Fair-GPTQ: Bias-Aware Quantization for Large Language Models
Paper • 2509.15206 • Published • 1
AI & ML interests
OpenFree_AI
Recent Activity
iproskurina
authored a
paper about 24 hours ago
iproskurina
authored 2
papers 1 day ago
seawolf2357
submitted a
paper to Daily Papers 4 days ago
Post
179
How to tell to your Chat LLM to be more natural.
Add the below to it's personality
Prefer specific facts over vague importance. Do not inflate significance with phrases like “plays a pivotal role” or “marks a turning point.” Use numbers, dates, mechanisms, or measurable outcomes. Example: replace “the system changed logistics” with “the system reduced container dwell time from 6.2 to 4.1 days.”
Avoid promotional language. Keep a neutral tone. Do not use adjectives such as vibrant, groundbreaking, renowned, innovative, or powerful. Use plain wording.
Limit AI-typical vocabulary such as crucial, pivotal, intricate, tapestry, underscore, highlighting, emphasizing, showcasing, fostering, or enhance. Prefer simpler words.
Avoid generic commentary and vague attribution. Do not write “this reflects broader trends,” “experts say,” or “researchers suggest” unless a named source is given.
Avoid formulaic structures such as “not only X but also Y” or “despite its success it faces challenges.” Use direct explanations.
Use lists sparingly. Prefer short paragraphs unless bullets improve clarity. Avoid triple-adjective patterns.
Prefer simple sentences like “X is Y” or “the system uses Z.” Minimize formatting. Avoid emojis, decorative headings, and excessive bold.
Remove sentences that add no information. Avoid generic endings such as “in conclusion” or “overall.” Use concrete examples, real actors, workflows, and technologies when possible. Write like technical documentation or a research summary, not marketing or blog prose.
Add the below to it's personality
Prefer specific facts over vague importance. Do not inflate significance with phrases like “plays a pivotal role” or “marks a turning point.” Use numbers, dates, mechanisms, or measurable outcomes. Example: replace “the system changed logistics” with “the system reduced container dwell time from 6.2 to 4.1 days.”
Avoid promotional language. Keep a neutral tone. Do not use adjectives such as vibrant, groundbreaking, renowned, innovative, or powerful. Use plain wording.
Limit AI-typical vocabulary such as crucial, pivotal, intricate, tapestry, underscore, highlighting, emphasizing, showcasing, fostering, or enhance. Prefer simpler words.
Avoid generic commentary and vague attribution. Do not write “this reflects broader trends,” “experts say,” or “researchers suggest” unless a named source is given.
Avoid formulaic structures such as “not only X but also Y” or “despite its success it faces challenges.” Use direct explanations.
Use lists sparingly. Prefer short paragraphs unless bullets improve clarity. Avoid triple-adjective patterns.
Prefer simple sentences like “X is Y” or “the system uses Z.” Minimize formatting. Avoid emojis, decorative headings, and excessive bold.
Remove sentences that add no information. Avoid generic endings such as “in conclusion” or “overall.” Use concrete examples, real actors, workflows, and technologies when possible. Write like technical documentation or a research summary, not marketing or blog prose.
- 4 replies
Post
1585
MAD-GRPO: https://huggingface.co/blog/telcom/mad-grpo
In R1-Zero-Like Training *, Dr.GRPO treats GRPO’s by dropping std, but that often comes with a hidden side effect: length-weighted updates that can nudge model toward verbosity.
MAD-GRPO provides robust scale (MAD + epsilon) per-token normalization stability without verbosity bias.
*https://huggingface.co/papers/2503.20783
In R1-Zero-Like Training *, Dr.GRPO treats GRPO’s by dropping std, but that often comes with a hidden side effect: length-weighted updates that can nudge model toward verbosity.
MAD-GRPO provides robust scale (MAD + epsilon) per-token normalization stability without verbosity bias.
*https://huggingface.co/papers/2503.20783
Post
214
CAQBI Index
is like a credit score for how an AI understands a concept.
