SkillFlow:Benchmarking Lifelong Skill Discovery and Evolution for Autonomous Agents
Paper β’ 2604.17308 β’ Published β’ 23
AI & ML interests
ποΈ Creators of models with the most cumulative new downloads each month (users only, no orgs)
Recent Activity
View all activity
Lin-ChenΒ
authored 4
papers about 2 months ago
VideoSeeker: Incentivizing Instance-level Video Understanding via Native Agentic Tool Invocation
Paper β’ 2605.16079 β’ Published β’ 29
SaaSBench: Exploring the Boundaries of Coding Agents in Long-Horizon Enterprise SaaS Engineering
Paper β’ 2605.17526 β’ Published β’ 7
AsyncTool: Evaluating the Asynchronous Function Calling Capability under Multi-Task Scenarios
Paper β’ 2605.27995 β’ Published β’ 16
Lin-ChenΒ
authored 4
papers 2 months ago
ACC: Compiling Agent Trajectories for Long-Context Training
Paper β’ 2605.21850 β’ Published β’ 61
VimRAG: Navigating Massive Visual Context in Retrieval-Augmented Generation via Multimodal Memory Graph
Paper β’ 2602.12735 β’ Published β’ 8
Flow-OPD: On-Policy Distillation for Flow Matching Models
Paper β’ 2605.08063 β’ Published β’ 102
SCOPE: Structured Decomposition and Conditional Skill Orchestration for Complex Image Generation
Paper β’ 2605.08043 β’ Published β’ 10
Post
42701
You may have noticed that my upload of MiMo-V2.5 upload didn't have the author in the model name:
bartowski/MiMo-V2.5-GGUF
Going forward, I plan to upload models from major 1st party developers without the author name attached for cleanliness, I feel it results in a nicer and more expected user experience
I will continue to uploaded fine tunes with that author + "_" appended for clarity, I personally feel it's nice to know at a glance who's tune it is, but it's also for the reason I first started doing it, to avoid it being confused for a new version of the official release
I hope this change makes sense, it seemed most reasonable to me and a poll I did (forever ago, I move slow sometimes) made it seem likely others would find it reasonable as well (feel free to let me know if you disagree, may not change my mind but I do value knowing what others think)
Thanks for downloading :)
bartowski/MiMo-V2.5-GGUF
Going forward, I plan to upload models from major 1st party developers without the author name attached for cleanliness, I feel it results in a nicer and more expected user experience
I will continue to uploaded fine tunes with that author + "_" appended for clarity, I personally feel it's nice to know at a glance who's tune it is, but it's also for the reason I first started doing it, to avoid it being confused for a new version of the official release
I hope this change makes sense, it seemed most reasonable to me and a poll I did (forever ago, I move slow sometimes) made it seem likely others would find it reasonable as well (feel free to let me know if you disagree, may not change my mind but I do value knowing what others think)
Thanks for downloading :)
- 5 replies
MaziyarPanahiΒ
posted an update 4 months ago
Post
4580
Training mRNA Language Models Across 25 Species for $165
We built an end-to-end protein AI pipeline covering structure prediction, sequence design, and codon optimization. After comparing multiple transformer architectures for codon-level language modeling, CodonRoBERTa-large-v2 emerged as the clear winner with a perplexity of 4.10 and a Spearman CAI correlation of 0.40, significantly outperforming ModernBERT. We then scaled to 25 species, trained 4 production models in 55 GPU-hours, and built a species-conditioned system that no other open-source project offers. Complete results, architectural decisions, and runnable code below.
https://huggingface.co/blog/OpenMed/training-mrna-models-25-species
We built an end-to-end protein AI pipeline covering structure prediction, sequence design, and codon optimization. After comparing multiple transformer architectures for codon-level language modeling, CodonRoBERTa-large-v2 emerged as the clear winner with a perplexity of 4.10 and a Spearman CAI correlation of 0.40, significantly outperforming ModernBERT. We then scaled to 25 species, trained 4 production models in 55 GPU-hours, and built a species-conditioned system that no other open-source project offers. Complete results, architectural decisions, and runnable code below.
https://huggingface.co/blog/OpenMed/training-mrna-models-25-species
MaziyarPanahiΒ
posted an update 4 months ago
Post
2396
We annotated 119K medical images with two frontier VLMs (Qwen 3.5, Kimi K2.5), cross-validated at 93% agreement, and produced 110K training records, all for under $500. Fine-tuning 3 small models (2-3B params) improved all benchmarks: best model reaches +15.0% average exact match.
Everything is open-sourced: datasets, adapters, and code.
https://huggingface.co/blog/OpenMed/synthvision
Everything is open-sourced: datasets, adapters, and code.
https://huggingface.co/blog/OpenMed/synthvision
- 2 replies
MaziyarPanahiΒ
posted an update 5 months ago
Post
4965
DNA, mRNA, proteins, AI. I spent the last year going deep into computational biology as an ML engineer. This is Part I of what I found. π§¬
In 2024, AlphaFold won the Nobel Prize in Chemistry.
