Notizie

In this tutorial, we explore TileGym GPU programming by building a practical Colab workflow that runs across different hardware conditions. We begin by probing the available CUDA environment, checking whether NVIDIA cuTile runs directly, and falling back to Triton when standard Colab GPUs lack the required cuTile stack. Through this setup, we learn the core tile-programming idea: instead of writing code for one thread at a time, we operate on entire data tiles, load them into the kernel, compute on them efficiently, and store the results back. We use this model to implement vector addition, fused GELU, row-wise softmax, tiled matrix multiplication, and flash attention, while comparing each result against PyTorch for correctness and benchmarking. CUDA Environment Probe Copy CodeCopiedUse a different Browser import os, sys, math, time, textwrap def rule(t=””): print(“n” + “=” * 78) if t: print(t) print(“=” * 78) rule(“0. ENVIRONMENT PROBE”) try: import torch except ImportError: print(“Installing torch …”) os.system(f”{sys.executable} -m pip install -q torch”) import torch HAS_CUDA = torch.cuda.is_available() DEV = “cuda” if HAS_CUDA else “cpu” cc = (0, 0) if HAS_CUDA: cc = torch.cuda.get_device_capability() print(f”GPU : {torch.cuda.get_device_name(0)}”) print(f”Compute capability : {cc[0]}.{cc[1]}”) print(f”Torch CUDA runtime : {torch.version.cuda}”) print(f”Driver / torch : {torch.__version__}”) else: print(“No CUDA GPU found. In Colab: Runtime -> Change runtime type -> GPU (T4).”) print(“The tutorial will still run its correctness math on CPU where possible.”) CUTILE_HW_OK = HAS_CUDA and cc[0] >= 8 CUDA_MAJOR = int((torch.version.cuda or “0”).split(“.”)[0]) if HAS_CUDA else 0 CUTILE_TOOLKIT_OK = CUDA_MAJOR >= 13 rule(“1. ATTEMPTING REAL cuTile (NVIDIA CUDA Tile) BACKEND”) ct = None CUTILE_READY = False if CUTILE_HW_OK and CUTILE_TOOLKIT_OK: try: import cuda.tile as ct CUTILE_READY = True print(“cuda.tile is already importable.”) except Exception: print(“Installing cuda-tile[tileiras] (this can take a while)…”) os.system(f”{sys.executable} -m pip install -q ‘cuda-tile[tileiras]’ cupy-cuda13x”) try: import cuda.tile as ct CUTILE_READY = True except Exception as e: print(“cuTile import still failed:”, repr(e)) else: reasons = [] if not HAS_CUDA: reasons.append(“no CUDA GPU”) if HAS_CUDA and cc[0] < 8: reasons.append(f”compute capability {cc[0]}.{cc[1]} < 8.0 (Turing/T4 unsupported)”) if not CUTILE_TOOLKIT_OK: reasons.append(f”CUDA {torch.version.cuda} < 13.1 required by tileiras”) print(“Skipping real cuTile install because:”, “; “.join(reasons) + “.”) print(“This is expected on a standard Colab T4 — we fall back to Triton below,”) print(“which teaches the exact same tile-based programming model.”) if CUTILE_READY: BACKEND = “cutile” else: try: import triton, triton.language as tl BACKEND = “triton” if HAS_CUDA else “torch” except ImportError: if HAS_CUDA: print(“Installing triton …”) os.system(f”{sys.executable} -m pip install -q triton”) try: import triton, triton.language as tl BACKEND = “triton” except Exception: BACKEND = “torch” else: BACKEND = “torch” rule(f”ACTIVE EXECUTION BACKEND: {BACKEND.upper()}”) print({ “cutile”: “Running NVIDIA cuTile kernels on your Ampere+/CUDA13 GPU. Nice hardware!”, “triton”: “Running Triton tile kernels on your GPU (the standard Colab path).”, “torch”: “No usable GPU kernel backend; showing reference math on CPU only.”, }[BACKEND]) print(textwrap.dedent(“”” —————————————————————— SIMT (classic CUDA) | TILE model (cuTile / Triton) —————————————————————— You write code for ONE | You write code for ONE BLOCK that thread. You compute a global | owns a whole TILE (e.g. 1024 elems index, bounds-check it, and | or a 128×128 sub-matrix). You load touch a single element. | the tile, do math on the WHOLE tile, | store it. The compiler maps the tile C[i] = A[i] + B[i] | onto threads / tensor cores for you. —————————————————————— cuTile primitives: ct.bid(0), ct.load(…), ct.store(…), a @ b, ct.launch Triton primitives: tl.program_id, tl.load, tl.store, tl.dot, grid[…] Same idea, two