z-lab/Qwen3.5-9B-PARO
Image-Text-to-Text β’ 3B β’ Updated
β’ 176 β’ 23
Jack Voide
Mindweller
AI & ML interests
None yet
Recent Activity
liked
a model about 6 hours ago
z-lab/Qwen3.5-9B-PARO liked
a model 3 days ago
EganAI/qwen3.5-9b-terminal-merge liked
a model 5 days ago
kiriyk/GigaAM-v3-onnx-rnnt-e2e Organizations
None yet
liked a model about 6 hours ago
liked a model 3 days ago
liked a model 5 days ago
reacted to DavidAU's post with π 11 days ago
Post
4931
Gemma 3 27B - The record breaker (Heretic'ed (uncensored) ; then training on Unsloth):
arc_challenge,arc_easy,boolq,hellaswag,openbookqa,piqa, winogrande
0.661 ,0.816 ,0.878,0.763 ,0.464 ,0.808 ,0.762
For comparison:
Qwen3.5-27B-Text
qx86-hi 0.443,0.498,0.857,0.701,0.372,0.770,0.752
Trained on a HERETIC uncensored base too ;
DavidAU/Gemma3-27B-it-vl-Polaris-HI16-Heretic-Uncensored-INSTRUCT
arc_challenge,arc_easy,boolq,hellaswag,openbookqa,piqa, winogrande
0.661 ,0.816 ,0.878,0.763 ,0.464 ,0.808 ,0.762
For comparison:
Qwen3.5-27B-Text
qx86-hi 0.443,0.498,0.857,0.701,0.372,0.770,0.752
Trained on a HERETIC uncensored base too ;
DavidAU/Gemma3-27B-it-vl-Polaris-HI16-Heretic-Uncensored-INSTRUCT
liked a model 11 days ago
reacted to GVA21q2's post with π 11 days ago
Post
1516
# Ο.Guy.AI β AI-Powered Neuropedagogy Math Lessons
Students with math anxiety, ADHD, dyslexia, or low working memory need different learning experiences β but teachers can't create individualized materials for every student.
**Ο.Guy.AI** generates interactive HTML math lessons adapted to 7 cognitive profiles, using a multi-agent AI pipeline:
1. **Neuro-Interpreter** β enriches prompts with profile-specific adaptations
2. **Creative Agent** β generates a 12-slide lesson with SVG visualizations
3. **Quality Control** β validates against 8 neuropedagogy principles
Each lesson is a standalone HTML file with inline CSS/JS/SVG β works offline, no dependencies.
## The Model
Fine-tuned **Qwen2.5-7B-Instruct** with LoRA on 313 curated Hebrew math lessons.
- Model: [GVA21q2/piguyai-lessons-v2-enhanced](https://huggingface.co/GVA21q2/piguyai-lessons-v2-enhanced)
- Dataset: [GVA21q2/pi-guy-ai-lessons](https://huggingface.co/datasets/GVA21q2/pi-guy-ai-lessons)
- Demo: [GVA21q2/pi-guy-ai-demo]( GVA21q2/pi-guy-ai-demo)
- Web app: [gva21q2.github.io/pi.guy.ai](https://gva21q2.github.io/pi.guy.ai/)
7 profiles: math anxiety, ADHD, dyslexia, dysgraphia, low working memory, visual processing, weak inhibition.
Built by [Guy Assal](https://www.guyassal.education)
Students with math anxiety, ADHD, dyslexia, or low working memory need different learning experiences β but teachers can't create individualized materials for every student.
**Ο.Guy.AI** generates interactive HTML math lessons adapted to 7 cognitive profiles, using a multi-agent AI pipeline:
1. **Neuro-Interpreter** β enriches prompts with profile-specific adaptations
2. **Creative Agent** β generates a 12-slide lesson with SVG visualizations
3. **Quality Control** β validates against 8 neuropedagogy principles
Each lesson is a standalone HTML file with inline CSS/JS/SVG β works offline, no dependencies.
## The Model
Fine-tuned **Qwen2.5-7B-Instruct** with LoRA on 313 curated Hebrew math lessons.
