Upload folder using huggingface_hub

0.codes.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:dc67ee4f6cccc41f215d2328f806590d32dc07ae35b669f3ba792182645a1cb4
-size 2440476
+oid sha256:5b0afe0ba1ce8b78962e0a76d8a747069dd7a9ec472fdd29d7c7562ba67ed6bd
+size 2453596

0.metadata.json

@@ -1,6 +1,6 @@
 {
     "passage_offset": 0,
-    "num_passages": 3543,
-    "num_embeddings": 609827,
+    "num_passages": 3562,
+    "num_embeddings": 613116,
     "embedding_offset": 0
 }

0.residuals.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f2ff523423cd1c29f5787ec6d5af7fc89786e338eee29d5da187bb19a8f55a5a
-size 78059056
+oid sha256:684bfb55795333e7f4080887f3188db88cd47e9d4886fe6206d22c32709f4228
+size 78480048

avg_residual.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d0f91567f64b689fa3c82d58a5396b6a2859b8881fabb5f3f6c0c6356945d910
+oid sha256:1fce5300a231680c4333b453d274bc826806178b8298b82e0ecaea173f6e23f4
 size 1205

buckets.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3bccc29a811e81449ed5dd1c80c13dcb7f4f17d1235184974317e7acbb4ebd28
+oid sha256:32ef0207b2d1fb9bf0b2aaf34c7881ff7aa712878c0f88e855b6a67f6d8bdd7b
 size 2904

centroids.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a2e73f166e4ee2d91a5b0b46169ab528db6d1a49a20c74cfb771a00ebada13d9
+oid sha256:d8a9431a36e0a18bce161d31a867a0a4bdf4d66615642a498ca07dd0e2ef948c
 size 2098342

collection.json

@@ -3541,5 +3541,24 @@
   "Since dynamic scenes always introduce new content, we also propose a new pixel-wise densification strategy for Gaussian points to integrate new visual content. Moreover, GFlow transcends the boundaries of mere 4D reconstruction; it also enables tracking of any points across frames without the need for prior training and segments moving objects from the scene in an unsupervised way. Additionally, the camera poses of each frame can be derived from GFlow, allowing for rendering novel views of a video scene through changing camera pose. By employing the explicit representation, we may readily conduct scene-level or object-level editing as desired, underscoring its versatility and power. Visit our project website at: https://littlepure2333.github.io/GFlow",
   "Recent advances in text-to-music editing, which employ text queries to modify music (e.g.\\ by changing its style or adjusting instrumental components), present unique challenges and opportunities for AI-assisted music creation. Previous approaches in this domain have been constrained by the necessity to train specific editing models from scratch, which is both resource-intensive and inefficient; other research uses large language models to predict edited music, resulting in imprecise audio reconstruction. To Combine the strengths and address these limitations, we introduce Instruct-MusicGen, a novel approach that finetunes a pretrained MusicGen model to efficiently follow editing instructions such as adding, removing, or separating stems. Our approach involves a modification of the original MusicGen architecture by incorporating a text fusion module and an audio fusion module, which allow the model to process instruction texts and audio inputs concurrently and yield the desired edited music. Remarkably, Instruct-MusicGen only introduces 8% new parameters to the original MusicGen model and only trains for 5K steps, yet it achieves superior performance across all tasks compared to existing baselines, and demonstrates performance comparable to the models trained for specific tasks.",
   "Remarkably, Instruct-MusicGen only introduces 8% new parameters to the original MusicGen model and only trains for 5K steps, yet it achieves superior performance across all tasks compared to existing baselines, and demonstrates performance comparable to the models trained for specific tasks. This advancement not only enhances the efficiency of text-to-music editing but also broadens the applicability of music language models in dynamic music production environments.",
-  "Scene image editing is crucial for entertainment, photography, and advertising design. Existing methods solely focus on either 2D individual object or 3D global scene editing. This results in a lack of a unified approach to effectively control and manipulate scenes at the 3D level with different levels of granularity. In this work, we propose 3DitScene, a novel and unified scene editing framework leveraging language-guided disentangled Gaussian Splatting that enables seamless editing from 2D to 3D, allowing precise control over scene composition and individual objects. We first incorporate 3D Gaussians that are refined through generative priors and optimization techniques. Language features from CLIP then introduce semantics into 3D geometry for object disentanglement. With the disentangled Gaussians, 3DitScene allows for manipulation at both the global and individual levels, revolutionizing creative expression and empowering control over scenes and objects. Experimental results demonstrate the effectiveness and versatility of 3DitScene in scene image editing. Code and online demo can be found at our project homepage: https://zqh0253.github.io/3DitScene/."
