ankitdhiman/haryanvi-tts
Viewer • Updated • 5.52k • 156 • 3
Ankit Dhiman
ankitd-nx
AI & ML interests
Speech and NLP
Recent Activity
liked a dataset 8 days ago
ankitdhiman/haryanvi-tts upvoted a changelog 8 days ago
Introducing Buckets: S3-like storage on the Hub liked a dataset 4 months ago
ankitdhiman/colloquial-hinglish-conversationsOrganizations
liked a
dataset 8 days ago
upvoted a changelog 8 days ago
Hugging Face Changelog
Introducing Buckets: S3-like storage on the Hub
• 176
liked a
dataset 4 months ago
reactedto sergiopaniego's post with 🤗 4 months ago
liked a
model 4 months ago
liked a
model 5 months ago
reactedto Kseniase's post with 🔥 5 months ago
Post
11216
11 Fascinating new Policy Optimization techniques
Policy optimization (PO) algorithms are central to training AI models with preference-based feedback. In recent weeks, numerous new PO methods have emerged that build on or replace the popular PPO and GRPO, solving their issues. Here are 11 of them:
1. BAlanced Policy Optimization (BAPO) → BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping (2510.18927)
Dynamically adjusting the clipping bounds in PPO-style updates to balance positive and negative gradients and prevent entropy collapse
2. Training-Free GRPO → Training-Free Group Relative Policy Optimization (2510.08191)
Instead of using numeric rewards, it compares rollouts semantically to distill useful knowledge as a token prior, which is then applied during inference to guide the model’s behavior
3. Asymmetric Importance Sampling Policy Optimization (ASPO) → ASPO: Asymmetric Importance Sampling Policy Optimization (2510.06062)
Fixes imbalanced token weighting in LLM training. It flips the importance sampling ratios for positive tokens to correct over- and under-updates, and adds a soft dual-clipping step to keep gradients stable
4. In-Context Steered Policy Optimization (ICPO) → https://arxiv.org/abs/2510.26519
Uses a model’s own in-context learning ability to guide training with existing data. It combines Mixed-Policy GRPO with Implicit Expert Forcing to expand exploration and adds Expert Region Reject Sampling and Annealed Expert-Bonus Reward Shaping to ensure stability and balanced expert influence
5. Graph-Enhanced Policy Optimization (GEPO) → https://arxiv.org/abs/2510.26270
Builds a graph of an agent’s experiences to understand how different states connect, guide exploration and assign rewards more effectively
6. Information Gain-based Policy Optimization (IGPO) → Information Gain-based Policy Optimization: A Simple and Effective Approach for Multi-Turn LLM Agents (2510.14967)
Uses the model’s own belief updates to create dense, informative feedback for smoother multi-turn learning
Read further below ⬇️
If you like this, also subscribe to the Turing post: https://www.turingpost.com/subscribe
Policy optimization (PO) algorithms are central to training AI models with preference-based feedback. In recent weeks, numerous new PO methods have emerged that build on or replace the popular PPO and GRPO, solving their issues. Here are 11 of them:
1. BAlanced Policy Optimization (BAPO) → BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping (2510.18927)
Dynamically adjusting the clipping bounds in PPO-style updates to balance positive and negative gradients and prevent entropy collapse
2. Training-Free GRPO → Training-Free Group Relative Policy Optimization (2510.08191)
Instead of using numeric rewards, it compares rollouts semantically to distill useful knowledge as a token prior, which is then applied during inference to guide the model’s behavior
3. Asymmetric Importance Sampling Policy Optimization (ASPO) → ASPO: Asymmetric Importance Sampling Policy Optimization (2510.06062)
Fixes imbalanced token weighting in LLM training. It flips the importance sampling ratios for positive tokens to correct over- and under-updates, and adds a soft dual-clipping step to keep gradients stable
4. In-Context Steered Policy Optimization (ICPO) → https://arxiv.org/abs/2510.26519
Uses a model’s own in-context learning ability to guide training with existing data. It combines Mixed-Policy GRPO with Implicit Expert Forcing to expand exploration and adds Expert Region Reject Sampling and Annealed Expert-Bonus Reward Shaping to ensure stability and balanced expert influence
5. Graph-Enhanced Policy Optimization (GEPO) → https://arxiv.org/abs/2510.26270
Builds a graph of an agent’s experiences to understand how different states connect, guide exploration and assign rewards more effectively
6. Information Gain-based Policy Optimization (IGPO) → Information Gain-based Policy Optimization: A Simple and Effective Approach for Multi-Turn LLM Agents (2510.14967)
Uses the model’s own belief updates to create dense, informative feedback for smoother multi-turn learning
Read further below ⬇️
If you like this, also subscribe to the Turing post: https://www.turingpost.com/subscribe
- 2 replies
liked a
model 5 months ago
reactedto cgeorgiaw's post with ❤️ 7 months ago
Post
6067
🚀🚀🚀 The largest ever dataset of co-folded 3D protein-ligand structures just dropped on HF!!
