AIEducation
- 7 replies
updated
a
collection
about 5 hours ago
liked
a
Space
about 5 hours ago
reacted to
Beegbrain's
post with ๐
about 5 hours ago
Post
424
Hello, I've just written an article explaining the project I've made with my team at the Mistral AI Robotic Hackathon one week ago : https://huggingface.co/blog/Beegbrain/guess-who-so100-mistral ; Feel free to take a look, we are open-sourcing the code and begin to launch a community project around the idea, reach out to participate
-
- 1 reply
reacted to
zhiminy's
post with ๐
about 5 hours ago
Post
425
# ๐ SE Arena: Evaluating Foundation Models for Software Engineering
**SE Arena** is the first open-source platform for evaluating foundation models in real-world software engineering workflows.
## What makes it unique?
- **RepoChat**: Automatically injects repository context (issues, commits, PRs) into conversations for more realistic evaluations
- **Multi-round interactions**: Tests models through iterative workflows, not just single prompts
- **Novel metrics**: Includes a "consistency score" that measures model determinism through self-play matches
Try it now: SE-Arena/Software-Engineering-Arena
## Why it matters
Traditional evaluation frameworks don't capture how developers actually use models in their daily work. SE Arena creates a testing environment that mirrors real engineering workflows, helping you choose the right model for your specific software development needs.
From debugging to requirement refinement, see which models truly excel at software engineering tasks!
**SE Arena** is the first open-source platform for evaluating foundation models in real-world software engineering workflows.
## What makes it unique?
- **RepoChat**: Automatically injects repository context (issues, commits, PRs) into conversations for more realistic evaluations
- **Multi-round interactions**: Tests models through iterative workflows, not just single prompts
- **Novel metrics**: Includes a "consistency score" that measures model determinism through self-play matches
Try it now: SE-Arena/Software-Engineering-Arena
## Why it matters
Traditional evaluation frameworks don't capture how developers actually use models in their daily work. SE Arena creates a testing environment that mirrors real engineering workflows, helping you choose the right model for your specific software development needs.
From debugging to requirement refinement, see which models truly excel at software engineering tasks!
updated
a
collection
about 5 hours ago
liked
a
Space
about 5 hours ago
reacted to
MonsterMMORPG's
post with ๐
about 5 hours ago
Post
317
Tencent InstantCharacter 1-Click Installers for Windows, RunPod and Massed Compute, Supports RTX 5000 series as well
Latest installer zip file : https://www.patreon.com/posts/126995127
Use above link to get installer zip file
Official repo : https://github.com/Tencent/InstantCharacter
I have significantly improved the official Repo app
Put FLUX LoRAs into loras folder, it will download 3 LoRAs by default
It will download necessary models into models folder automatically
Lower Character Scale makes it more stylized like 0.6, 0.8 etc
Also official repo Gradio was completely broken, fixed, improved, added new features like automatically save every generated image, number of generations and more
Currently you need min 48GB GPUs, I am trying to make it work with lower VRAM via quantization
Latest installer zip file : https://www.patreon.com/posts/126995127
Use above link to get installer zip file
Official repo : https://github.com/Tencent/InstantCharacter
I have significantly improved the official Repo app
Put FLUX LoRAs into loras folder, it will download 3 LoRAs by default
It will download necessary models into models folder automatically
Lower Character Scale makes it more stylized like 0.6, 0.8 etc
Also official repo Gradio was completely broken, fixed, improved, added new features like automatically save every generated image, number of generations and more
Currently you need min 48GB GPUs, I am trying to make it work with lower VRAM via quantization
-
- 2 replies
reacted to
JLouisBiz's
post with ๐ฅ
about 5 hours ago
Post
420
Back to LLM integration.
ClickDefine.sh -- quickly define or explain anything within your whole desktop environment
You only need to run the model locally, maybe with the **llama.cpp** or **ollama**
- https://github.com/ggml-org/llama.cpp
- https://ollama.com/download
And you get universal explaining tool that works anywhere on your X Org Desktop (on operating systems which are usually Fully Free Software like Debian GNU/Linux)
ClickDefine - Interactive Text Processor Script for Iterative LLM Query Handling:
https://hyperscope.link/9/6/0/9/8/ClickDefine-Interactive-Text-Processor-Script-for-Iterative-LLM-Query-Handling-96098.html
Watch the demonstration here: https://www.youtube.com/watch?v=mQxCYAiReu0&t=2s
ClickDefine.sh -- quickly define or explain anything within your whole desktop environment
You only need to run the model locally, maybe with the **llama.cpp** or **ollama**
- https://github.com/ggml-org/llama.cpp
- https://ollama.com/download
And you get universal explaining tool that works anywhere on your X Org Desktop (on operating systems which are usually Fully Free Software like Debian GNU/Linux)
ClickDefine - Interactive Text Processor Script for Iterative LLM Query Handling:
https://hyperscope.link/9/6/0/9/8/ClickDefine-Interactive-Text-Processor-Script-for-Iterative-LLM-Query-Handling-96098.html
Watch the demonstration here: https://www.youtube.com/watch?v=mQxCYAiReu0&t=2s