+  "Scene image editing is crucial for entertainment, photography, and advertising design. Existing methods solely focus on either 2D individual object or 3D global scene editing. This results in a lack of a unified approach to effectively control and manipulate scenes at the 3D level with different levels of granularity. In this work, we propose 3DitScene, a novel and unified scene editing framework leveraging language-guided disentangled Gaussian Splatting that enables seamless editing from 2D to 3D, allowing precise control over scene composition and individual objects. We first incorporate 3D Gaussians that are refined through generative priors and optimization techniques. Language features from CLIP then introduce semantics into 3D geometry for object disentanglement. With the disentangled Gaussians, 3DitScene allows for manipulation at both the global and individual levels, revolutionizing creative expression and empowering control over scenes and objects. Experimental results demonstrate the effectiveness and versatility of 3DitScene in scene image editing. Code and online demo can be found at our project homepage: https://zqh0253.github.io/3DitScene/.",
+  "Large Language Models (LLMs) have made great strides in recent years to achieve unprecedented performance across different tasks. However, due to commercial interest, the most competitive models like GPT, Gemini, and Claude have been gated behind proprietary interfaces without disclosing the training details. Recently, many institutions have open-sourced several strong LLMs like LLaMA-3, comparable to existing closed-source LLMs. However, only the model's weights are provided with most details (e.g., intermediate checkpoints, pre-training corpus, and training code, etc.) being undisclosed. To improve the transparency of LLMs, the research community has formed to open-source truly open LLMs (e.g., Pythia, Amber, OLMo), where more details (e.g., pre-training corpus and training code) are being provided. These models have greatly advanced the scientific study of these large models including their strengths, weaknesses, biases and risks. However, we observe that the existing truly open LLMs on reasoning, knowledge, and coding tasks are still inferior to existing state-of-the-art LLMs with similar model sizes.",
+  "These models have greatly advanced the scientific study of these large models including their strengths, weaknesses, biases and risks. However, we observe that the existing truly open LLMs on reasoning, knowledge, and coding tasks are still inferior to existing state-of-the-art LLMs with similar model sizes. To this end, we open-source MAP-Neo, a highly capable and transparent bilingual language model with 7B parameters trained from scratch on 4.5T high-quality tokens. Our MAP-Neo is the first fully open-sourced bilingual LLM with comparable performance compared to existing state-of-the-art LLMs. Moreover, we open-source all details to reproduce our MAP-Neo, where the cleaned pre-training corpus, data cleaning pipeline, checkpoints, and well-optimized training/evaluation framework are provided. Finally, we hope our MAP-Neo will enhance and strengthen the open research community and inspire more innovations and creativities to facilitate the further improvements of LLMs.",
+  "This paper examines the extent to which large language models (LLMs) have developed higher-order theory of mind (ToM); the human ability to reason about multiple mental and emotional states in a recursive manner (e.g. I think that you believe that she knows). This paper builds on prior work by introducing a handwritten test suite -- Multi-Order Theory of Mind Q&A -- and using it to compare the performance of five LLMs to a newly gathered adult human benchmark. We find that GPT-4 and Flan-PaLM reach adult-level and near adult-level performance on ToM tasks overall, and that GPT-4 exceeds adult performance on 6th order inferences. Our results suggest that there is an interplay between model size and finetuning for the realisation of ToM abilities, and that the best-performing LLMs have developed a generalised capacity for ToM. Given the role that higher-order ToM plays in a wide range of cooperative and competitive human behaviours, these findings have significant implications for user-facing LLM applications.",
+  "Diffusion-based text-to-video (T2V) models have achieved significant success but continue to be hampered by the slow sampling speed of their iterative sampling processes. To address the challenge, consistency models have been proposed to facilitate fast inference, albeit at the cost of sample quality. In this work, we aim to break the quality bottleneck of a video consistency model (VCM) to achieve both fast and high-quality video generation. We introduce T2V-Turbo, which integrates feedback from a mixture of differentiable reward models into the consistency distillation (CD) process of a pre-trained T2V model. Notably, we directly optimize rewards associated with single-step generations that arise naturally from computing the CD loss, effectively bypassing the memory constraints imposed by backpropagating gradients through an iterative sampling process. Remarkably, the 4-step generations from our T2V-Turbo achieve the highest total score on VBench, even surpassing Gen-2 and Pika. We further conduct human evaluations to corroborate the results, validating that the 4-step generations from our T2V-Turbo are preferred over the 50-step DDIM samples from their teacher models, representing more than a tenfold acceleration while improving video generation quality.",
+  "The creation of high-fidelity, digital versions of human heads is an important stepping stone in the process of further integrating virtual components into our everyday lives. Constructing such avatars is a challenging research problem, due to a high demand for photo-realism and real-time rendering performance. In this work, we propose Neural Parametric Gaussian Avatars (NPGA), a data-driven approach to create high-fidelity, controllable avatars from multi-view video recordings. We build our method around 3D Gaussian Splatting for its highly efficient rendering and to inherit the topological flexibility of point clouds. In contrast to previous work, we condition our avatars' dynamics on the rich expression space of neural parametric head models (NPHM), instead of mesh-based 3DMMs. To this end, we distill the backward deformation field of our underlying NPHM into forward deformations which are compatible with rasterization-based rendering. All remaining fine-scale, expression-dependent details are learned from the multi-view videos. To increase the representational capacity of our avatars, we augment the canonical Gaussian point cloud using per-primitive latent features which govern its dynamic behavior.",
+  "All remaining fine-scale, expression-dependent details are learned from the multi-view videos. To increase the representational capacity of our avatars, we augment the canonical Gaussian point cloud using per-primitive latent features which govern its dynamic behavior. To regularize this increased dynamic expressivity, we propose Laplacian terms on the latent features and predicted dynamics. We evaluate our method on the public NeRSemble dataset, demonstrating that NPGA significantly outperforms the previous state-of-the-art avatars on the self-reenactment task by 2.6 PSNR. Furthermore, we demonstrate accurate animation capabilities from real-world monocular videos.",
+  "Sound content is an indispensable element for multimedia works such as video games, music, and films. Recent high-quality diffusion-based sound generation models can serve as valuable tools for the creators. However, despite producing high-quality sounds, these models often suffer from slow inference speeds. This drawback burdens creators, who typically refine their sounds through trial and error to align them with their artistic intentions. To address this issue, we introduce Sound Consistency Trajectory Models (SoundCTM). Our model enables flexible transitioning between high-quality 1-step sound generation and superior sound quality through multi-step generation. This allows creators to initially control sounds with 1-step samples before refining them through multi-step generation. While CTM fundamentally achieves flexible 1-step and multi-step generation, its impressive performance heavily depends on an additional pretrained feature extractor and an adversarial loss, which are expensive to train and not always available in other domains. Thus, we reframe CTM's training framework and introduce a novel feature distance by utilizing the teacher's network for a distillation loss. Additionally, while distilling classifier-free guided trajectories, we train conditional and unconditional student models simultaneously and interpolate between these models during inference.",
+  "Thus, we reframe CTM's training framework and introduce a novel feature distance by utilizing the teacher's network for a distillation loss. Additionally, while distilling classifier-free guided trajectories, we train conditional and unconditional student models simultaneously and interpolate between these models during inference. We also propose training-free controllable frameworks for SoundCTM, leveraging its flexible sampling capability. SoundCTM achieves both promising 1-step and multi-step real-time sound generation without using any extra off-the-shelf networks. Furthermore, we demonstrate SoundCTM's capability of controllable sound generation in a training-free manner.",
