Law practice management software is software designed to manage the business operations of a law firm. This can include software that manages cases, client intake, court communications, electronic discovery, time tracking, trust accounting, and billing. == Features of law practice management software == Common features of practice management software include: Case management Time tracking Document assembly Contact management Calendaring Docket management Client portal Contract Management Court Case Status Tracker Trust accounting == Examples of law practice management software == Smokeball LEAP Legal Software PracticeEvolve Dye & Durham
Residual neural network
A residual neural network (also referred to as a residual network or ResNet) is a deep learning architecture in which the layers learn residual functions with reference to the layer inputs. It was developed in 2015 for image recognition, and won the ImageNet Large Scale Visual Recognition Challenge (ILSVRC) of that year. As a point of terminology, "residual connection" refers to the specific architectural motif of x ↦ f ( x ) + x {\displaystyle x\mapsto f(x)+x} , where f {\displaystyle f} is an arbitrary neural network module. The motif had been used previously (see §History for details). However, the publication of ResNet made it widely popular for feedforward networks, appearing in neural networks that are seemingly unrelated to ResNet. The residual connection stabilizes the training and convergence of deep neural networks with hundreds of layers, and is a common motif in deep neural networks, such as transformer models (e.g., BERT, and GPT models such as ChatGPT), the AlphaGo Zero system, the AlphaStar system, and the AlphaFold system. == Mathematics == === Residual connection === In a multilayer neural network model, consider a (non-residual) subnetwork with a certain number of stacked layers (e.g., 2 or 3). Let H ( x ; α ) {\displaystyle H(x;\alpha )} denote the subnetwork. Suppose H ∗ {\displaystyle H^{}} is the desired optimal output of this subnetwork. Residual learning simply adds x {\displaystyle x} directly to the output, such that the optimal learned output now becomes be H ∗ − x {\displaystyle H^{}-x} , which is interpreted as a "residual" with respect to x {\displaystyle x} . The operation of "adding x {\displaystyle x} " is implemented via a "skip connection" that performs an identity mapping to connect the input of the subnetwork with its output. This connection is referred to as a "residual connection" in later work. Let F ( x ; α ) = H ( x ; a ) + x {\displaystyle F(x;\alpha )=H(x;a)+x} . The function F {\displaystyle F} is often represented by matrix multiplication interlaced with activation functions and normalization operations (e.g., batch normalization or layer normalization). As a whole, one of these subnetworks is referred to as a "residual block". A deep residual network is constructed by simply stacking these blocks. Long short-term memory (LSTM) has a memory mechanism that serves as a residual connection. In an LSTM without a forget gate, an input x t {\displaystyle x_{t}} is processed by a function F {\displaystyle F} and added to a memory cell c t {\displaystyle c_{t}} , resulting in c t + 1 = c t + F ( x t ) {\displaystyle c_{t+1}=c_{t}+F(x_{t})} . An LSTM with a forget gate essentially functions as a highway network. To stabilize the variance of the layers' inputs, it is recommended to replace the residual connections x + f ( x ) {\displaystyle x+f(x)} with x / L + f ( x ) {\displaystyle x/L+f(x)} , where L {\displaystyle L} is the total number of residual layers. === Projection connection === If the function F {\displaystyle F} is of type F : R n → R m {\displaystyle F:\mathbb {R} ^{n}\to \mathbb {R} ^{m}} where n ≠ m {\displaystyle n\neq m} , then F ( x ) + x {\displaystyle F(x)+x} is undefined. To handle this special case, a projection connection is used: y = F ( x ) + P ( x ) {\displaystyle y=F(x)+P(x)} where P {\displaystyle P} is typically a linear projection, defined by P ( x ) = M x {\displaystyle P(x)=Mx} where M {\displaystyle M} is a m × n {\displaystyle m\times n} matrix. The matrix is trained via backpropagation, as is any other parameter of the model. === Signal propagation === The introduction of identity mappings facilitates signal propagation in both forward and backward paths. ==== Forward propagation ==== If the output of the ℓ {\displaystyle \ell } -th residual block is the input to the ( ℓ + 1 ) {\displaystyle (\ell +1)} -th residual block (assuming no activation function between blocks), then the ( ℓ + 1 ) {\displaystyle (\ell +1)} -th input is: x ℓ + 1 = F ( x ℓ ) + x ℓ {\displaystyle x_{\ell +1}=F(x_{\ell })+x_{\ell }} Applying this formulation recursively, e.g.: x ℓ + 2 = F ( x ℓ + 1 ) + x ℓ + 1 = F ( x ℓ + 1 ) + F ( x ℓ ) + x ℓ {\displaystyle {\begin{aligned}x_{\ell +2}&=F(x_{\ell +1})+x_{\ell +1}\\&=F(x_{\ell +1})+F(x_{\ell })+x_{\ell }\end{aligned}}} yields the general relationship: x L = x ℓ + ∑ i = ℓ L − 1 F ( x i ) {\displaystyle x_{L}=x_{\ell }+\sum _{i=\ell }^{L-1}F(x_{i})} where L {\textstyle L} is the index of a residual block and ℓ {\textstyle \ell } is the index of some earlier block. This formulation suggests that there is always a signal that is directly sent from a shallower block ℓ {\textstyle \ell } to a deeper block L {\textstyle L} . ==== Backward propagation ==== The residual learning formulation provides the added benefit of mitigating the vanishing gradient problem to some extent. However, it is crucial to acknowledge that the vanishing gradient issue is not the root cause of the degradation problem, which is tackled through the use of normalization. To observe the effect of residual blocks on backpropagation, consider the partial derivative of a loss function E {\displaystyle {\mathcal {E}}} with respect to some residual block input x ℓ {\displaystyle x_{\ell }} . Using the equation above from forward propagation for a later residual block L > ℓ {\displaystyle L>\ell } : ∂ E ∂ x ℓ = ∂ E ∂ x L ∂ x L ∂ x ℓ = ∂ E ∂ x L ( 1 + ∂ ∂ x ℓ ∑ i = ℓ L − 1 F ( x i ) ) = ∂ E ∂ x L + ∂ E ∂ x L ∂ ∂ x ℓ ∑ i = ℓ L − 1 F ( x i ) {\displaystyle {\begin{aligned}{\frac {\partial {\mathcal {E}}}{\partial x_{\ell }}}&={\frac {\partial {\mathcal {E}}}{\partial x_{L}}}{\frac {\partial x_{L}}{\partial x_{\ell }}}\\&={\frac {\partial {\mathcal {E}}}{\partial x_{L}}}\left(1+{\frac {\partial }{\partial x_{\ell }}}\sum _{i=\ell }^{L-1}F(x_{i})\right)\\&={\frac {\partial {\mathcal {E}}}{\partial x_{L}}}+{\frac {\partial {\mathcal {E}}}{\partial x_{L}}}{\frac {\partial }{\partial x_{\ell }}}\sum _{i=\ell }^{L-1}F(x_{i})\end{aligned}}} This formulation suggests that the gradient computation of a shallower layer, ∂ E ∂ x ℓ {\textstyle {\frac {\partial {\mathcal {E}}}{\partial x_{\ell }}}} , always has a later term ∂ E ∂ x L {\textstyle {\frac {\partial {\mathcal {E}}}{\partial x_{L}}}} that is directly added. Even if the gradients of the F ( x i ) {\displaystyle F(x_{i})} terms are small, the total gradient ∂ E ∂ x ℓ {\textstyle {\frac {\partial {\mathcal {E}}}{\partial x_{\ell }}}} resists vanishing due to the added term ∂ E ∂ x L {\textstyle {\frac {\partial {\mathcal {E}}}{\partial x_{L}}}} . == Variants of residual blocks == === Basic block === A basic block is the simplest building block studied in the original ResNet. This block consists of two sequential 3x3 convolutional layers and a residual connection. The input and output dimensions of both layers are equal. === Bottleneck block === A bottleneck block consists of three sequential convolutional layers and a residual connection. The first layer in this block is a 1×1 convolution for dimension reduction (e.g., to 1/2 of the input dimension); the second layer performs a 3×3 convolution; the last layer is another 1×1 convolution for dimension restoration. The models of ResNet-50, ResNet-101, and ResNet-152 are all based on bottleneck blocks. === Pre-activation block === The pre-activation residual block applies activation functions before applying the residual function F {\displaystyle F} . Formally, the computation of a pre-activation residual block can be written as: x ℓ + 1 = F ( ϕ ( x ℓ ) ) + x ℓ {\displaystyle x_{\ell +1}=F(\phi (x_{\ell }))+x_{\ell }} where ϕ {\displaystyle \phi } can be any activation (e.g. ReLU) or normalization (e.g. LayerNorm) operation. This design reduces the number of non-identity mappings between residual blocks, and allows an identity mapping directly from the input to the output. This design was used to train models with 200 to over 1000 layers, and was found to consistently outperform variants where the residual path is not an identity function. The pre-activation ResNet with 200 layers took 3 weeks to train for ImageNet on 8 GPUs in 2016. Since GPT-2, transformer blocks have been mostly implemented as pre-activation blocks. This is often referred to as "pre-normalization" in the literature of transformer models. == Applications == Originally, ResNet was designed for computer vision. All transformer architectures include residual connections. Indeed, very deep transformers cannot be trained without them. The original ResNet paper made no claim on being inspired by biological systems. However, later research has related ResNet to biologically-plausible algorithms. A study published in Science in 2023 disclosed the complete connectome of an insect brain (specifically that of a fruit fly larva). This study discovered "multilayer shortcuts" that resemble the skip connections in artificial neural networks, including ResNets. == History == === Previous work === Residual connections were noticed in neu
Stockfish (chess)