spellings. Below, every kernel is shown in BOTH. “””)) CUTILE_SOURCE = { “vector_add”: ”’ import cuda.tile as ct @ct.kernel def vector_add(a, b, c, tile_size: ct.Constant[int]): pid = ct.bid(0) a_tile = ct.load(a, index=(pid,), shape=(tile_size,)) b_tile = ct.load(b, index=(pid,), shape=(tile_size,)) ct.store(c, index=(pid,), tile=a_tile + b_tile) ”’, “matmul”: ”’ import cuda.tile as ct @ct.kernel def matmul(A, B, C, K: ct.Constant[int], BM: ct.Constant[int], BN: ct.Constant[int], BK: ct.Constant[int]): m, n = ct.bid(0), ct.bid(1) acc = ct.zeros((BM, BN), dtype=ct.float32) for k in range(ct.cdiv(K, BK)): a = ct.load(A, index=(m, k), shape=(BM, BK)) b = ct.load(B, index=(k, n), shape=(BK, BN)) acc = a @ b + acc ct.store(C, index=(m, n), tile=acc) ”’} We begin by setting up the environment, importing the required libraries, and checking whether CUDA is available on the current runtime. We inspect the GPU capabilities, CUDA version, and PyTorch setup to determine whether the real cuTile backend is usable. We then select the active execution backend, explain the tile programming model, and store reference cuTile kernel source strings for comparison. Defining Triton Kernels Copy CodeCopiedUse a different Browser if BACKEND == “triton”: @triton.jit def _vadd_kernel(a_ptr, b_ptr, c_ptr, n, BLOCK: tl.constexpr): pid = tl.program_id(0) offs = pid * BLOCK + tl.arange(0, BLOCK) mask = offs < n a = tl.load(a_ptr + offs, mask=mask) b = tl.load(b_ptr + offs, mask=mask) tl.store(c_ptr + offs, a + b, mask=mask) @triton.jit def _fused_gelu_kernel(x_ptr, w_ptr, b_ptr, o_ptr, n, BLOCK: tl.constexpr): pid = tl.program_id(0) offs = pid * BLOCK + tl.arange(0, BLOCK) mask = offs < n x = tl.load(x_ptr + offs, mask=mask) w = tl.load(w_ptr + offs, mask=mask) b = tl.load(b_ptr + offs, mask=mask) h = x * w + b c = 0.7978845608028654 z = c * (h + 0.044715 * h * h * h) e = tl.exp(-2.0 * z) tanh = (1.0 – e) / (1.0 + e) g = 0.5 * h * (1.0 + tanh) tl.store(o_ptr + offs, g, mask=mask) @triton.jit def _softmax_kernel(x_ptr, o_ptr, stride, n_cols, BLOCK: tl.constexpr): row = tl.program_id(0) cols = tl.arange(0, BLOCK) mask = cols < n_cols ptr = x_ptr + row * stride + cols x = tl.load(ptr, mask=mask, other=-float(“inf”)) x = x – tl.max(x, axis=0) num = tl.exp(x) den = tl.sum(num, axis=0) tl.store(o_ptr + row * stride + cols, num / den, mask=mask) @triton.jit def _matmul_kernel(A, B, C, M, N, K, sam, sak, sbk, sbn, scm, scn, BM: tl.constexpr, BN: tl.constexpr, BK: tl.constexpr): pid_m = tl.program_id(0) pid_n = tl.program_id(1) offs_m = pid_m * BM + tl.arange(0, BM) offs_n = pid_n * BN + tl.arange(0, BN) offs_k = tl.arange(0, BK) a_ptr = A + offs_m[:, None] * sam + offs_k[None, :] * sak b_ptr

arXiv:2607.08346v1 Announce Type: new Abstract: Form 8-K filings are the primary channel through which U.S. public companies disclose material events, but the SEC item codes attached to them are coarse: a single item spans routine administrative changes and chief executive departures, and many of the most market-moving disclosures fall into a catch-all item. Large language models make fine-grained labelling feasible at corpus scale, but only if the labels can be traced to the source text and shown to be reliable. We present a two-stage system that tags 8-K disclosures against a three-tier taxonomy of 119 event types. The first stage constrains output to valid taxonomy entries and anchors every tag to a verbatim quote via fuzzy n-gram validation; the second re-grades each cited quote against the category definition to produce a quality score. Applying the system to 292,984 filings from 2022 to 2026 yields 601,088 grounded event tags, which we release. Over 5,125 stratified tags, an LLM judge finds precision rises monotonically with the quality score, from 12% to 96%, while unsupported tags fall from 8% to near zero. Ablation shows the score is calibrated only when assigned in a dedicated second pass. An event study on unsigned abnormal returns confirms, without any language model, that the taxonomy separates economically distinct events sharing an item code.