+  "Existing diffusion-based text-to-3D generation methods primarily focus on producing visually realistic shapes and appearances, often neglecting the physical constraints necessary for downstream tasks. Generated models frequently fail to maintain balance when placed in physics-based simulations or 3D printed. This balance is crucial for satisfying user design intentions in interactive gaming, embodied AI, and robotics, where stable models are needed for reliable interaction. Additionally, stable models ensure that 3D-printed objects, such as figurines for home decoration, can stand on their own without requiring additional supports. To fill this gap, we introduce Atlas3D, an automatic and easy-to-implement method that enhances existing Score Distillation Sampling (SDS)-based text-to-3D tools. Atlas3D ensures the generation of self-supporting 3D models that adhere to physical laws of stability under gravity, contact, and friction. Our approach combines a novel differentiable simulation-based loss function with physically inspired regularization, serving as either a refinement or a post-processing module for existing frameworks. We verify Atlas3D's efficacy through extensive generation tasks and validate the resulting 3D models in both simulated and real-world environments.",
+  "This paper presents EasyAnimate, an advanced method for video generation that leverages the power of transformer architecture for high-performance outcomes. We have expanded the DiT framework originally designed for 2D image synthesis to accommodate the complexities of 3D video generation by incorporating a motion module block. It is used to capture temporal dynamics, thereby ensuring the production of consistent frames and seamless motion transitions. The motion module can be adapted to various DiT baseline methods to generate video with different styles. It can also generate videos with different frame rates and resolutions during both training and inference phases, suitable for both images and videos. Moreover, we introduce slice VAE, a novel approach to condense the temporal axis, facilitating the generation of long duration videos. Currently, EasyAnimate exhibits the proficiency to generate videos with 144 frames. We provide a holistic ecosystem for video production based on DiT, encompassing aspects such as data pre-processing, VAE training, DiT models training (both the baseline model and LoRA model), and end-to-end video inference. Code is available at: https://github.com/aigc-apps/EasyAnimate. We are continuously working to enhance the performance of our method.",
+  "The dominant framework for alignment of large language models (LLM), whether through reinforcement learning from human feedback or direct preference optimisation, is to learn from preference data. This involves building datasets where each element is a quadruplet composed of a prompt, two independent responses (completions of the prompt) and a human preference between the two independent responses, yielding a preferred and a dis-preferred response. Such data is typically scarce and expensive to collect. On the other hand, single-trajectory datasets where each element is a triplet composed of a prompt, a response and a human feedback is naturally more abundant. The canonical element of such datasets is for instance an LLM's response to a user's prompt followed by a user's feedback such as a thumbs-up/down. Consequently, in this work, we propose DRO, or Direct Reward Optimisation, as a framework and associated algorithms that do not require pairwise preferences. DRO uses a simple mean-squared objective that can be implemented in various ways. We validate our findings empirically, using T5 encoder-decoder language models, and show DRO's performance over selected baselines such as Kahneman-Tversky Optimization (KTO).",
+  "DRO uses a simple mean-squared objective that can be implemented in various ways. We validate our findings empirically, using T5 encoder-decoder language models, and show DRO's performance over selected baselines such as Kahneman-Tversky Optimization (KTO). Thus, we confirm that DRO is a simple and empirically compelling method for single-trajectory policy optimisation.",
+  "Integrating multiple generative foundation models, especially those trained on different modalities, into something greater than the sum of its parts poses significant challenges. Two key hurdles are the availability of aligned data (concepts that contain similar meaning but is expressed differently in different modalities), and effectively leveraging unimodal representations in cross-domain generative tasks, without compromising their original unimodal capabilities.   We propose Zipper, a multi-tower decoder architecture that addresses these concerns by using cross-attention to flexibly compose multimodal generative models from independently pre-trained unimodal decoders. In our experiments fusing speech and text modalities, we show the proposed architecture performs very competitively in scenarios with limited aligned text-speech data. We also showcase the flexibility of our model to selectively maintain unimodal (e.g., text-to-text generation) generation performance by freezing the corresponding modal tower (e.g. text). In cross-modal tasks such as automatic speech recognition (ASR) where the output modality is text, we show that freezing the text backbone results in negligible performance degradation.",
+  "text). In cross-modal tasks such as automatic speech recognition (ASR) where the output modality is text, we show that freezing the text backbone results in negligible performance degradation. In cross-modal tasks such as text-to-speech generation (TTS) where the output modality is speech, we show that using a pre-trained speech backbone results in superior performance to the baseline.",
+  "Large language models (LLMs) often hallucinate and lack the ability to provide attribution for their generations. Semi-parametric LMs, such as kNN-LM, approach these limitations by refining the output of an LM for a given prompt using its nearest neighbor matches in a non-parametric data store. However, these models often exhibit slow inference speeds and produce non-fluent texts. In this paper, we introduce Nearest Neighbor Speculative Decoding (NEST), a novel semi-parametric language modeling approach that is capable of incorporating real-world text spans of arbitrary length into the LM generations and providing attribution to their sources. NEST performs token-level retrieval at each inference step to compute a semi-parametric mixture distribution and identify promising span continuations in a corpus. It then uses an approximate speculative decoding procedure that accepts a prefix of the retrieved span or generates a new token. NEST significantly enhances the generation quality and attribution rate of the base LM across a variety of knowledge-intensive tasks, surpassing the conventional kNN-LM method and performing competitively with in-context retrieval augmentation. In addition, NEST substantially improves the generation speed, achieving a 1.8x speedup in inference time when applied to Llama-2-Chat 70B.",
+  "Reinforcement learning from human feedback (RLHF) has demonstrated great promise in aligning large language models (LLMs) with human preference. Depending on the availability of preference data, both online and offline RLHF are active areas of investigation. A key bottleneck is understanding how to incorporate uncertainty estimation in the reward function learned from the preference data for RLHF, regardless of how the preference data is collected. While the principles of optimism or pessimism under uncertainty are well-established in standard reinforcement learning (RL), a practically-implementable and theoretically-grounded form amenable to large language models is not yet available, as standard techniques for constructing confidence intervals become intractable under arbitrary policy parameterizations.   In this paper, we introduce a unified approach to online and offline RLHF -- value-incentivized preference optimization (VPO) -- which regularizes the maximum-likelihood estimate of the reward function with the corresponding value function, modulated by a sign to indicate whether the optimism or pessimism is chosen. VPO also directly optimizes the policy with implicit reward modeling, and therefore shares a simpler RLHF pipeline similar to direct preference optimization.",
+  "VPO also directly optimizes the policy with implicit reward modeling, and therefore shares a simpler RLHF pipeline similar to direct preference optimization. Theoretical guarantees of VPO are provided for both online and offline settings, matching the rates of their standard RL counterparts. Moreover, experiments on text summarization and dialog verify the practicality and effectiveness of VPO.",
+  "Preference optimization, particularly through Reinforcement Learning from Human Feedback (RLHF), has achieved significant success in aligning Large Language Models (LLMs) to adhere to human intentions. Unlike offline alignment with a fixed dataset, online feedback collection from humans or AI on model generations typically leads to more capable reward models and better-aligned LLMs through an iterative process. However, achieving a globally accurate reward model requires systematic exploration to generate diverse responses that span the vast space of natural language. Random sampling from standard reward-maximizing LLMs alone is insufficient to fulfill this requirement. To address this issue, we propose a bilevel objective optimistically biased towards potentially high-reward responses to actively explore out-of-distribution regions. By solving the inner-level problem with the reparameterized reward function, the resulting algorithm, named Self-Exploring Language Models (SELM), eliminates the need for a separate RM and iteratively updates the LLM with a straightforward objective. Compared to Direct Preference Optimization (DPO), the SELM objective reduces indiscriminate favor of unseen extrapolations and enhances exploration efficiency.",
+  "Compared to Direct Preference Optimization (DPO), the SELM objective reduces indiscriminate favor of unseen extrapolations and enhances exploration efficiency. Our experimental results demonstrate that when finetuned on Zephyr-7B-SFT and Llama-3-8B-Instruct models, SELM significantly boosts the performance on instruction-following benchmarks such as MT-Bench and AlpacaEval 2.0, as well as various standard academic benchmarks in different settings. Our code and models are available at https://github.com/shenao-zhang/SELM."