By 2026, the open-source community had built alternatives that outperform it.
That's the story I find most interesting about protein AI right now. Not just the science (which is incredible), but the speed at which open-source caught up. Multiple teams, independently, reproduced and then exceeded AlphaFold 3's accuracy with permissive licenses. The field went from prediction to generation: we're not just modeling known proteins anymore, we're designing new ones.
I spent months mapping this landscape for ML engineers. What the architectures actually are (spoiler: transformers and diffusion models), which tools to use for what, and which ones you can actually ship commercially.
New post on the Hugging Face blog: https://huggingface.co/blog/MaziyarPanahi/protein-ai-landscape
Hope you all enjoy! π€
In 2024, AlphaFold won the Nobel Prize in Chemistry.
By 2026, the open-source community had built alternatives that outperform it.
That's the story I find most interesting about protein AI right now. Not just the science (which is incredible), but the speed at which open-source caught up. Multiple teams, independently, reproduced and then exceeded AlphaFold 3's accuracy with permissive licenses. The field went from prediction to generation: we're not just modeling known proteins anymore, we're designing new ones.
I spent months mapping this landscape for ML engineers. What the architectures actually are (spoiler: transformers and diffusion models), which tools to use for what, and which ones you can actually ship commercially.
New post on the Hugging Face blog: https://huggingface.co/blog/MaziyarPanahi/protein-ai-landscape
Hope you all enjoy! π€
- 2 replies
MaziyarPanahiΒ
authored a
paper 5 months ago
MaziyarPanahiΒ
posted an update 5 months ago
Post
2554
Announcing: OpenMed Multilingual PII Detection Models
Today I am releasing 105 open-source models for Personally Identifiable Information (PII) detection in French, German, and Italian.
All Apache 2.0 licensed. Free for commercial use. No restrictions.
Performance:
- French: 97.97% F1 (top model)
- German: 97.61% F1 (top model)
- Italian: 97.28% F1 (top model)
All top-10 models per language exceed 96% F1
Coverage:
55+ PII entity types per language
Native ID formats: NSS (French), Sozialversicherungsnummer (German), Codice Fiscale (Italian)
Language-specific address, phone, and name patterns
Training Data:
French: 49,580 samples
German: 42,250 samples
Italian: 40,944 samples
Why Multilingual?
European healthcare operates in European languages. Clinical notes, patient records, and medical documents are generated in French, German, Italian, and other languages.
Effective de-identification requires:
- Native language understanding β not translation
- Local ID format recognition β each country has unique patterns
- Cultural context awareness β names, addresses, and formats vary
- These models deliver production-ready accuracy without requiring data to leave your infrastructure or language.
HIPAA & GDPR Compliance
Built for US and European privacy regulations:
- On-premise deployment: Process data locally with zero external dependencies
- Data sovereignty: No API calls, no cloud services, no cross-border transfers
- Air-gapped capable: Deploy in fully isolated environments if required
- Regulatory-grade accuracy: Supporting Expert Determination standards
- HIPAA and GDPR compliance across languages, without compliance gaps.
Use Cases
- Hospital EHR systems: Automated patient record de-identification
- Clinical research: Multilingual dataset preparation for studies
- Insurance companies: Claims processing across
https://huggingface.co/collections/OpenMed/multilingual-pii-and-de-identification
Today I am releasing 105 open-source models for Personally Identifiable Information (PII) detection in French, German, and Italian.
All Apache 2.0 licensed. Free for commercial use. No restrictions.
Performance:
- French: 97.97% F1 (top model)
- German: 97.61% F1 (top model)
- Italian: 97.28% F1 (top model)
All top-10 models per language exceed 96% F1
Coverage:
55+ PII entity types per language
Native ID formats: NSS (French), Sozialversicherungsnummer (German), Codice Fiscale (Italian)
Language-specific address, phone, and name patterns
Training Data:
French: 49,580 samples
German: 42,250 samples
Italian: 40,944 samples
Why Multilingual?
European healthcare operates in European languages. Clinical notes, patient records, and medical documents are generated in French, German, Italian, and other languages.
Effective de-identification requires:
- Native language understanding β not translation
- Local ID format recognition β each country has unique patterns
- Cultural context awareness β names, addresses, and formats vary
- These models deliver production-ready accuracy without requiring data to leave your infrastructure or language.
HIPAA & GDPR Compliance
Built for US and European privacy regulations:
- On-premise deployment: Process data locally with zero external dependencies
- Data sovereignty: No API calls, no cloud services, no cross-border transfers
- Air-gapped capable: Deploy in fully isolated environments if required
- Regulatory-grade accuracy: Supporting Expert Determination standards
- HIPAA and GDPR compliance across languages, without compliance gaps.