arXiv:2607.08017v1 Announce Type: new Abstract: Large-Language Models (LLMs) can be prone to flawed and unfaithful reasoning that decoding strategies like Self-Consistency (SC) fail to detect as they evaluate only final-answer agreement while ignoring the logical validity of intermediate steps. This raises three fundamental questions: How can we reliably quantify uncertainty in LLM reasoning? Can semantic, structural, and causal awareness select more faithful reasoning compared to na”ive majority voting? and How robust is reasoning topology under adversarial conditions? To address these questions, we introduce GRAPHEVAL, a graph-based reasoning framework that re-frames uncertainty quantification (UQ) as a holistic reasoning fidelity problem. We propose a novel UQ metric, Graph Reasoning Coherence Score (GRCS), that quantifies semantic-structural consensus of the reasoning space and captures pathological mode collapse and confident hallucinations. We find that GRCS is the only metric that is consistently negatively correlated with reasoning faithfulness across both more capable and smaller models. Additionally, we introduce Graph Self-Consistency (GSC), a medoid-based decoding strategy that trades nominal accuracy for reasoning fidelity, exposing the degree to which SC is inflated by unfaithful lucky guesses in smaller models, while preserving or improving accuracy in more capable ones. Finally, through adversarial medoid ablation, we demonstrate that the GSC-selected path acts as a “load-bearing path” and forcing models away from it degrades reasoning faithfulness and, in targeted cases, causes drops in accuracy.

arXiv:2606.29031v2 Announce Type: replace Abstract: In regulated domains such as banking and healthcare, where privacy constraints make real speech costly to collect and retain, synthetic speech from modern text-to-speech (TTS) is an appealing alternative for training automatic speech recognition (ASR) without exposing sensitive customer recordings. Yet a persistent distributional gap between synthetic and real data limits how far it can replace genuine recordings. Prior work largely treats this gap as a black box to be engineered around, but in our work, we instead examine its origin directly by probing a SLAM-ASR architecture. Then, we localise where its LLM backbone separates real from synthetic speech and find the discriminative signal concentrated in the early-to-middle layers, where temporal and prosodic perturbations disrupt it most. We further show that representation-level separability, help, but does not directly predict downstream ASR gains. On the other hand, convolving synthetic audio with room impulse responses (RIRs) narrows the gap not by making synthetic speech sound cleaner or more natural, but by reproducing the acoustic irregularities of real recordings. Translating these findings into the training procedure, by adding a layer-selection module combined with RIR augmentation matches a fully real-data baseline using only 25% of the real speech (13.6h) and surpasses it at all higher proportions.

arXiv:2508.09767v3 Announce Type: replace-cross Abstract: We propose UtterTune, a lightweight method for adapting a multilingual text-to-speech (TTS) system built on a large language model (LLM). It improves control of pronunciation in the target language while preserving performance in the others. Although LLM architectures have enabled TTS models to achieve remarkable naturalness, accurately modeling grapheme-to-phoneme (G2P) mapping and prosody remains challenging, especially when the model omits an explicit G2P module and directly processes minimally encoded text (e.g., byte-pair encoding). UtterTune leverages low-rank adaptation to enable the control of segmental pronunciation and pitch accent at the phoneme level for Japanese speech, the target language in this paper, while maintaining naturalness and speaker similarity in a zero-shot setting. Objective and subjective evaluations confirm its effectiveness.