 ]

doclens.0.json

@@ -1 +1 @@
-[178,205,218,148,184,163,221,185,200,228,172,155,210,222,88,206,226,67,132,212,91,206,104,212,174,205,132,159,230,175,216,198,227,190,212,198,122,213,169,204,92,197,118,191,191,224,69,219,197,72,218,77,175,111,155,217,220,170,231,91,221,217,95,146,177,123,195,205,151,209,207,36,202,200,226,176,232,53,167,199,89,184,213,104,154,153,216,214,215,174,205,72,211,78,221,212,232,223,73,158,220,158,202,222,189,165,205,175,222,132,126,179,219,110,209,158,208,98,176,192,226,34,158,205,126,178,224,182,227,100,152,191,169,195,163,172,208,117,199,217,167,217,157,163,194,217,200,217,23,221,209,146,150,204,200,125,215,232,68,147,212,41,223,178,152,173,210,139,198,182,196,207,95,176,205,223,83,216,85,207,210,52,177,178,230,197,119,226,99,182,210,212,77,138,199,123,179,111,219,69,223,65,204,215,83,197,87,211,132,216,135,178,157,166,216,85,170,195,208,190,175,134,220,200,67,221,211,66,227,222,226,190,209,205,67,207,139,208,127,186,205,168,221,179,223,117,148,221,216,80,189,125,199,202,64,218,77,195,190,221,181,98,143,214,220,97,187,127,219,122,216,87,138,212,194,112,219,227,101,220,100,164,234,109,221,102,223,89,184,205,219,77,188,223,172,171,175,152,175,137,213,197,114,205,221,138,181,174,227,73,147,144,178,147,215,152,182,204,80,210,123,211,121,209,224,224,219,211,163,133,187,148,151,163,221,94,133,213,72,187,224,216,162,154,224,184,118,204,220,154,117,220,162,202,223,195,110,197,151,224,88,182,217,221,214,118,218,118,164,205,97,221,183,154,206,197,74,170,219,103,230,215,192,224,78,184,72,201,227,221,191,181,104,190,224,221,99,123,206,102,228,202,74,195,96,225,176,232,231,96,225,206,61,173,101,190,211,90,213,199,203,184,209,173,160,207,203,99,201,195,132,195,214,148,211,227,192,215,212,127,162,213,114,178,111,207,71,129,182,212,75,176,209,137,213,224,220,70,232,116,228,179,200,184,222,93,202,71,219,123,213,119,204,173,135,118,207,96,216,107,210,202,189,172,211,76,146,195,169,83,227,62,219,223,125,158,202,226,98,214,225,76,156,211,204,110,190,224,200,64,228,223,72,213,102,214,178,219,128,211,127,187,213,160,216,134,202,150,186,201,195,230,199,206,193,205,219,216,84,213,223,192,222,109,187,209,79,153,204,224,72,167,190,183,183,228,133,227,103,197,225,203,205,201,207,213,210,78,138,195,133,195,180,190,185,207,189,100,191,229,198,129,226,135,205,164,226,190,73,200,231,209,216,147,221,228,165,213,209,181,157,222,117,216,58,148,219,92,222,219,63,191,218,186,197,158,172,204,109,204,102,223,229,91,192,217,186,204,144,173,209,95,186,222,225,178,183,222,71,193,182,232,63,227,228,82,209,162,219,183,222,122,212,195,60,219,130,223,69,216,111,198,224,80,223,74,164,148,227,193,216,139,207,198,148,227,199,169,230,225,86,138,165,206,91,213,164,215,229,173,216,210,92,166,205,102,196,78,202,74,187,202,229,191,222,57,206,115,183,94,225,173,198,109,212,224,126,215,44,221,80,196,95,177,218,217,198,109,218,191,221,214,227,71,207,123,206,90,219,218,108,181,214,123,150,204,214,208,225,199,191,192,213,93,217,149,178,199,106,204,80,208,60,131,132,232,218,213,115,202,94,175,227,217,114,200,222,178,180,185,192,193,205,173,226,217,222,197,184,205,211,136,214,221,62,213,89,169,184,135,198,98,216,225,71,230,133,213,154,216,189,122,224,108,204,88,207,227,93,191,204,214,66,153,219,209,203,218,237,229,222,148,174,218,107,214,169,223,216,173,220,148,199,188,175,203,142,188,210,98,167,174,224,103,156,97,220,111,216,70,130,221,216,115,166,164,162,210,216,69,173,203,77,216,175,172,210,98,199,213,88,221,82,184,131,201,220,133,211,116,156,219,213,227,128,221,237,172,217,64,180,221,70,169,220,216,134,215,177,190,197,229,221,226,74,189,190,219,86,222,74,213,106,184,182,205,221,229,143,201,198,162,209,128,221,209,60,216,146,228,106,215,154,214,202,82,194,141,168,217,213,80,140,114,220,149,229,215,150,211,170,199,109,218,108,206,125,204,164,183,148,193,153,188,143,160,199,210,124,225,105,178,75,222,112,205,75,199,83,222,215,158,104,205,70,215,223,158,220,99,124,194,91,219,65,209,149,234,188,222,218,200,66,208,101,208,188,201,207,198,160,205,219,118,170,195,141,128,182,189,211,51,179,216,112,152,196,224,80,223,177,218,127,186,113,151,117,159,184,200,231,212,134,222,68,216,204,90,151,202,158,173,146,144,209,222,89,171,188,206,215,117,191,223,80,215,68,151,159,210,74,208,105,204,105,174,228,67,147,202,214,41,222,163,183,142,227,188,200,208,132,157,191,195,94,166,201,155,214,132,185,195,165,214,219,156,217,100,147,220,105,220,223,225,117,176,166,207,134,218,202,172,183,217,76,229,139,217,87,208,181,214,169,154,199,231,208,200,157,213,222,49,207,191,139,219,218,82,122,154,137,190,176,192,180,231,80,208,136,185,112,189,187,192,208,219,110,208,102,186,225,