Use Cases
- Hospital EHR systems: Automated patient record de-identification
- Clinical research: Multilingual dataset preparation for studies
- Insurance companies: Claims processing across
https://huggingface.co/collections/OpenMed/multilingual-pii-and-de-identification
- 1 reply
MaziyarPanahiΒ
posted an update 6 months ago
Post
1387
From Golden Gate Bridge to Broken JSON: Why Anthropic's SAE Steering Fails for Structured Output
I ran 6 experiments trying to use Anthropic's SAE steering for JSON generation.
- Base model: 86.8% valid JSON
- Steering only: 24.4%
- Fine-tuned: 96.6%
- FSM constrained: 100%
Steering is for semantics, not syntax.
https://huggingface.co/blog/MaziyarPanahi/sae-steering-json
I ran 6 experiments trying to use Anthropic's SAE steering for JSON generation.
- Base model: 86.8% valid JSON
- Steering only: 24.4%
- Fine-tuned: 96.6%
- FSM constrained: 100%
Steering is for semantics, not syntax.
https://huggingface.co/blog/MaziyarPanahi/sae-steering-json
MaziyarPanahiΒ
posted an update 6 months ago
Post
4143
π¨ Day 8/8: OpenMed Medical Reasoning Dataset Release - THE GRAND FINALE
Today I complete my 8-day release series with Medical-Reasoning-SFT-Mega.
The largest open medical reasoning dataset, combining 7 state-of-the-art AI models with fair distribution deduplication.
THE 7 SOURCE MODELS (Original Sample Counts):
1. Trinity-Mini: 810,284 samples
2. Qwen3-Next-80B: 604,249 samples
3. GPT-OSS-120B: 506,150 samples
4. Nemotron-Nano-30B: 444,544 samples
5. GLM-4.5-Air: 225,179 samples
6. MiniMax-M2.1: 204,773 samples
7. Baichuan-M3-235B: 124,520 samples
TOTAL BEFORE DEDUPLICATION: 2,919,699 samples
TOKEN COUNTS:
- Content tokens: 2.22 Billion
- Reasoning tokens: 1.56 Billion
- Total tokens: 3.78 Billion
- Samples with chain-of-thought: 100%
Quick Start:
All datasets Apache 2.0 licensed. Free for research and commercial use.
Thank you for following OpenMed's release series. I can't wait to see what you build. π₯
OpenMed/Medical-Reasoning-SFT-Mega
OpenMed/Medical-Reasoning-SFT-GPT-OSS-120B-V2
OpenMed/Medical-Reasoning-SFT-Trinity-Mini
OpenMed/Medical-Reasoning-SFT-GLM_4.5_Air
OpenMed/Medical-Reasoning-SFT-MiniMax-M2.1
OpenMed/Medical-Reasoning-SFT-Qwen3-Next-80B
OpenMed/Medical-Reasoning-SFT-Nemotron-Nano-30B
OpenMed/Medical-Reasoning-SFT-Baichuan-M3-235B
https://huggingface.co/collections/OpenMed/medical-datasets
Today I complete my 8-day release series with Medical-Reasoning-SFT-Mega.
The largest open medical reasoning dataset, combining 7 state-of-the-art AI models with fair distribution deduplication.
THE 7 SOURCE MODELS (Original Sample Counts):
1. Trinity-Mini: 810,284 samples
2. Qwen3-Next-80B: 604,249 samples
3. GPT-OSS-120B: 506,150 samples
4. Nemotron-Nano-30B: 444,544 samples
5. GLM-4.5-Air: 225,179 samples
6. MiniMax-M2.1: 204,773 samples
7. Baichuan-M3-235B: 124,520 samples
TOTAL BEFORE DEDUPLICATION: 2,919,699 samples
TOKEN COUNTS:
- Content tokens: 2.22 Billion
- Reasoning tokens: 1.56 Billion
- Total tokens: 3.78 Billion
- Samples with chain-of-thought: 100%
Quick Start:
from datasets import load_dataset
ds = load_dataset("OpenMed/Medical-Reasoning-SFT-Mega")All datasets Apache 2.0 licensed. Free for research and commercial use.
Thank you for following OpenMed's release series. I can't wait to see what you build. π₯
OpenMed/Medical-Reasoning-SFT-Mega
OpenMed/Medical-Reasoning-SFT-GPT-OSS-120B-V2
OpenMed/Medical-Reasoning-SFT-Trinity-Mini
OpenMed/Medical-Reasoning-SFT-GLM_4.5_Air
OpenMed/Medical-Reasoning-SFT-MiniMax-M2.1
OpenMed/Medical-Reasoning-SFT-Qwen3-Next-80B
OpenMed/Medical-Reasoning-SFT-Nemotron-Nano-30B
OpenMed/Medical-Reasoning-SFT-Baichuan-M3-235B
https://huggingface.co/collections/OpenMed/medical-datasets
- 6 replies
Lin-ChenΒ
authored 2
papers 6 months ago
Lin-ChenΒ
authored a
paper 7 months ago
Lin-ChenΒ
submitted a
paper to Daily Papers 7 months ago