Stockfish is a free and open-source chess engine, available for various desktop and mobile platforms. It can be used in chess software through the Universal Chess Interface. Stockfish has been one of the strongest chess engines in the world for several years. It has won all main events of the Top Chess Engine Championship (TCEC) and the Chess.com Computer Chess Championship (CCC) since 2020 and, as of May 2026, is the strongest CPU chess engine in the world with an estimated Elo rating of 3653 in a time control of 40/15 (15 minutes to make 40 moves), according to CCRL. The Stockfish engine was developed by Tord Romstad, Marco Costalba, and Joona Kiiski, and was derived from Glaurung, an open-source engine by Tord Romstad released in 2004. It is now being developed and maintained by the Stockfish community. Stockfish historically used only a classical hand-crafted function to evaluate board positions, but with the introduction of the efficiently updatable neural network (NNUE) in August 2020, Stockfish 12 adopted a hybrid evaluation system that primarily used the neural network and occasionally relied on the hand-crafted evaluation. In July 2023, Stockfish removed the hand-crafted evaluation and transitioned to a fully neural network-based approach. == Features == Stockfish uses a tree-search algorithm based on alpha–beta search with several hand-designed heuristics. Stockfish represents positions using bitboards. Stockfish supports Chess960, a feature it inherited from Glaurung. Support for Syzygy tablebases, previously available in a fork maintained by Ronald de Man, was integrated into Stockfish in 2014. In 2018, support for the 7-man Syzygy was added, shortly after the tablebase was made available. Stockfish supports an unlimited number of CPU threads in multiprocessor systems, with a maximum transposition table size of 32 TB. Stockfish has been a very popular engine on various platforms. On desktop, it is the default chess engine bundled with the Internet Chess Club interface programs BlitzIn and Dasher. On mobile, it has been bundled with the Stockfish app, SmallFish and Droidfish. Other Stockfish-compatible graphical user interfaces (GUIs) include Fritz, Arena, Stockfish for Mac, and PyChess. Stockfish can be compiled to WebAssembly or JavaScript, allowing it to run in the browser. Both Chess.com and Lichess provide Stockfish in this form in addition to a server-side program. Release versions and development versions are available as C++ source code and as precompiled versions for Microsoft Windows, macOS, Linux 32-bit/64-bit and Android. == History == The program originated from Glaurung, an open-source chess engine created by Tord Romstad and first released in 2004. Four years later, Marco Costalba forked the project, naming it Stockfish because it was "produced in Norway and cooked in Italy" (Romstad is Norwegian and Costalba is Italian). The first version, Stockfish 1.0, was released in November 2008. For a while, new ideas and code changes were transferred between the two programs in both directions, until Romstad decided to discontinue Glaurung in favor of Stockfish, which was the stronger engine at the time. The last Glaurung version (2.2) was released in December 2008. Around 2011, Romstad decided to abandon his involvement with Stockfish in order to spend more time on his new iOS chess app. On 18 June 2014 Marco Costalba announced that he had "decided to step down as Stockfish maintainer" and asked that the community create a fork of the current version and continue its development. An official repository, managed by a volunteer group of core Stockfish developers, was created soon after and currently manages the development of the project. === Fishtest === Since 2013, Stockfish has been developed using a distributed testing framework named Fishtest, where volunteers can donate CPU time for testing improvements to the program. Changes to game-playing code are accepted or rejected based on results of playing of tens of thousands of games on the framework against an older "reference" version of the program, using sequential probability ratio testing. Tests on the framework are verified using the chi-squared test, and only if the results are statistically significant are they deemed reliable and used to revise the software code. After the inception of Fishtest, Stockfish gained 120 Elo points in 12 months, propelling it to the top of all major rating lists. As of May 2026, the framework has used a total of more than 20,100 years of CPU time to play over 10 billion chess games. === NNUE === In June 2020, Stockfish introduced the efficiently updatable neural network (NNUE) approach, based on earlier work by computer shogi programmers. Instead of using manually designed heuristics to evaluate the board, this approach introduced a neural network trained on millions of positions which could be evaluated quickly on CPU. On 2 September 2020, the twelfth version of Stockfish was released, incorporating NNUE, and reportedly winning ten times more game pairs than it loses when matched against version eleven. In July 2023, the classical evaluation was completely