220,218,235,214,96,133,198,199,95,152,83,219,190,124,148,222,213,207,145,217,168,200,32,208,198,171,205,218,104,218,136,230,77,139,178,212,214,129,211,233,93,221,95,155,153,221,151,201,108,193,216,172,208,72,177,217,208,225,60,81,172,234,180,146,184,180,199,222,224,182,106,183,105,221,208,81,221,137,211,44,198,224,205,82,187,164,147,213,163,229,192,219,218,190,228,221,218,116,231,97,182,126,213,193,173,207,103,207,122,161,156,193,153,192,92,198,155,223,224,106,223,170,177,225,40,206,203,221,152,208,116,203,105,197,221,65,226,111,224,215,215,142,154,211,84,184,135,169,212,213,76,229,225,130,207,85,209,177,116,140,150,229,209,105,164,186,214,219,62,148,203,230,118,207,216,217,211,114,201,134,217,87,197,124,209,186,184,164,189,163,212,85,219,106,209,220,168,221,212,112,189,183,229,200,192,106,224,114,220,95,141,223,222,219,226,79,187,202,212,222,193,128,223,209,160,178,197,222,111,152,114,164,206,179,189,183,122,212,149,150,130,214,115,195,118,204,208,120,195,163,159,220,125,222,174,211,144,140,220,80,173,204,87,215,149,227,97,229,221,119,213,139,212,82,215,68,213,136,186,168,202,85,213,93,168,218,63,222,121,213,90,218,94,197,221,182,220,112,224,221,83,215,194,138,181,197,221,91,186,194,99,179,213,210,76,188,214,216,79,213,92,220,94,201,108,213,131,221,153,219,74,159,172,218,217,219,108,209,35,217,112,157,206,148,211,126,214,107,191,105,164,224,220,126,204,201,211,119,203,129,165,205,130,218,137,211,178,204,215,206,119,209,75,193,182,115,178,207,217,220,184,199,167,189,230,140,208,167,173,195,136,220,205,64,186,159,194,75,213,189,110,189,208,120,176,182,162,196,183,137,134,198,172,193,184,172,212,189,214,200,95,167,199,110,195,116,180,204,66,217,126,222,226,193,166,143,121,218,224,196,140,226,142,206,66,214,140,218,144,201,221,152,204,154,221,101,225,69,216,212,128,181,200,180,206,136,193,226,226,219,141,168,202,209,105,178,128,208,197,145,163,218,124,220,147,220,156,211,69,212,226,132,223,165,211,110,229,194,193,204,208,90,218,106,152,215,134,180,182,204,211,80,211,108,204,102,208,102,192,94,191,206,203,140,165,213,197,105,186,186,210,62,208,165,213,174,186,115,211,164,193,128,201,169,217,167,217,174,222,167,129,170,217,218,182,159,113,208,85,219,126,219,89,172,213,212,69,205,135,124,205,67,208,52,186,160,225,125,190,65,227,168,180,166,151,194,205,88,138,204,110,124,223,80,193,106,179,183,163,220,148,205,218,200,221,198,202,190,213,87,227,239,81,213,115,211,219,201,204,105,216,83,210,95,170,219,197,86,228,86,209,125,231,83,198,194,171,144,183,222,216,216,226,204,109,216,115,180,227,116,207,223,84,190,215,117,212,79,214,230,218,134,208,92,209,99,181,220,61,212,85,199,185,174,171,220,221,57,146,224,214,81,183,218,152,163,193,72,203,111,130,222,204,217,229,209,201,189,220,213,72,192,214,221,157,197,124,209,100,223,162,223,73,85,221,68,210,46,205,76,118,160,180,200,85,218,34,196,170,204,88,168,224,71,188,207,166,181,210,101,148,215,76,219,217,104,199,180,210,208,69,201,202,216,224,89,206,88,128,125,196,203,106,183,182,84,228,200,120,171,166,153,215,210,82,210,65,163,131,213,83,180,205,92,152,228,99,212,212,209,94,215,218,95,206,179,236,219,204,102,196,110,160,218,140,228,102,218,177,172,193,95,209,215,203,220,78,213,191,189,168,204,218,192,89,213,223,208,230,168,202,169,222,56,155,203,73,197,156,208,222,127,212,223,160,217,201,223,205,217,100,221,208,143,218,22,208,224,86,209,90,220,144,182,139,219,79,194,183,225,216,209,190,102,221,220,203,152,215,113,230,106,212,221,106,197,111,185,211,83,162,200,208,219,124,210,92,161,216,221,79,208,166,211,142,179,222,197,79,185,226,182,193,195,170,202,113,192,231,228,197,180,152,223,155,215,97,213,85,205,145,200,89,209,201,139,217,163,133,206,174,112,205,213,217,123,201,136,132,221,206,105,200,139,183,160,215,134,221,208,181,208,101,210,225,116,183,217,80,186,120,162,194,201,97,188,209,174,122,229,214,205,120,218,228,228,83,134,201,119,171,210,55,227,107,225,64,174,203,130,213,226,95,206,157,139,154,171,139,224,219,154,190,206,218,217,221,86,135,217,213,183,118,187,186,128,212,79,184,186,156,188,134,216,172,155,195,210,176,223,125,128,217,213,214,217,94,189,192,133,167,85,187,217,60,158,221,122,217,117,224,95,206,89,199,74,205,210,70,209,83,196,77,169,188,228,114,188,105,229,182,174,223,59,181,205,127,225,222,128,190,219,112,197,115,222,205,115,184,214,210,238,196,212,85,207,86,183,215,178,187,112,194,176,212,90,217,80,164,179,187,138,221,214,203,96,227,198,82,209,207,96,208,76,228,224,205,207,186,178,221,224,189,143,229,182,229,132,213,201,186,218,67,222,221,83,216,44,220,68,214,103,137,210,48,210,211,88,222,169,225,155,181,200,207,62,219,152,179,130,215,86,183,90,202,164,179,205,128,219