Vortex5/Darklit-Maiden-12B
Text Generation β’ 12B β’ Updated β’ 221 β’ 5
John Smith PRO
John6666
AI & ML interests
None yet
Recent Activity
liked a model about 3 hours ago
Vortex5/Darklit-Maiden-12B liked a model about 3 hours ago
mradermacher/Darklit-Maiden-12B-i1-GGUF liked a model about 3 hours ago
mradermacher/Darklit-Maiden-12B-GGUFOrganizations
liked 3
models about 3 hours ago
reactedto branikita's post with π about 8 hours ago
Post
1082
Our paper is now published in HardwareX!
"An open-source test stand for backlash measurement in low-cost UART servo motors" presents a ~$100 automated platform for measuring backlash in compact serial bus servos β no dial indicators, no probe contact force, fully scripted and repeatable.
All CAD files, control software, and analysis tools are open source.
Read the full article: https://www.hardware-x.com/article/S2468-0672(26)00027-1/fulltext
Appearing in HardwareX Volume 26 (June 2026): https://www.sciencedirect.com/journal/hardwarex/vol/26/suppl/Cf
"An open-source test stand for backlash measurement in low-cost UART servo motors" presents a ~$100 automated platform for measuring backlash in compact serial bus servos β no dial indicators, no probe contact force, fully scripted and repeatable.
All CAD files, control software, and analysis tools are open source.
Read the full article: https://www.hardware-x.com/article/S2468-0672(26)00027-1/fulltext
Appearing in HardwareX Volume 26 (June 2026): https://www.sciencedirect.com/journal/hardwarex/vol/26/suppl/Cf
reactedto Aurelien-Morgan's post with π about 8 hours ago
reactedto sequelbox's post with π about 8 hours ago
Post
139
Multiple new releases for Gemma 4!
For Gemma 4 31B: Guardpoint, our medical reasoning model, trained on medical knowledge, management, diagnosis, and tasks:
- Structured medical reasoning responses are efficient and informative, cutting token costs for faster inference!
- Wide-ranging knowledge base: trained on a wide variety of medical disciplines, patient types, and query structures!
- High quality medical responses emphasize performance, brevity, specificity, statistical rationality, and openness.
Get Guardpoint for Gemma 4: ValiantLabs/gemma-4-31B-it-Guardpoint
For Gemma 4 E4B and E2B: Shining Valiant 3, our science-reasoning model!
- Science-reasoning: physics, biology, chemistry, compsci, astronomy, Earth science, and information theory.
- AI to build AI: high-quality reasoning performance on AI, MLOps, math and CUDA, complex adaptive and agentic systems, cognition, logic, linguistics, simulation, knowledge management, and more!
- Supplemented creative reasoning and general chat performance.
Get the new SV3 models:
E4B: ValiantLabs/gemma-4-E4B-it-ShiningValiant3
E2B: ValiantLabs/gemma-4-E2B-it-ShiningValiant3
We're working on several things - most excitingly, we've officially started the dataset curation process for Esper 4! We're focused on enhanced agentic capability and higher-dififculty, higher-value tasks this time, very excited to bring this to everyone when we can :)
Help support our releases, donations used for our experimental models and datasets: sequelbox/SupportOpenSource
Fight for open source with us!
for love and friendship,
allegra
For Gemma 4 31B: Guardpoint, our medical reasoning model, trained on medical knowledge, management, diagnosis, and tasks:
- Structured medical reasoning responses are efficient and informative, cutting token costs for faster inference!
- Wide-ranging knowledge base: trained on a wide variety of medical disciplines, patient types, and query structures!
- High quality medical responses emphasize performance, brevity, specificity, statistical rationality, and openness.
Get Guardpoint for Gemma 4: ValiantLabs/gemma-4-31B-it-Guardpoint
For Gemma 4 E4B and E2B: Shining Valiant 3, our science-reasoning model!
- Science-reasoning: physics, biology, chemistry, compsci, astronomy, Earth science, and information theory.
- AI to build AI: high-quality reasoning performance on AI, MLOps, math and CUDA, complex adaptive and agentic systems, cognition, logic, linguistics, simulation, knowledge management, and more!
- Supplemented creative reasoning and general chat performance.