Robbyant, Ant Group’s embodied-intelligence unit, has released LingBot-World-Infinity (LingBot-World 2.0). It is a causal video generation model that behaves as an interactive world simulator. It is how the team attacks two failure modes: long-horizon drift and interactive latency. What is LingBot-World-Infinity? An interactive world model generates video frame by frame, conditioned on a stream of user actions. Each state depends only on past frames and current input. The research team formalizes this as a causal factorization: Copy CodeCopiedUse a different Browser p_θ(x_1:T | a_1:T) = Π_t p_θ(x_t | x_<t, a_≤t) Here x_t is the visual state at time t. The action a_t combines a camera pose and a text prompt. Camera pose uses Plücker embeddings, injected through adaptive layer normalization (AdaLN). Text enters as chunk-wise prompts through cross-attention. The research team claims four upgrades over LingBot-World: An unbounded interaction horizon with consistent output quality. A distilled real-time variant sufficient to drive 720p video streams at 60 fps. A broader action space, including attacking, archery, spell-casting, and shooting. An agentic harness pairing a pilot agent with a director agent. The primary model is 14B. A lightweight 1.3B counterpart is described as deployable on a single GPU. The Architecture: MoBA and Two-Stage Training The core contribution is the Mixture of Bidirectional and Autoregressive (MoBA) Attention Mask. It explains the drift. Standard autoregressive video training uses a teacher forcing mask. Each noisy frame attends to itself and its clean context. The research team found a failure here. As context grows, the model leans on that context instead of predicting future frames. The result is overfitting and visual quality degradation. MoBA appends a bidirectional full-attention block to the teacher forcing mask. That block acts as a regularizer. It also helps the model handle flexible-length generation. The cross-attention mask mirrors the split. The autoregressive component attends to a background prompt along with chunk-wise prompts in a lower-triangular pattern. That prevents future semantics from leaking backward. The bidirectional component attends to one global prompt. Pre-training optimizes a conditional flow-matching objective with rectified-flow interpolation. Post-training then compresses the multi-step teacher into a few-step student: Consistency distillation: Latents on the same teacher probability-flow ODE (PF-ODE) trajectory must map to identical predictions. Distribution matching distillation (DMD): The generator follows the KL gradient between noised student and noised data distributions. The important detail sits in the DMD. The research team applies it over long self-rollout trajectories, not only teacher-forced states. The student is therefore optimized on the state distribution its own predictions induce. That is the stated mechanism behind anti-drift. The Agentic Harness: The Feature Worth Taking Seriously A frame predictor does not play itself. The Robbyant research team wraps the generator in a Director-Pilot Co-Simulation Framework. As described in the research paper, a Vision-Language Model is the Director. It governs macroscopic semantic rules and causal reasoning. The Diffusion Transformer video generator is the Pilot. It simulates low-level physical dynamics and renders transitions. The harness exposes two interaction modes: Mode A: Direct Semantic Interaction. The VLM reads the current frame and generates event cards. No object masks are required. Mode B: Tracking-Assisted Object Interaction. A SAM-based (Segment Anything Model) action-proposal loop tracks objects across chunks. Users select a tracked object and trigger actions. The research paper shows door-opening and ball-rotating rollouts. Users can also intervene textually. Global state shifts change time of day or weather. Local entity injection spawns creatures, and the VLM picks plausible entry points. The interface follows game conventions. WASD drives movement, IJKL controls view. Space triggers a jump; P triggers a wing glide. Keys U and O carry VLM-proposed character actions. Keys F and G carry environmental events. Numeric keys are user-registered event slots. Hands-On: What Ships and What Doesn’t Expectations need calibration here. The codebase is built on Wan2.2. Only lingbot-world-v2-14b-causal-fast is downloadable today. The causal-pretrained 14B, the