Hugging Face's logo
abdo hamdy's picture
1

abdo hamdy

abdohamdy12739
Fishtiks's profile picture
·

AI & ML interests

None yet

Recent Activity

reacted to Yann-CV's post with 🔥 12 days ago
🚀 Introducing Goldener: The Python Data Orchestrator for more efficient ML Machine Learning workflows often rely on randomness: selecting/splitting data for training, batching it variably, and monitoring real-world performance. Nowadays, foundation models give access to the semantics of data. Goldener leverages this semantic to make the entire ML lifecycle more efficient! 🔗 Check it out: https://github.com/goldener-data/goldener 🔨 Give it a try: pip install goldener
reacted to ManniX-ITA's post with 🔥 12 days ago
Two custom releases — both unusual takes on common problems, on a single RTX 3090 + a vast.ai pod. 🔹 ManniX-ITA/Qwen3.5-27B-Omnimerge-v2 3-source weight-space merge over Qwen3.5-27B combining OBIM-lite magnitude masking + DAREx rescaling + EMR election (sign from consensus, amplitude from max-abs across sources). GPU-accelerated, ~35× over CPU. Sources: Claude-4.6-Opus-distill (0.40), Esper3.1 code (0.35), Gemini-3.1-Pro-distill (0.25). density 0.53, DAREx q 0.75. Q6_K vs best source: • GPQA Diamond: 53.03 → 69.19 (+16.16 pp) • MBPP pass@1: 71.20 → 74.60 (+3.40) • HumanEval pass@1: 76.22 → 79.27 (+3.05) vs Omnimerge v1 (vanilla DARE-TIES): +8.08 pp GPQA, +2.80 MBPP. Amplitude-from-max + sign-from-consensus is what unlocked the GPQA jump. 🔹 ManniX-ITA/gemma-4-A4B-98e-v3-it Gemma 4 26B-A4B pruned 128 → 98 experts/layer (-23.4% MoE capacity, -5.2B params), zero GPQA degradation. GPQA Diamond: • 128e reference: 75.25% • 98e v3 (this): 75.25% — +0.00 pp despite -23.4% capacity, -5.2B params • 109e v3 (older): 71.72% — -3.53 pp The win over 109e v3 came from changing the importance map: aggregate per-expert contribution across math/logic/code/science/creative via 128-token teacher-force, instead of GPQA-specific per-question top-16 (which overfitted). Result: more experts dropped, quality preserved. Findings worth flagging: • Experts NOT topic-specialized — 28/32 overlap math/creative top-32. • Expert weight cosine ≈ 0.05 max → merging destroys the model. Dropping is the only viable structural compression here. • Contribution Gini ≈ 0.38 → ~75 experts/layer carry 80% of signal. Eval: lm-eval gpqa_diamond_cot_zeroshot, llama-server --reasoning-format deepseek --reasoning-budget 8192, Gemma 4 official sampling. Feedback welcome.
liked a model 6 months ago
Qwen/Qwen3-VL-2B-Instruct-FP8
View all activity

Organizations

z2data's profile picture