removed in favor of the NNUE evaluation. == Competition results == === Top Chess Engine Championship === Stockfish is a TCEC multiple-time champion and the current leader in trophy count. Ever since TCEC restarted in 2013, Stockfish has finished first or second in every season except one. Stockfish finished second in TCEC Season 4 and 5, with scores of 23–25 first against Houdini 3 and later against Komodo 1142 in the Superfinal event. Season 5 was notable for the winning Komodo team as they accepted the award posthumously for the program's creator Don Dailey, who succumbed to an illness during the final stage of the event. In his honor, the version of Stockfish that was released shortly after that season was named "Stockfish DD". On 30 May 2014, Stockfish 170514 (a development version of Stockfish 5 with tablebase support) convincingly won TCEC Season 6, scoring 35.5–28.5 against Komodo 7x in the Superfinal. Stockfish 5 was released the following day. In TCEC Season 7, Stockfish again made the Superfinal, but lost to Komodo with a score of 30.5–33.5. In TCEC Season 8, despite losses on time caused by buggy code, Stockfish nevertheless qualified once more for the Superfinal, but lost 46.5–53.5 to Komodo. In Season 9, Stockfish defeated Houdini 5 with a score of 54.5–45.5. Stockfish finished third during season 10 of TCEC, the only season since 2013 in which Stockfish had failed to qualify for the superfinal. It did not lose a game but was still eliminated because it was unable to score enough wins against lower-rated engines. After this technical elimination, Stockfish went on a long winning streak, winning seasons 11 (59–41 against Houdini 6.03), 12 (60–40 against Komodo 12.1.1), and 13 (55–45 against Komodo 2155.00) convincingly. In Season 14, Stockfish faced a new challenger in Leela Chess Zero, eking out a win by one point (50.5–49.5). Its winning streak was finally ended in Season 15, when Leela qualified again and won 53.5–46.5, but Stockfish promptly won Season 16, defeating AllieStein 54.5–45.5, after Leela failed to qualify for the Superfinal. In Season 17, Stockfish faced Leela again in the superfinal, losing 52.5–47.5. However, Stockfish has won every Superfinal since: beating Leela 53.5–46.5 in Season 18, 54.5–45.5 in Season 19, 53–47 in Season 20, and 56–44 in Season 21. In Season 22, Komodo Dragon beat out Leela to qualify for the Superfinal, losing to Stockfish by a large margin 59.5–40.5. Stockfish did not lose an opening pair in this match. Leela made the Superfinal in Seasons 23 and 24, but was crushed by Stockfish both times (58.5–41.5 and 58–42). In Season 25, Stockfish once again defeated Leela, but this time by a narrower margin of 52–48. Stockfish also took part in the TCEC cup, winning the first edition, but was surprisingly upset by Houdini in the semifinals of the second edition. Stockfish recovered to beat Komodo in the third-place playoff. In the third edition, Stockfish made it to the finals, but was defeated by Leela Chess Zero after blundering in a 7-man endgame tablebase draw. It turned this result around in the fourth edition, defeating Leela in the final 4.5–3.5. In TCEC Cup 6, Stockfish finished third after losing to AllieStein in the semifinals, the first time it had failed to make the finals. Since then, Stockfish has consistently won the tournament, with the exception of the 11th edition which Leela won 8.5–7.5. === Chess.com Computer Chess Championship === Ever since Chess.com hosted its first Chess.com Computer Chess Championship in 2018, Stockfish has been the most successful engine. It dominated the earlier championships, winning six consecutive titles before finishing second in CCC7. Since then, its dominance has come under threat from the neural-network engines Leelenstein and Leela Chess Zero, but it has continued to perform w
Comet (browser)
Comet is an AI-powered web browser based on Chromium. It was released by Perplexity AI for Microsoft Windows and macOS on July 9, 2025, for Android on November 20, 2025, and for iOS on March 18, 2026. Initial access to the browser was limited to users subscribed to Perplexity's most expensive tier, with broader availability expected over time. The browser was released for free download in October 2025. == Features == Comet is integrated with Perplexity's AI-assisted search engine. The browser features an assistant which enables users to perform a variety of tasks such as generating article summaries, sending emails, or buying products. == Security concerns == Researchers at LayerX Security identified a malicious attack vector which they call CometJacking. The exploit could possibly exfiltrate a user's personal sensitive data to a remote server controlled by the attacker. LayerX attempted to responsibly disclose their findings to Comet's developer Perplexity AI in August 2025. Perplexity responded that they saw no security impact and marked the disclosure as not applicable.