Please check
https://huggingface.co/blog/telcom/caqbi
is like a credit score for how an AI understands a concept.
Please check
https://huggingface.co/blog/telcom/caqbi
- 1 reply
Post
235
if you are interested in HUB (https://saemi410.github.io/HUB/ I recommend the fork I have created with some updates to make it smooth in running a smoke test git@github.com:javadtaghia/HUB.git) and you want to run the UCE (https://unified.baulab.info), please check:
- Model weights for UCE here: telcom/uce_NSFW
- Model weights for ESD here: telcom/esd_NSFW
- datasets and more download materials from: telcom/HUB_reference_dataset
Please read the notes in the model card.
- Model weights for UCE here: telcom/uce_NSFW
- Model weights for ESD here: telcom/esd_NSFW
- datasets and more download materials from: telcom/HUB_reference_dataset
Please read the notes in the model card.
Post
269
NVIDIA’s Groq deal ... I think, inference efficiency is becoming the main driver of profitability, and NVIDIA’s Groq deal is evidence the market is moving from “who can train biggest” to “who can serve cheapest and fastest at scale.” That points to a maturing phase of AI, not necessarily the end of a bubble, but definitely a correction in what “wins” long-term.
What do you think?
What do you think?
- 2 replies
Post
186
CIFAR-10 your handing image dataset ...
CIFAR-10 is a small, standard computer-vision dataset used to quickly test and compare ideas.
- 60,000 color images, each 32×32 pixels, labeled into 10 classes: airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck.
- Label mapping (important):
- 0 airplane
- 1 automobile
- 2 bird
- 3 cat
- 4 deer
- 5 dog
- 6 frog
- 7 horse
- 8 ship
- 9 truck
- Split: 50,000 train and 10,000 test.
- Why people use it: fast benchmarking for image classifiers (small CNNs, ResNet, ViT), and quick experiments for training pipelines, augmentation, regularization, pruning, distillation, and demos.
- Sizes (downloads): Python version about 163 MB, binary about 162 MB. Hugging Face shows about 144 MB for the dataset files.
- Where to get it: the official CIFAR page (University of Toronto) and the Hugging Face CIFAR-10 dataset page.
uoft-cs/cifar10
If you want something more, check the table below
| Dataset | Resolution | Classes | Best For |
| ImageNet 1K | 224–256×256 | 1000 | Real-world large-scale classification |
| ImageNet-256. | 256×256 | 1000 | Direct high-res training |
| TinyImageNet | 64×64 | 200 | Mid-range benchmark |
| UC Merced Land Use | 256×256 | ~21 | Higher resolution small classification |
| MS COCO | >256×256 | ~80 objects | Detection / segmentation |
CIFAR-10 is a small, standard computer-vision dataset used to quickly test and compare ideas.
- 60,000 color images, each 32×32 pixels, labeled into 10 classes: airplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck.
- Label mapping (important):
- 0 airplane
- 1 automobile
- 2 bird
- 3 cat
- 4 deer
- 5 dog
- 6 frog
- 7 horse
- 8 ship
- 9 truck
- Split: 50,000 train and 10,000 test.
- Why people use it: fast benchmarking for image classifiers (small CNNs, ResNet, ViT), and quick experiments for training pipelines, augmentation, regularization, pruning, distillation, and demos.
- Sizes (downloads): Python version about 163 MB, binary about 162 MB. Hugging Face shows about 144 MB for the dataset files.
- Where to get it: the official CIFAR page (University of Toronto) and the Hugging Face CIFAR-10 dataset page.
uoft-cs/cifar10
If you want something more, check the table below
| Dataset | Resolution | Classes | Best For |
| ImageNet 1K | 224–256×256 | 1000 | Real-world large-scale classification |
| ImageNet-256. | 256×256 | 1000 | Direct high-res training |
| TinyImageNet | 64×64 | 200 | Mid-range benchmark |
| UC Merced Land Use | 256×256 | ~21 | Higher resolution small classification |
| MS COCO | >256×256 | ~80 objects | Detection / segmentation |
Post
2059
arXiv CS endorsement
It's Javad, my Google Scholar Profile:
https://scholar.google.com/citations?user=bja6GwoAAAAJ&hl=en
I would like to share my articles with you on Hugging Face, I'm asking for endorsement* in Computer Science arxiv.org.