- Model: [GVA21q2/piguyai-lessons-v2-enhanced](https://huggingface.co/GVA21q2/piguyai-lessons-v2-enhanced)
- Dataset: [GVA21q2/pi-guy-ai-lessons](https://huggingface.co/datasets/GVA21q2/pi-guy-ai-lessons)
- Demo: [GVA21q2/pi-guy-ai-demo]( GVA21q2/pi-guy-ai-demo)
- Web app: [gva21q2.github.io/pi.guy.ai](https://gva21q2.github.io/pi.guy.ai/)
7 profiles: math anxiety, ADHD, dyslexia, dysgraphia, low working memory, visual processing, weak inhibition.
Built by [Guy Assal](https://www.guyassal.education)
- 1 reply
liked a Space 12 days ago
liked a model 16 days ago
liked a model 23 days ago
liked a model 24 days ago
liked a Space 24 days ago
liked a model 25 days ago
reacted to mrs83's post with π 25 days ago
Post
2342
In 2017, my RNNs were babbling. Today, they are hallucinating beautifully.
10 years ago, getting an LSTM to output coherent English was a struggle.
10 years later, after a "cure" based on FineWeb-EDU and a custom synthetic mix for causal conversation, the results are fascinating.
We trained this on ~10B tokens on a single AMD GPU (ROCm). It is not a Transformer: Echo-DSRN (400M) is a novel recurrent architecture inspired by Hymba, RWKV, and xLSTM, designed to challenge the "Attention is All You Need" monopoly on the Edge.
The ambitious goal is to build a small instruct model with RAG and tool usage capabilities ( ethicalabs/Kurtis-EON1)
π The Benchmarks (Size: 400M)
For a model this size (trained on <10B tokens), the specialized performance is surprising:
*SciQ*: 73.8% π¦ (This rivals billion-parameter models in pure fact retrieval).
*PIQA*: 62.3% (Solid physical intuition for a sub-1B model).
The Reality Check:
HellaSwag (29.3%) and Winogrande (50.2%) show the limits of 400M parameters and 10B tokens training.
We are hitting the "Reasoning Wall" which confirms we need to scale to (hopefully) unlock deeper common sense. As you can see in the visualization (to be released soon on HF), the FineWeb-EDU bias is strong. The model is convinced it is in a classroom ("In this course, we explore...").
The Instruct Model is not ready yet and we are currently using curriculum learning to test model plasticity.
Source code and weights will not be released yet. This is not a fork or a fine-tune: the base model is built in-house at https://www.ethicalabs.ai/, with novel components that do not exist in current open libraries.
π€ Call for Collaboration: I am looking for Peer Reviewers interested in recurrent/hybrid architectures. If you want to explore what lies beyond Transformers, letβs connect!
Training diary: ethicalabs/Kurtis-EON1
10 years ago, getting an LSTM to output coherent English was a struggle.
10 years later, after a "cure" based on FineWeb-EDU and a custom synthetic mix for causal conversation, the results are fascinating.
We trained this on ~10B tokens on a single AMD GPU (ROCm). It is not a Transformer: Echo-DSRN (400M) is a novel recurrent architecture inspired by Hymba, RWKV, and xLSTM, designed to challenge the "Attention is All You Need" monopoly on the Edge.
The ambitious goal is to build a small instruct model with RAG and tool usage capabilities ( ethicalabs/Kurtis-EON1)
π The Benchmarks (Size: 400M)
For a model this size (trained on <10B tokens), the specialized performance is surprising:
*SciQ*: 73.8% π¦ (This rivals billion-parameter models in pure fact retrieval).
*PIQA*: 62.3% (Solid physical intuition for a sub-1B model).
The Reality Check:
HellaSwag (29.3%) and Winogrande (50.2%) show the limits of 400M parameters and 10B tokens training.
We are hitting the "Reasoning Wall" which confirms we need to scale to (hopefully) unlock deeper common sense. As you can see in the visualization (to be released soon on HF), the FineWeb-EDU bias is strong. The model is convinced it is in a classroom ("In this course, we explore...").
The Instruct Model is not ready yet and we are currently using curriculum learning to test model plasticity.
Source code and weights will not be released yet. This is not a fork or a fine-tune: the base model is built in-house at https://www.ethicalabs.ai/, with novel components that do not exist in current open libraries.
π€ Call for Collaboration: I am looking for Peer Reviewers interested in recurrent/hybrid architectures. If you want to explore what lies beyond Transformers, letβs connect!
Training diary: ethicalabs/Kurtis-EON1
- 6 replies
liked a model 26 days ago
liked a Space 26 days ago
liked a model 26 days ago
liked a model 27 days ago
liked a model 29 days ago
liked 2 models about 1 month ago