bidirectional 14B, and both 1.3B variants are marked TODO. Copy CodeCopiedUse a different Browser git clone https://github.com/robbyant/lingbot-world-v2.git cd lingbot-world-v2 pip install -r requirements.txt # torch >= 2.4.0 pip install flash-attn –no-build-isolation huggingface-cli download robbyant/lingbot-world-v2-14b-causal-fast –local-dir ./lingbot-world-v2-14b-causal-fast The provided generate.py runs causal inference with KV caching. It processes frames chunk-by-chunk rather than all at once. The reference command is eight-GPU and 480P: Copy CodeCopiedUse a different Browser torchrun –nproc_per_node=8 generate.py –task i2v-A14B –size 480*832 –ckpt_dir lingbot-world-v2-14b-causal-fast –image examples/03/image.jpg –action_path examples/03 –dit_fsdp –t5_fsdp –ulysses_size 8 –frame_num 361 –local_attn_size 18 –sink_size 6 –prompt “A serene lakeside scene with a lone tree standing in calm water…” The released reference script is 480×832 across eight GPUs. The 60 fps figure describes the deployed stream, which passes a spatio-temporal refiner. That refiner upsamples decoded frames, then synthesizes intermediate frames for a higher frame rate. Both stages compile into TensorRT engines. A Diffusers checkpoint also exists: Copy CodeCopiedUse a different Browser import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image, export_to_video pipe = DiffusionPipeline.from_pretrained( “robbyant/lingbot-world-v2-14b-causal-fast”, dtype=torch.bfloat16, device_map=”cuda”) frames = pipe(image=load_image(“seed.png”), prompt=”…”).frames[0] export_to_video(frames, “output.mp4″) For a hosted path, Reactor serves the model as reactor/lingbot-world-2. Its docs list 48 fps at 1664×960. Sessions are command-driven and stateful: Copy CodeCopiedUse a different Browser from reactor_sdk import Reactor, ReactorStatus reactor = Reactor(model_name=”reactor/lingbot-world-2”, api_key=KEY) @reactor.on_status(ReactorStatus.READY) async def on_ready(status): ref = await reactor.upload_file(“seed.jpg”) await reactor.send_command(“set_image”, {“image”: ref}) await reactor.send_command(“set_prompt”, {“prompt”: “A misty alpine valley.”}) await reactor.send_command(“start”, {}) Movement is persistent state, not a pulse. set_move_longitudinal: “forward” drives until you send “idle”. Commands land at the next chunk boundary.  Comparison The research paper’s comparison is qualitative. Every superiority claim rests on side-by-side frame grids. Property M-G 3.0 D-W LingBot-World HappyOyster Genie 3 LingBot-World-Infinity Generation duration Minutes Minutes Minutes Minutes Minutes Hours (Infinite) Semantic interaction None None None Few Few Infinite Domain Game General General General General General Dynamic degree Medium Medium High Medium Medium High Real-time Yes Yes Yes Yes Yes Yes Open-source Yes Yes Yes No No Yes Use Cases Game and level prototyping: Seed an image of a canal town. Hot-swap the prompt to summon a snowstorm. Iterate on mood before any asset pipeline exists. Embodied simulation: Generate first-person rollouts under scripted camera poses. Feed frames to

arXiv:2607.08010v1 Announce Type: new Abstract: Production LLM agents often waste latency and reliability by regenerating code for the same procedural steps on every request. We replace this inference-time coding loop with an agentic tool-making pipeline that compiles repeated SOP steps into validated, versioned tools before deployment. The tool-maker grounds synthesis in the live environment as it collects execution traces, observes backend schemas and values, generates candidate tools, and repairs them against labeled cases. At runtime, the production agent calls these tools directly and falls back to code generation only when needed. We deploy the approach in a Fulfillment Center alarm-triage system, where an agent diagnoses alarms against a 44-node SOP over heterogeneous metric backends. In production, tool calls reduce p50 latency by 42%. On 1,500 historical alarms, they reduce end-to-end error rate by up to 53% by suppressing run-to-run variance in repeated steps. Because tools return compact structured verdicts, they also enable a simpler direct-call architecture, reducing p50 latency by a further 62% in a controlled ablation. Versioned tools also improve auditability and expose specification gaps and upstream data drift. Our results show that self-evolving agents can make industrial LLM systems faster, more reliable, and easier to operate.