,224,148,190,117,213,221,184,175,146,223,216,82,230,92,201,231,218,172,207,209,189,150,216,142,226,46,197,219,92,228,168,216,111,221,204,216,190,180,189,206,108,204,101,224,74,223,65,181,142,211,203,122,220,214,66,179,185,222,116,120,209,222,101,172,209,232,56,223,177,212,107,191,168,193,193,105,212,105,163,209,191,117,194,170,211,190,121,221,90,161,207,67,226,90,216,60,171,219,89,220,184,214,194,144,182,214,208,196,133,226,214,77,190,215,63,115,208,92,227,204,68,197,77,178,188,204,216,202,194,152,175,181,208,131,196,223,128,177,116,207,102,220,121,216,199,202,207,74,211,196,217,177,214,218,221,93,219,192,215,143,139,183,226,83,102,219,211,100,194,47,119,204,160,143,180,149,183,226,96,172,212,186,198,207,182,207,221,164,146,224,61,200,140,147,196,214,60,226,215,111,216,186,180,161,234,196,124,208,214,221,189,178,112,168,208,136,169,181,202,124,201,74,211,143,192,175,119,163,202,67,191,172,148,179,223,122,217,117,180,190,186,209,72,193,182,218,80,156,138,201,202,218,208,149,190,109,224,64,192,114,219,199,210,208,97,232,205,197,188,189,87,213,152,196,164,131,225,94,219,205,163,191,172,196,189,206,110,201,73,191,122,210,173,208,73,221,136,222,185,224,74,213,86,185,221,170,210,69,165,156,210,102,211,210,221,197,209,154,127,212,180,208,139,231,207,50,166,96,150,158,206,228,214,202,95,162,220,191,136,217,54,155,201,140,179,191,91,213,220,74,145,216,232,45,208,217,209,182,160,182,100,221,155,219,227,160,180,209,147,174,212,67,209,82,213,148,208,51,176,88,210,71,117,174,218,82,211,188,170,210,186,136,176,220,157,189,167,190,212,160,212,135,201,219,49,162,165,209,117,175,213,152,176,220,221,124,150,204,220,73,228,194,218,57,195,173,159,173,175,206,176,229,79,164,201,203,152,214,116,137,219,66,222,214,80,175,202,90,225,113,219,206,78,190,89,214,50,209,154,188,227,194,157,195,74,186,206,130,198,73,212,60,204,122,222,99,205,196,229,213,83,230,108,171,126,192,104,216,207,117,217,197,214,79,207,110,221,79,217,144,206,160,206,172,197,183,207,217,207,113,210,221,71,161,221,164,227,214,142,177,185,180,103,130,198,123,205,74,216,102,219,160,217,75,204,114,192,213,166,188,118,222,227,92,195,219,161,200,221,69,203,143,198,198,217,198,66,212,50,208,116,199,125,210,207,167,225,116,207,97,184,99,220,184,203,184,219,177,167,202,214,55,207,161,197,122,212,226,187,96,216,201,188,135,224,207,139,225,230,220,121,221,107,212,66,170,169,210,199,102,220,94,159,184,207,92,207,231,214,125,227,220,205,58,193,203,215,223,229,78,196,170,185,196,162,234,56,201,123,171,231,196,86,162,199,213,220,68,200,68,205,88,225,135,220,82,182,215,222,79,152,230,62,162,218,184,224,67,206,99,189,124,214,197,73,204,105,221,179,102,218,232,80,214,181,170,204,165,216,207,217,212,195,176,215,106,192,160,221,182,217,57,211,88,198,233,113,171,204,138,193,209,225,59,176,184,134,223,151,193,200,217,100,225,79,180,142,190,123,222,80,232,216,133,216,148,211,110,198,96,187,224,95,208,112,178,227,94,171,96,181,209,170,225,196,206,94,216,87,217,171,191,82,218,127,227,176,219,207,230,79,214,203,105,213,143,174,188,125,193,220,60,215,172,214,101,211,110,161,117,187,180,125,218,220,62,208,203,217,87,198,156,216,226,161,161,223,224,72,178,198,213,195,219,208,140,175,217,74,201,201,66,186,154,229,89,226,169,204,87,184,85,161,133,201,80,176,188,114,224,77,207,126,202,83,219,200,125,172,169,190,216,80,88,221,68,218,133,216,117,217,157,217,170,190,124,214,210,156,231,84,207,204,113,200,70,222,162,208,227,92,223,136,167,195,221,221,77,173,213,109,214,117,211,217,89,217,91,210,152,194,206,202,110,216,177,190,207,227,185,172,230,172,207,171,199,234,207,149,194,192,179,212,209,210,101,198,225,85,164,211,110,194,182,211,224,65,228,218,79,224,81,122,208,154,129,206,92,193,171,148,188,221,80,220,161,165,166,161,214,99,210,64,174,224,221,105,200,122,230,216,94,223,128,225,161,219,126,187,137,191,222,214,148,151,198,218,210,110,208,228,184,211,35,202,218,195,216,115,212,95,177,199,101,184,208,202,212,134,193,129,192,81,182,223,70,226,230,134,167,183,198,222,227,227,226,63,213,109,187,177,219,223,203,144,179,209,103,177,181,158,221,90,222,166,207,175,230,207,99,205,234,210,210,168,223,143,210,187,209,204,150,209,213,208,193,221,214,77,215,199,81,197,82,177,190,210,231,79,179,221,64,182,199,82,204,204,95,172,187,178,209,86,222,220,118,192,223,88,220,77,174,104,224,137,182,186,96,207,198,74,152,196,217,206,79,214,208,204,180,94,215,81,177,160,201,164,173,205,76,199,220,228,91,215,155,226,79,133,181,136,182,226,96,221,109,209,223,71,202,95,217,87,202,204,183,210,187,212,81,226,184,224,88,170,214,198,226,142,212,81,209]