Meet SAIR (Structurally Augmented IC₅₀ Repository): 5M+ AI-generated complexes with experimentally measured drug potency data from SandboxAQ. 🚀🚀🚀
Check it out and explore here: SandboxAQ/SAIR
Meet SAIR (Structurally Augmented IC₅₀ Repository): 5M+ AI-generated complexes with experimentally measured drug potency data from SandboxAQ. 🚀🚀🚀
Check it out and explore here: SandboxAQ/SAIR
- 3 replies
liked a
dataset 7 months ago
liked a
model 7 months ago
Delete training_args.bin
#1 opened 7 months ago
by ankitd-nx
reactedto MonsterMMORPG's post with 🚀 7 months ago
Post
5811
Qwen Image Edit Full Tutorial: 26 Different Demo Cases, Prompts & Images, Pwns FLUX Kontext Dev
Tutorial Link
https://youtu.be/gLCMhbsICEQ
Extra Info
I tested newly arrived Qwen-Image-Edit-Lightning-8steps-V1.0 (arrived after tutorial recorded) and definitely our existing preset which uses Qwen-Image-Lightning-8steps-V1.1 is better than it, so this tutorial and presets are still 100% best quality and up-to-date
Info
Qwen Image Edit just has been published and since then I have been experimenting to prepare you this amazing tutorial. I have literally shown 26 unique cases and provided demo images and prompts. After watching this tutorial your image editing skills will move to next level i promise you that. Also this tutorial will give you a lot of ideas.
Tutorial Link
https://youtu.be/gLCMhbsICEQ
Extra Info
I tested newly arrived Qwen-Image-Edit-Lightning-8steps-V1.0 (arrived after tutorial recorded) and definitely our existing preset which uses Qwen-Image-Lightning-8steps-V1.1 is better than it, so this tutorial and presets are still 100% best quality and up-to-date
Info
Qwen Image Edit just has been published and since then I have been experimenting to prepare you this amazing tutorial. I have literally shown 26 unique cases and provided demo images and prompts. After watching this tutorial your image editing skills will move to next level i promise you that. Also this tutorial will give you a lot of ideas.
reactedto FlameF0X's post with 😔 7 months ago
Post
4177
I am very sad to say that the budget in creating of SnowflakeCore-G1 1b and 7b MoE models ran out and I can't pre-train them anymore.
- 7 replies
reactedto neph1's post with 🔥 7 months ago
Post
3656
I'm building a mmo-ish RPG with LLM agents that can (hopefully) complete player tasks, as an experiment. I've started documenting my progress here: https://huggingface.co/blog/neph1/rpg-llm-agents
Let me know if you want to see more of it.
Let me know if you want to see more of it.
- 4 replies
liked a
model 8 months ago
reactedto mitkox's post with 🚀 8 months ago
Post
2125
I run Qwen3-Coder 480B locally on my Z8, with a 1-million token context window. It’s the equivalent of parallel-parking a Nimitz-class carrier in a kiddie pool. Thanks to whatever dark pact the llama.cpp, CUDA, and kernel folks signed, hybrid inferencing + VRAM↔RAM offload let me stream the model’s synapses across Xeon, RAM, and four lonely A6000s without summoning either the OOM killer or a small house fire.