Ties van der Meer doesn’t know how many siblings he has. The 47-year-old was conceived at a private fertility clinic in the Netherlands using sperm provided by an anonymous donor. After the Netherlands banned anonymous donation in 2004, the doctor who ran the clinic destroyed records that might have identified those donors, he says. He describes the situation as “problematic.” Children have a right to know their biological parents, he says. While he did ultimately track down one sibling, who helped him identify his father along with other genetic relatives, he may have others he’ll never find. Other donor-conceived people who have been able to track down siblings have found they have tens or even hundreds of them. One donor-conceived woman who found 25 half-siblings over the course of seven years told the Guardian, “It does make you feel a bit mass-produced.” We need international limits on the number of children a single donor can contribute to, a European fertility organization argued yesterday. At a conference in London, members laid out plans to start with a Europe-wide limit. Today many countries, including the UK, have banned anonymous egg and sperm donation. But anonymity can’t be guaranteed even in places where it is technically allowed. Genetic tests offered by companies like Ancestry and 23andMe, along with genetic registries, have made it much easier for donor-conceived people to find parents and siblings who share their genes. And because sperm can be frozen and stored for years before it is eventually used, the current set-up can result in situations where donor-conceived people discover the identity of a genetic parent only after the person’s death. They might also find that they have siblings of very different ages, all around the world. Some people are finding hundreds of siblings. Sperm from Jonathan Meijer, a Dutch man who began donating in 2007, was used to conceive between 550 and 600 children. (Stichting Donorkind, a foundation and advocacy group for donor-conceived people that’s chaired by van der Meer, took him to court, and he was ordered to stop donating in 2023.) Stories like these can be distressing for donor-conceived people. And there are other reasons why limits are considered important. The offspring of a prolific donor might be at risk of unknowingly forming romantic or sexual relationships, for instance. And some people are concerned that a donor with a harmful genetic mutation might pass that down to many children. This is unlikely, given the level of screening that most donors undergo. But it has happened. A man who donated his sperm to a sperm bank in Denmark was found to have a genetic mutation that significantly increased the risk of multiple cancers. But his sperm had already been used to conceive at least 197 children across Europe. Some of those children developed cancer. Some died. Many countries already have legal limits for donors. In Malta and Cyprus, for example, both egg and sperm donors are allowed to contribute to the birth of just a single child, according to data presented at the European Society of Human Reproduction and Embryology (ESHRE) meeting in London on July 8. Other countries set limits based on the number of families a single donor can contribute to, allowing recipients to have children who share a genetic link. In the UK, that limit is set at 10 families per donor. But these limits are difficult to enforce, partly because donated gametes don’t necessarily stay in their original country. In Denmark, the national limit is set at 12 families. But the country is a major exporter of sperm. In the UK, for example, more than half of sperm donations in 2020 were imported—with most of those coming from either Denmark or the US. “The only thing that really makes sense is a transnational limit,” Jackson Kirkman-Brown, a professor of reproductive biology at the University of Birmingham, said at the meeting. Kirkman-Brown and his colleagues have spent months putting together a document that represents ESHRE’s position on these limits. After consulting with fertility specialists, clinics, sperm and egg banks, donors, and donor-conceived people, the team has developed a plan to start with a Europe-wide limit on sperm and egg donations. ESHRE is calling on sperm and egg banks, as well as fertility clinics, to respect an initial limit of 50 families per donor. That’s still very high, according to a handful of people I spoke to at the meeting. But at least it’s a start. Europe should move toward setting limits at 15 families per donor, Kirkman-Brown said. “We may find that 15 is also too high,” says Vasanti Jadva, who studies the psychological well-being of people conceived using donated eggs, sperm, and embryos at City St George’s in London. “We still don’t know what the right number is.” It will be difficult to enforce these limits, too. And if they end up limiting the supply of donor sperm, there’s a chance that some people will turn to unregulated sperm donations from people who do not undergo health screening. Unregulated donations can lead to other problems for prospective parents, including the possibility that donors will seek parental rights over the children conceived using their sperm. And it will be even harder to establish international limits. When I asked the American Society of Reproductive Medicine for its thoughts on ESHRE’s proposed limits, a representative directed me to a guidance document saying “it has been suggested” that for a population of 800,000, single donors should be limited to “no more than 25 births” in order to avoid the risk that relatives will have children together. (Considering the US has a population of over 340 million, the total figure could be pretty high, but many sperm banks opt to limit the number of families contributed to by a single donor at around 25.) Van der Meer thinks that even a limit of five families from a single donor would be high. International donation makes it even harder for donor-conceived people to connect with genetic relatives, so the limit for international contributions should be set at two