+[178,205,218,148,184,163,221,185,200,228,172,155,210,222,88,206,226,67,132,212,91,206,104,212,174,205,132,159,230,175,216,198,227,190,212,198,122,213,169,204,92,197,118,191,191,224,69,219,197,72,218,77,175,111,155,217,220,170,231,91,221,217,95,146,177,123,195,205,151,209,207,36,202,200,226,176,232,53,167,199,89,184,213,104,154,153,216,214,215,174,205,72,211,78,221,212,232,223,73,158,220,158,202,222,189,165,205,175,222,132,126,179,219,110,209,158,208,98,176,192,226,34,158,205,126,178,224,182,227,100,152,191,169,195,163,172,208,117,199,217,167,217,157,163,194,217,200,217,23,221,209,146,150,204,200,125,215,232,68,147,212,41,223,178,152,173,210,139,198,182,196,207,95,176,205,223,83,216,85,207,210,52,177,178,230,197,119,226,99,182,210,212,77,138,199,123,179,111,219,69,223,65,204,215,83,197,87,211,132,216,135,178,157,166,216,85,170,195,208,190,175,134,220,200,67,221,211,66,227,222,226,190,209,205,67,207,139,208,127,186,205,168,221,179,223,117,148,221,216,80,189,125,199,202,64,218,77,195,190,221,181,98,143,214,220,97,187,127,219,122,216,87,138,212,194,112,219,227,101,220,100,164,234,109,221,102,223,89,184,205,219,77,188,223,172,171,175,152,175,137,213,197,114,205,221,138,181,174,227,73,147,144,178,147,215,152,182,204,80,210,123,211,121,209,224,224,219,211,163,133,187,148,151,163,221,94,133,213,72,187,224,216,162,154,224,184,118,204,220,154,117,220,162,202,223,195,110,197,151,224,88,182,217,221,214,118,218,118,164,205,97,221,183,154,206,197,74,170,219,103,230,215,192,224,78,184,72,201,227,221,191,181,104,190,224,221,99,123,206,102,228,202,74,195,96,225,176,232,231,96,225,206,61,173,101,190,211,90,213,199,203,184,209,173,160,207,203,99,201,195,132,195,214,148,211,227,192,215,212,127,162,213,114,178,111,207,71,129,182,212,75,176,209,137,213,224,220,70,232,116,228,179,200,184,222,93,202,71,219,123,213,119,204,173,135,118,207,96,216,107,210,202,189,172,211,76,146,195,169,83,227,62,219,223,125,158,202,226,98,214,225,76,156,211,204,110,190,224,200,64,228,223,72,213,102,214,178,219,128,211,127,187,213,160,216,134,202,150,186,201,195,230,199,206,193,205,219,216,84,213,223,192,222,109,187,209,79,153,204,224,72,167,190,183,183,228,133,227,103,197,225,203,205,201,207,213,210,78,138,195,133,195,180,190,185,207,189,100,191,229,198,129,226,135,205,164,226,190,73,200,231,209,216,147,221,228,165,213,209,181,157,222,117,216,58,148,219,92,222,219,63,191,218,186,197,158,172,204,109,204,102,223,229,91,192,217,186,204,144,173,209,95,186,222,225,178,183,222,71,193,182,232,63,227,228,82,209,162,219,183,222,122,212,195,60,219,130,223,69,216,111,198,224,80,223,74,164,148,227,193,216,139,207,198,148,227,199,169,230,225,86,138,165,206,91,213,164,215,229,173,216,210,92,166,205,102,196,78,202,74,187,202,229,191,222,57,206,115,183,94,225,173,198,109,212,224,126,215,44,221,80,196,95,177,218,217,198,109,218,191,221,214,227,71,207,123,206,90,219,218,108,181,214,123,150,204,214,208,225,199,191,192,213,93,217,149,178,199,106,204,80,208,60,131,132,232,218,213,115,202,94,175,227,217,114,200,222,178,180,185,192,193,205,173,226,217,222,197,184,205,211,136,214,221,62,213,89,169,184,135,198,98,216,225,71,230,133,213,154,216,189,122,224,108,204,88,207,227,93,191,204,214,66,153,219,209,203,218,237,229,222,148,174,218,107,214,169,223,216,173,220,148,199,188,175,203,142,188,210,98,167,174,224,103,156,97,220,111,216,70,130,221,216,115,166,164,162,210,216,69,173,203,77,216,175,172,210,98,199,213,88,221,82,184,131,201,220,133,211,116,156,219,213,227,128,221,237,172,217,64,180,221,70,169,220,216,134,215,177,190,197,229,221,226,74,189,190,219,86,222,74,213,106,184,182,205,221,229,143,201,198,162,209,128,221,209,60,216,146,228,106,215,154,214,202,82,194,141,168,217,213,80,140,114,220,149,229,215,150,211,170,199,109,218,108,206,125,204,164,183,148,193,153,188,143,160,199,210,124,225,105,178,75,222,112,205,75,199,83,222,215,158,104,205,70,215,223,158,220,99,124,194,91,219,65,209,149,234,188,222,218,200,66,208,101,208,188,201,207,198,160,205,219,118,170,195,141,128,182,189,211,51,179,216,112,152,196,224,80,223,177,218,127,186,113,151,117,159,184,200,231,212,134,222,68,216,204,90,151,202,158,173,146,144,209,222,89,171,188,206,215,117,191,223,80,215,68,151,159,210,74,208,105,204,105,174,228,67,147,202,214,41,222,163,183,142,227,188,200,208,132,157,191,195,94,166,201,155,214,132,185,195,165,214,219,156,217,100,147,220,105,220,223,225,117,176,166,207,134,218,202,172,183,217,76,229,139,217,87,208,181,214,169,154,199,231,208,200,157,213,222,49,207,191,139,219,218,82,122,154,137,190,176,192,180,231,80,208,136,185,112,189,187,192,208,219,110,208,102,186,225,220,218,235,214,96,133,198,199,95,152,83,219,190,124,148,222,213,207,145,217,168,200,32,208,198,171,205,218,104,218,136