Fix: total_mem -> total_memory"

Add quick_test.py - standalone timing test, prints immediately, no imports from other files"

Fix: force unbuffered stdout for Kaggle notebook output + use print() with flush"

Add --log_batch_every: prints loss every N batches within an epoch"

Add --num_gpus: splits tasks across GPUs in parallel processes"

Add --log_every arg to control epoch print frequency (default 100)

Fix: add build_trm_nano_onnx function (alias used by train_task.py)

Add solved_tasks.json - 51 tasks solved by analytical solvers

Replace train_task.py with full batch runner: loops all 400 tasks, skips solved, logs everything, uses manual ONNX builder

Add manual ONNX builder with weight reuse - fits 1.44MB with full h=96 T=3 n=4"

Add TRM-Nano README with architecture details and usage"

Add TRM-Nano per-task training script with augmentation and ONNX export

Add TRM-Nano model implementation (h=96, 4 heads, 2 layers, unrolled recursion)"

Update SKILLS/kilo_agent/references/howto.md

Create kilo_client.py

Update classifier to support Kilo local server

Update TODO with TRM findings, fix outdated info

Add TRM research findings and architecture notes

Add composition solvers for chained transforms

Add DeepSeek task classifier for LLM-guided solver routing

reorg kilo skills folder

Add TRM solver skill for agent guidance

add skills for kilo agent

Upload trm_solver/__init__.py

Add TRM solver README

Add Kilo Bridge for LLM-driven ARC task analysis and ONNX export

Add TRM solver NN executor with 14 transform types and ONNX export

Upload SKILLS/kilo_agent/references/howto.md

Upload abstract reasoning corpus rl - Google Search.pdf

delete fles

move fles into pemf folder

Fix pyproject.toml: tell hatchling where the package is (itt_solver/)

Add pyproject.toml for uv install

Add Lightning.ai multi-GPU notebook with Ollama + Qwen2.5-Coder (auto-shards across GPUs)

Add Jupyter notebook for LLM synthesis (NVIDIA NIM / GLM 4.7 / DeepSeek V4)

Fix: default model is z-ai/glm4.7 on NVIDIA NIM

Add NVIDIA NIM provider (DeepSeek V4 Pro, GLM) — free tier

Add multi-provider LLM solver: Gemini, DeepSeek, GLM, Ollama

Add arc_results/already_solved.json

Add scripts/merge_results.py

Add scripts/kaggle_llm_solver.py

Update results: 70/400 (17.5%) — ITT:16 + Predicate:25 + DSL:29, zero regressions

Add predicate enumeration engine: neighborhood rules + enclosed fill + object predicates — 70/400 (17.5%)

Fix fill_color=0 bug + mixed->recolor fallback (980 lines)

Fix fill_color=0 bug in ITT rule learning + add mixed→recolor fallback

Update results: 47/400 (11.8%) — ITT solves 16, DSL solves 31, +7 new from ITT physics"

Add ITT physics engine (960 lines): PhiField, ρ_q, SigmaResidue, Fan Signatures, rule learning, FieldInvariants — 47/400 solved

Add ITT physics engine: PhiField dual-field, ρ_q boundary charge, SigmaResidue, Fan Signatures, TransformationRule, FieldInvariants — 47/400 (11.8%)"

Update LEARNING.md — add ITT source repo findings, object layer lessons, architecture direction

Update results: 40/400 (10.0%) solved — +9 new tasks, 0 regressions, greedy stacker working