Chris Olah
Christopher Olah (born 1992 or 1993) is a Canadian machine learning researcher and a co-founder of Anthropic. He is known for his work on neural network interpretability, particularly mechanistic interpretability, and for research and tools that visualise internal representations in neural networks. In 2025, Forbes reported he had become a billionaire due to his ownership in Anthropic. == Early life and education == Olah was born in Canada. According to Wired, he left university at age 18 without earning a degree and later received a Thiel Fellowship, which supported him in pursuing independent work. == Career == Olah has worked on interpretability research at Google Brain, OpenAI, and Anthropic. Time called him one of the pioneers of mechanistic interpretability and noted that he pursued this research line first at Google, then at OpenAI, and later at Anthropic, which he co-founded. Wired reported that Olah was involved in neural network visualisation work including DeepDream in 2015, as part of efforts to better understand what neural networks learn. Later coverage linked him to more structured interpretability approaches such as "activation atlases". The Verge covered activation atlases as a collaboration between Google and OpenAI researchers to help inspect neural network representations. At Anthropic, Olah has been identified in major press coverage as leading interpretability work aimed at mapping internal "features" in large language models and relating interpretability findings to AI safety. Quanta Magazine has also quoted Olah in reporting on interpretability and the internal structure of modern language models. Time included Olah in its TIME100 AI list in 2024. === Vatican address on AI ethics === On May 25, 2026, Olah spoke at the Vatican during the official presentation of Magnifica Humanitas, the first encyclical of Pope Leo XIV, which addresses artificial intelligence and human dignity. Olah said AI could lead to large-scale displacement of human labor and exacerbate global inequality. He said the commercial and geopolitical incentives driving frontier AI labs often conflict with the public good, and described AI systems as "grown" rather than strictly engineered. Olah called for external moral oversight from religious institutions, scholars, and civil society to hold the technology sector accountable.
Tail latency
Tail latency is a term used to describe the high-percentile response times seen in a system. This is usually measured at the 95th, 99th, or 99.9th percentile, not the average latency. In distributed systems, cloud computing, and large-scale web services, even a small number of slow requests can make the user experience and system performance much worse. Tail latency often happens because of things like resource contention, network variability, garbage collection pauses, and hardware heterogeneity. A major problem in system design is managing tail latency, because lowering average latency doesn't always make the worst-case performance better. To lessen its effects, people often use techniques like request hedging, replication, load balancing, and adaptive timeouts. In latency-sensitive applications like search engines, financial systems, and real-time services, where service-level objectives (SLOs) are often based on high-percentile latencies, it is especially important to understand and improve tail latency.
Open Neural Network Exchange
The Open Neural Network Exchange (ONNX) [ˈɒnɪks] is an open-source artificial intelligence ecosystem of technology companies and research organizations that establish open standards for representing machine learning algorithms and software tools to enable a standard format for representing machine learning models. ONNX is available on GitHub. == History == ONNX was originally named Toffee and was developed by the PyTorch team at Facebook. In September 2017 it was renamed to ONNX and announced by Facebook and Microsoft. Later, IBM, Huawei, Intel, AMD, Arm and Qualcomm announced support for the initiative. In October 2017, Microsoft announced that it would add its Cognitive Toolkit and Project Brainwave platform to the initiative. In November 2019 ONNX was accepted as graduate project in Linux Foundation AI. In October 2020 Zetane Systems became a member of the ONNX ecosystem. == Intent == The initiative targets: === Framework interoperability === Enable developers to move machine learning models between different frameworks, which may be used at different stages of the development process, such as training, architecture design, or deployment on mobile devices. === Shared optimization === Provide a common representation that can be used by hardware vendors and other developers to apply optimizations to artificial neural network models across multiple machine learning frameworks. == Contents == ONNX provides definitions of an extensible computation graph model, built-in operators and standard data types, focused on inferencing (evaluation).. The container format is Protocol Buffers. Each computation dataflow graph is a list of nodes that form an acyclic graph. Nodes have inputs and outputs. Each node is a call to an operator. Metadata documents the graph. Built-in operators are to be available on each ONNX-supporting framework. ONNX models can be trained in a single framework, such as PyTorch or TensorFlow, and then exported to ONNX. This format allows models to be transferred from the training framework to other environments for testing or deployment. Once a model is in ONNX format, it can be executed in different runtime systems or on various hardware platforms, such as GPUs or specialized AI accelerators. Using a common format enables the same model representation to be used across multiple systems and frameworks.