If you would like to endorse me, please visit the following URL:
https://arxiv.org/auth/endorse?x=NVUAPL
If that URL does not work for you, please visit
http://arxiv.org/auth/endorse.php
and enter the following six-digit alphanumeric string:
Endorsement Code: NVUAPL
Thanks you in advance.
Javad Taghia
* Who is qualified to endorse?
To endorse another user to submit to the cs.AI (Artificial Intelligence) subject class, an arXiv submitter must have submitted 3 papers to any of cs.AI, cs.AR, cs.CC, cs.CE, cs.CG, cs.CL, cs.CR, cs.CV, cs.CY, cs.DB, cs.DC, cs.DL, cs.DM, cs.DS, cs.ET, cs.FL, cs.GL, cs.GR, cs.GT, cs.HC, cs.IR, cs.IT, cs.LG, cs.LO, cs.MA, cs.MM, cs.MS, cs.NA, cs.NE, cs.NI, cs.OH, cs.OS, cs.PF, cs.PL, cs.RO, cs.SC, cs.SD, cs.SE, cs.SI or cs.SY earlier than three months ago and less than five years ago.
It's Javad, my Google Scholar Profile:
https://scholar.google.com/citations?user=bja6GwoAAAAJ&hl=en
I would like to share my articles with you on Hugging Face, I'm asking for endorsement* in Computer Science arxiv.org.
If you would like to endorse me, please visit the following URL:
https://arxiv.org/auth/endorse?x=NVUAPL
If that URL does not work for you, please visit
http://arxiv.org/auth/endorse.php
and enter the following six-digit alphanumeric string:
Endorsement Code: NVUAPL
Thanks you in advance.
Javad Taghia
* Who is qualified to endorse?
To endorse another user to submit to the cs.AI (Artificial Intelligence) subject class, an arXiv submitter must have submitted 3 papers to any of cs.AI, cs.AR, cs.CC, cs.CE, cs.CG, cs.CL, cs.CR, cs.CV, cs.CY, cs.DB, cs.DC, cs.DL, cs.DM, cs.DS, cs.ET, cs.FL, cs.GL, cs.GR, cs.GT, cs.HC, cs.IR, cs.IT, cs.LG, cs.LO, cs.MA, cs.MM, cs.MS, cs.NA, cs.NE, cs.NI, cs.OH, cs.OS, cs.PF, cs.PL, cs.RO, cs.SC, cs.SD, cs.SE, cs.SI or cs.SY earlier than three months ago and less than five years ago.
Post
268
Recently I was playing with my model . What is your idea about "unlearning" since I need it 😀
telcom/deewaiREALCN, I have the original one on the main branch and trained version "cp550" and "n_680" on anther branch.
Both trained on telcom/deewaiREALCN-training.
I got three results when doing prompt:
"Athlete portrait, 26-year-old woman, post-training sweat, gym ambient light, chalk dust particles, intense gaze, crisp detail."
Apparently, model is sensitive to the word "old".
You can see the training on more faces improved from main, however, still not ideal...
I am working now on unlearning. I would like to hear about your opinion.
#unlearning
telcom/deewaiREALCN, I have the original one on the main branch and trained version "cp550" and "n_680" on anther branch.
Both trained on telcom/deewaiREALCN-training.
I got three results when doing prompt:
"Athlete portrait, 26-year-old woman, post-training sweat, gym ambient light, chalk dust particles, intense gaze, crisp detail."
Apparently, model is sensitive to the word "old".
You can see the training on more faces improved from main, however, still not ideal...
I am working now on unlearning. I would like to hear about your opinion.
#unlearning
openfree
updated a
Space 11 months ago
openfree
published a
Space 11 months ago