Latent-Identity Tuning in Text-to-Image Personalization Models
Paper β’ 2607.11885 β’ Published β’ 10
ZeroGPU Explorers
community
AI & ML interests
None defined yet.
Recent Activity
View all activity
garibidaΒ
submitted a
paper to Daily Papers 4 days ago
sayakpaulΒ
authored 3
papers 8 days ago
sayakpaulΒ
submitted a
paper to Daily Papers 8 days ago
caizhongangΒ
authored a
paper 8 days ago
Post
2008
You can explore and try my Agentic Space Factory:
fffiloni/agentic-space-factory-UI-public
Be kind π€
fffiloni/agentic-space-factory-UI-public
Be kind π€
Post
1799
I made a Hugging Face Space for SCAIL-2 π€
Reference character + driving motion β animated result.
A simple demo to explore the paperβs core workflow with curated examples.
π fffiloni/SCAIL-2
Reference character + driving motion β animated result.
A simple demo to explore the paperβs core workflow with curated examples.
π fffiloni/SCAIL-2
- 1 reply
Post
805
β±οΈ Built a small Space for Visual Chronometer / Pulse of Motion.
Upload a video and estimate its Physical FPS: the frame rate implied by visual motion, independent of metadata.
Useful to inspect βchronometric hallucinationβ in generated videos: clips that look smooth, but move with the wrong physical time scale.
Try it here: fffiloni/Pulse-of-Motion
Upload a video and estimate its Physical FPS: the frame rate implied by visual motion, independent of metadata.
Useful to inspect βchronometric hallucinationβ in generated videos: clips that look smooth, but move with the wrong physical time scale.
Try it here: fffiloni/Pulse-of-Motion
- 1 reply
ymoslemΒ
authored a
paper 18 days ago
ymoslemΒ
submitted a
paper to Daily Papers 18 days ago
IliaLarchenkoΒ
submitted a
paper to Daily Papers 19 days ago
Post
366
SPROG-9M β a 9.37M parameter model trained from scratch to solve GSM8K-style math without using an LLM at inference.
The model, codelion/sprog-9m, predicts symbolic programs over number slots, then a deterministic executor does the arithmetic. With a simple verifier, it reaches ~11.8% on GSM8K test.
We also released the dataset: codelion/gsm8k-synth, 117K validated synthetic GSM8K-style problems.
Tiny model, no pretraining, no LLM at inference, runs on a laptop.
The model, codelion/sprog-9m, predicts symbolic programs over number slots, then a deterministic executor does the arithmetic. With a simple verifier, it reaches ~11.8% on GSM8K test.
We also released the dataset: codelion/gsm8k-synth, 117K validated synthetic GSM8K-style problems.
Tiny model, no pretraining, no LLM at inference, runs on a laptop.
IliaLarchenkoΒ
authored a
paper 20 days ago
IliaLarchenkoΒ
posted an update 21 days ago
Post
188
I placed π₯ 2nd in the LeHome Challenge (ICRA 2026), and π₯ 1st of 62 teams in the first simulation round. Now I'm open-sourcing the full solution β code, tech report, and final weights.
The task: teach a cheap two-armed robot (SO-ARM101) to fold 4 garment types β long/short tops and pants. Garment category is hidden at eval. Round 1 in sim (auto-scored), round 2 on a real robot (jury-scored).
I trained a VLA policy with an RL loop on top. The key ideas:
π§ The policy is its own value function. From the same forward pass that picks the next action chunk, cheap heads predict success probability, task completion %, garment type, and future keypoint distances + a Q-residual. Those become the advantage signal for RL β no separate critic.
π A fully asynchronous RL loop coordinated only through the HF Hub: 1 trainer (H200) ships a fresh checkpoint ~every 40 min while N rollout workers (and a human doing teleop / DAgger corrections) collect data in parallel. Nobody waits β it uses the off-policy nature of the loop to the fullest.
π Binary success is too sparse, so I densify it into per-frame advantage via GAE β from objective keypoint checkpoints, the success-probability value baseline, and completion %.
ποΈ The RL combines AWR + RECAP. I also tune the inference knobs β execution length, playback speed, inpainting overlap, CFG scale, best-of-N β with a per-parameter Thompson-sampling bandit folded into rollout collection.
π§ Round 2: with only ~1 week and no access to the eval robot β so the pipeline was sim β my robot β their robot, leaning on heavy augmentation to make the policy more robust.
π Blog: https://ilialarchenko.com/projects/lehome2026
π Tech report: Learning to Fold: prizewinning solution at LeHome Challenge 2026 (1st place online, 2nd offline) (2606.27163)
π§ Code: https://github.com/IliaLarchenko/lehome_solution
π€ Sim policy: IliaLarchenko/lehome_sim
π€ Real policy: IliaLarchenko/lehome_real
π Challenge: https://lehome-challenge.com/
The task: teach a cheap two-armed robot (SO-ARM101) to fold 4 garment types β long/short tops and pants. Garment category is hidden at eval. Round 1 in sim (auto-scored), round 2 on a real robot (jury-scored).
I trained a VLA policy with an RL loop on top. The key ideas:
π§ The policy is its own value function. From the same forward pass that picks the next action chunk, cheap heads predict success probability, task completion %, garment type, and future keypoint distances + a Q-residual. Those become the advantage signal for RL β no separate critic.
π A fully asynchronous RL loop coordinated only through the HF Hub: 1 trainer (H200) ships a fresh checkpoint ~every 40 min while N rollout workers (and a human doing teleop / DAgger corrections) collect data in parallel. Nobody waits β it uses the off-policy nature of the loop to the fullest.
π Binary success is too sparse, so I densify it into per-frame advantage via GAE β from objective keypoint checkpoints, the success-probability value baseline, and completion %.
ποΈ The RL combines AWR + RECAP. I also tune the inference knobs β execution length, playback speed, inpainting overlap, CFG scale, best-of-N β with a per-parameter Thompson-sampling bandit folded into rollout collection.
π§ Round 2: with only ~1 week and no access to the eval robot β so the pipeline was sim β my robot β their robot, leaning on heavy augmentation to make the policy more robust.
π Blog: https://ilialarchenko.com/projects/lehome2026
π Tech report: Learning to Fold: prizewinning solution at LeHome Challenge 2026 (1st place online, 2nd offline) (2606.27163)
π§ Code: https://github.com/IliaLarchenko/lehome_solution
π€ Sim policy: IliaLarchenko/lehome_sim
π€ Real policy: IliaLarchenko/lehome_real
π Challenge: https://lehome-challenge.com/
Post
1560
A few weeks ago, @victor opened the door: coding agents can now ship Hugging Face Spaces autonomously.
I pulled on that thread.
As someone who builds and ships Gradio demos regularly, I didnβt just want to reproduce the loop. I wanted to see what happens when that loop is plugged into the whole Hugging Face stack.
The interesting part is not only that an agent can ship a Space.
Itβs what happens when Space generation becomes a first-class Hugging Face workflow.
That became Agentic Space Factory.
More soon. π€
I pulled on that thread.
As someone who builds and ships Gradio demos regularly, I didnβt just want to reproduce the loop. I wanted to see what happens when that loop is plugged into the whole Hugging Face stack.
The interesting part is not only that an agent can ship a Space.
Itβs what happens when Space generation becomes a first-class Hugging Face workflow.
That became Agentic Space Factory.
More soon. π€
- 1 reply
omer11aΒ
submitted a
paper to Daily Papers 23 days ago
wenbopanΒ
authored 2
papers about 1 month ago
wenbopanΒ
submitted a
paper to Daily Papers about 1 month ago