Get the new SV3 models:
E4B: ValiantLabs/gemma-4-E4B-it-ShiningValiant3
E2B: ValiantLabs/gemma-4-E2B-it-ShiningValiant3
We're working on several things - most excitingly, we've officially started the dataset curation process for Esper 4! We're focused on enhanced agentic capability and higher-dififculty, higher-value tasks this time, very excited to bring this to everyone when we can :)
Help support our releases, donations used for our experimental models and datasets: sequelbox/SupportOpenSource
Fight for open source with us!
for love and friendship,
allegra
reactedto AGIreflex's post with π about 8 hours ago
Post
123
Building OmniBench Aegis Env: A Reproducible Multi-Domain OpenEnv for Agent Evaluation
As AI systems move from static prediction toward more general, agentic behavior, the quality of the environment becomes just as important as the quality of the agent. A good environment is not just a wrapper around tasks. It is the place where agents explore, act, fail, recover, and improve. For that reason, I built OmniBench Aegis Env, a reproducible, multi-domain OpenEnv environment inside AegisForge_agent, with a strong focus on evaluation clarity, controlled variation, and practical agent benchmarking. The OpenEnv Challenge explicitly asks for an HF Hub environment, public training artifacts, and a Hugging Face blog, and also states that judging will focus heavily on the submission blog itself.
Why I built this environment
A recurring problem in agent repos is that too many responsibilities get mixed into the same layer. Environment state, task rules, reward shaping, evaluation logic, and even agent behavior often end up entangled. That makes it difficult to answer simple but important questions:
Did the agent actually improve?
Did the environment change?
Did the harness become more forgiving?
Did a success come from a real fix or from a hidden shortcut?
OmniBench Aegis Env was designed to separate those concerns cleanly. The environment manages state, actions, observations, and reward. The agent remains responsible for planning, routing, policies, and higher-level decision making. Evaluation scripts stay external and reproducible. Curriculum, transfer, and variant generation are explicit artifacts rather than hidden internal assumptions. That separation is one of the core ideas behind the project.
What OmniBench Aegis Env is
As AI systems move from static prediction toward more general, agentic behavior, the quality of the environment becomes just as important as the quality of the agent. A good environment is not just a wrapper around tasks. It is the place where agents explore, act, fail, recover, and improve. For that reason, I built OmniBench Aegis Env, a reproducible, multi-domain OpenEnv environment inside AegisForge_agent, with a strong focus on evaluation clarity, controlled variation, and practical agent benchmarking. The OpenEnv Challenge explicitly asks for an HF Hub environment, public training artifacts, and a Hugging Face blog, and also states that judging will focus heavily on the submission blog itself.
Why I built this environment
A recurring problem in agent repos is that too many responsibilities get mixed into the same layer. Environment state, task rules, reward shaping, evaluation logic, and even agent behavior often end up entangled. That makes it difficult to answer simple but important questions:
Did the agent actually improve?
Did the environment change?
Did the harness become more forgiving?
Did a success come from a real fix or from a hidden shortcut?
OmniBench Aegis Env was designed to separate those concerns cleanly. The environment manages state, actions, observations, and reward. The agent remains responsible for planning, routing, policies, and higher-level decision making. Evaluation scripts stay external and reproducible. Curriculum, transfer, and variant generation are explicit artifacts rather than hidden internal assumptions. That separation is one of the core ideas behind the project.
What OmniBench Aegis Env is
reactedto nightmedia's post with π about 8 hours ago
Post
1239
Updated gemma-4-E4B-it metrics
I noticed the chat template got updated, and tried it on the E4B, with surprising results in stabilizing the brainwave.
Old numbers
I will re-do all baselines soon based on the new template. It is completely expected that the model behavior will change as a result.
Here are the effects of the new template on few known distills from DavidAU
gemma-4-E4B-it-The-DECKARD-Expresso-Universe-HERETIC-UNCENSORED
gemma-4-E4B-it-GLM-4.7-Flash-HERETIC-UNCENSORED-Thinking
gemma-4-E4B-it-Claude-Opus-4.5-HERETIC-UNCENSORED-Thinking
I noticed the chat template got updated, and tried it on the E4B, with surprising results in stabilizing the brainwave.
quant arc arc/e boolq hswag obkqa piqa wino
mxfp8 0.480,0.656,0.797,0.608,0.400,0.755,0.665
mxfp4 0.455,0.607,0.851,0.585,0.402,0.744,0.651
Quant Perplexity Peak Memory Tokens/sec
mxfp8 35.937 Β± 0.525 14.80 GB 1153
mxfp4 36.746 Β± 0.534 11.06 GB 1030Old numbers
quant arc arc/e boolq hswag obkqa piqa wino
mxfp8 0.404,0.489,0.825,0.586,0.392,0.734,0.661
mxfp4 0.414,0.508,0.854,0.562,0.378,0.717,0.645
Quant Perplexity Peak Memory Tokens/sec
mxfp8 34.652 Β± 0.502 14.80 GB 1146
mxfp4 35.203 Β± 0.506 11.06 GB 1200I will re-do all baselines soon based on the new template. It is completely expected that the model behavior will change as a result.
Here are the effects of the new template on few known distills from DavidAU
gemma-4-E4B-it-The-DECKARD-Expresso-Universe-HERETIC-UNCENSORED
quant arc arc/e boolq hswag obkqa piqa wino
New template
mxfp8 0.518,0.709,0.755,0.657,0.418,0.759,0.626
mxfp4 0.485,0.682,0.792,0.641,0.432,0.746,0.635
Old template
mxfp8 0.506,0.697,0.754,0.661,0.416,0.757,0.627
mxfp4 0.487,0.670,0.792,0.644,0.430,0.748,0.624gemma-4-E4B-it-GLM-4.7-Flash-HERETIC-UNCENSORED-Thinking
mxfp8 0.461,0.599,0.779,0.630,0.406,0.766,0.629
Old template
mxfp8 0.456,0.580,0.786,0.629,0.410,0.764,0.633gemma-4-E4B-it-Claude-Opus-4.5-HERETIC-UNCENSORED-Thinking
mxfp8 0.509,0.705,0.806,0.646,0.416,0.773,0.650
Old template
mxfp8 0.502,0.692,0.809,0.650,0.420,0.771,0.651
- 2 replies
reactedto victor's post with π€π₯ about 8 hours ago
Post
3195
Want to share my enthusiasm for zai-org/GLM-5.1 here too π₯
I think we have it: our open source Claude Code = GLM-5.1 + Pi (https://pi.dev/) - Built a Three.js racing game to eval and it's extremely impressive. Thoughts:
- One-shot car physics with real drift mechanics (this is hard)
- My fav part: Awesome at self iterating (with no vision!) created 20+ Bun.WebView debugging tools to drive the car programmatically and read game state. Proved a winding bug with vector math without ever seeing the screen
- 531-line racing AI in a single write: 4 personalities, curvature map, racing lines, tactical drifting. Built telemetry tools to compare player vs AI speed curves and data-tuned parameters
- All assets from scratch: 3D models, procedural textures, sky shader, engine sounds, spatial AI audio!
- Can do hard math: proved road normals pointed DOWN via vector cross products, computed track curvature normalized by arc length to tune AI cornering speed
You are going to hear about this model a lot in the next months - open source let's go - and thanks z-aiππ
I think we have it: our open source Claude Code = GLM-5.1 + Pi (https://pi.dev/) - Built a Three.js racing game to eval and it's extremely impressive. Thoughts:
- One-shot car physics with real drift mechanics (this is hard)
- My fav part: Awesome at self iterating (with no vision!) created 20+ Bun.WebView debugging tools to drive the car programmatically and read game state. Proved a winding bug with vector math without ever seeing the screen
- 531-line racing AI in a single write: 4 personalities, curvature map, racing lines, tactical drifting. Built telemetry tools to compare player vs AI speed curves and data-tuned parameters
- All assets from scratch: 3D models, procedural textures, sky shader, engine sounds, spatial AI audio!
- Can do hard math: proved road normals pointed DOWN via vector cross products, computed track curvature normalized by arc length to tune AI cornering speed
You are going to hear about this model a lot in the next months - open source let's go - and thanks z-aiππ
- 3 replies
reactedto kanaria007's post with π§ about 8 hours ago
Post
106
β
Article highlight: *Operational Rights as Autonomy Envelopes* (art-60-062, v0.1)
TL;DR:
This article turns βAI rightsβ into a concrete runtime object.
Instead of treating rights as a moral trophy, it models them as *bounded autonomy envelopes*: explicit effect permissions with scope, budgets, gates, rollback requirements, and auditability. The point is not to romanticize autonomy, but to make local discretion governable.
Read:
kanaria007/agi-structural-intelligence-protocols
Why it matters:
β’ makes βAI rightsβ legible as systems engineering rather than sentiment
β’ defines a practical object for local discretion under latency, partitions, or mission distance
β’ shows that bounded permission is not the same thing as trust
β’ treats envelope expansion itself as a high-stakes governance action
Whatβs inside:
β’ βrightsβ as *runtime budgets for effectful autonomy*
β’ *autonomy envelopes* as typed, scoped, rate-limited, gated, rollback-bounded, auditable, revisable objects
β’ the rule that loosening an envelope must go through evaluation / approval / audit
β’ a concrete deep-space style example of local operational discretion
β’ a migration path from *LLM proposal engines* to governed autonomous SI nodes
Key idea:
Do not grant autonomy as a blank check.
Grant it as a bounded envelope:
*what effects are allowed, in what scope, at what rate, under what gates, with what rollback, and under what audit trail?*
TL;DR:
This article turns βAI rightsβ into a concrete runtime object.
Instead of treating rights as a moral trophy, it models them as *bounded autonomy envelopes*: explicit effect permissions with scope, budgets, gates, rollback requirements, and auditability. The point is not to romanticize autonomy, but to make local discretion governable.
Read:
kanaria007/agi-structural-intelligence-protocols
Why it matters:
β’ makes βAI rightsβ legible as systems engineering rather than sentiment
β’ defines a practical object for local discretion under latency, partitions, or mission distance
β’ shows that bounded permission is not the same thing as trust
β’ treats envelope expansion itself as a high-stakes governance action
Whatβs inside:
β’ βrightsβ as *runtime budgets for effectful autonomy*
β’ *autonomy envelopes* as typed, scoped, rate-limited, gated, rollback-bounded, auditable, revisable objects
β’ the rule that loosening an envelope must go through evaluation / approval / audit
β’ a concrete deep-space style example of local operational discretion
β’ a migration path from *LLM proposal engines* to governed autonomous SI nodes
Key idea:
Do not grant autonomy as a blank check.
Grant it as a bounded envelope:
*what effects are allowed, in what scope, at what rate, under what gates, with what rollback, and under what audit trail?*
updated a
collection about 8 hours ago
liked a
Space about 8 hours ago
reactedto prithivMLmods's post with π€π₯ about 8 hours ago
Post
5538
A new comparator on Spaces showcases Standard FLUX.2 Decoder vs. FLUX.2 Small Decoder. The Small Decoder is ~1.4Γ faster, uses ~1.4Γ less VRAM, and maintains near-identical image quality. It has ~28M parameters with narrower channels [96, 192, 384, 384] vs. [128, 256, 512, 512], and the demo supports sequence generation by running both decoders simultaneously and comparing the results side by side.
π€ Comparator: prithivMLmods/Flux.2-4B-Decoder-Comparator
π FLUX.2-small-decoder: black-forest-labs/FLUX.2-small-decoder
π GitHub: https://github.com/PRITHIVSAKTHIUR/Flux.2-4B-Encoder-Comparator
π Collection: https://huggingface.co/collections/prithivMLmods/image-generation-apps-collection
π€ > App built on the Gradio SDK. To learn more, visit the app page or the respective model pages.
π€ Comparator: prithivMLmods/Flux.2-4B-Decoder-Comparator
π FLUX.2-small-decoder: black-forest-labs/FLUX.2-small-decoder
π GitHub: https://github.com/PRITHIVSAKTHIUR/Flux.2-4B-Encoder-Comparator
π Collection: https://huggingface.co/collections/prithivMLmods/image-generation-apps-collection
π€ > App built on the Gradio SDK. To learn more, visit the app page or the respective model pages.
liked 5
models 2 days ago
mradermacher/Qwen3.5-9B-Claude-4.6-OS-INSTRUCT-GGUF
9B β’ Updated β’ 559 β’ 1
mradermacher/Qwen3.5-9B-Claude-4.6-OS-INSTRUCT-i1-GGUF
9B β’ Updated β’ 4.61k β’ 1
goonsai-com/Qwen3-gabliterated-image-generation
Text Generation β’ 4B β’ Updated β’ 305 β’ 1
mradermacher/Qwen3-gabliterated-image-generation-GGUF
4B β’ Updated β’ 878 β’ 1
mradermacher/Qwen_14B_rp-i1-GGUF
15B β’ Updated β’ 1.94k β’ 2