,230,77,139,178,212,214,129,211,233,93,221,95,155,153,221,151,201,108,193,216,172,208,72,177,217,208,225,60,81,172,234,180,146,184,180,199,222,224,182,106,183,105,221,208,81,221,137,211,44,198,224,205,82,187,164,147,213,163,229,192,219,218,190,228,221,218,116,231,97,182,126,213,193,173,207,103,207,122,161,156,193,153,192,92,198,155,223,224,106,223,170,177,225,40,206,203,221,152,208,116,203,105,197,221,65,226,111,224,215,215,142,154,211,84,184,135,169,212,213,76,229,225,130,207,85,209,177,116,140,150,229,209,105,164,186,214,219,62,148,203,230,118,207,216,217,211,114,201,134,217,87,197,124,209,186,184,164,189,163,212,85,219,106,209,220,168,221,212,112,189,183,229,200,192,106,224,114,220,95,141,223,222,219,226,79,187,202,212,222,193,128,223,209,160,178,197,222,111,152,114,164,206,179,189,183,122,212,149,150,130,214,115,195,118,204,208,120,195,163,159,220,125,222,174,211,144,140,220,80,173,204,87,215,149,227,97,229,221,119,213,139,212,82,215,68,213,136,186,168,202,85,213,93,168,218,63,222,121,213,90,218,94,197,221,182,220,112,224,221,83,215,194,138,181,197,221,91,186,194,99,179,213,210,76,188,214,216,79,213,92,220,94,201,108,213,131,221,153,219,74,159,172,218,217,219,108,209,35,217,112,157,206,148,211,126,214,107,191,105,164,224,220,126,204,201,211,119,203,129,165,205,130,218,137,211,178,204,215,206,119,209,75,193,182,115,178,207,217,220,184,199,167,189,230,140,208,167,173,195,136,220,205,64,186,159,194,75,213,189,110,189,208,120,176,182,162,196,183,137,134,198,172,193,184,172,212,189,214,200,95,167,199,110,195,116,180,204,66,217,126,222,226,193,166,143,121,218,224,196,140,226,142,206,66,214,140,218,144,201,221,152,204,154,221,101,225,69,216,212,128,181,200,180,206,136,193,226,226,219,141,168,202,209,105,178,128,208,197,145,163,218,124,220,147,220,156,211,69,212,226,132,223,165,211,110,229,194,193,204,208,90,218,106,152,215,134,180,182,204,211,80,211,108,204,102,208,102,192,94,191,206,203,140,165,213,197,105,186,186,210,62,208,165,213,174,186,115,211,164,193,128,201,169,217,167,217,174,222,167,129,170,217,218,182,159,113,208,85,219,126,219,89,172,213,212,69,205,135,124,205,67,208,52,186,160,225,125,190,65,227,168,180,166,151,194,205,88,138,204,110,124,223,80,193,106,179,183,163,220,148,205,218,200,221,198,202,190,213,87,227,239,81,213,115,211,219,201,204,105,216,83,210,95,170,219,197,86,228,86,209,125,231,83,198,194,171,144,183,222,216,216,226,204,109,216,115,180,227,116,207,223,84,190,215,117,212,79,214,230,218,134,208,92,209,99,181,220,61,212,85,199,185,174,171,220,221,57,146,224,214,81,183,218,152,163,193,72,203,111,130,222,204,217,229,209,201,189,220,213,72,192,214,221,157,197,124,209,100,223,162,223,73,85,221,68,210,46,205,76,118,160,180,200,85,218,34,196,170,204,88,168,224,71,188,207,166,181,210,101,148,215,76,219,217,104,199,180,210,208,69,201,202,216,224,89,206,88,128,125,196,203,106,183,182,84,228,200,120,171,166,153,215,210,82,210,65,163,131,213,83,180,205,92,152,228,99,212,212,209,94,215,218,95,206,179,236,219,204,102,196,110,160,218,140,228,102,218,177,172,193,95,209,215,203,220,78,213,191,189,168,204,218,192,89,213,223,208,230,168,202,169,222,56,155,203,73,197,156,208,222,127,212,223,160,217,201,223,205,217,100,221,208,143,218,22,208,224,86,209,90,220,144,182,139,219,79,194,183,225,216,209,190,102,221,220,203,152,215,113,230,106,212,221,106,197,111,185,211,83,162,200,208,219,124,210,92,161,216,221,79,208,166,211,142,179,222,197,79,185,226,182,193,195,170,202,113,192,231,228,197,180,152,223,155,215,97,213,85,205,145,200,89,209,201,139,217,163,133,206,174,112,205,213,217,123,201,136,132,221,206,105,200,139,183,160,215,134,221,208,181,208,101,210,225,116,183,217,80,186,120,162,194,201,97,188,209,174,122,229,214,205,120,218,228,228,83,134,201,119,171,210,55,227,107,225,64,174,203,130,213,226,95,206,157,139,154,171,139,224,219,154,190,206,218,217,221,86,135,217,213,183,118,187,186,128,212,79,184,186,156,188,134,216,172,155,195,210,176,223,125,128,217,213,214,217,94,189,192,133,167,85,187,217,60,158,221,122,217,117,224,95,206,89,199,74,205,210,70,209,83,196,77,169,188,228,114,188,105,229,182,174,223,59,181,205,127,225,222,128,190,219,112,197,115,222,205,115,184,214,210,238,196,212,85,207,86,183,215,178,187,112,194,176,212,90,217,80,164,179,187,138,221,214,203,96,227,198,82,209,207,96,208,76,228,224,205,207,186,178,221,224,189,143,229,182,229,132,213,201,186,218,67,222,221,83,216,44,220,68,214,103,137,210,48,210,211,88,222,169,225,155,181,200,207,62,219,152,179,130,215,86,183,90,202,164,179,205,128,219,224,148,190,117,213,221,184,175,146,223,216,82,230,92,201,231,218,172,207,209,189,150,216,142,226,46,197,219,92,228,168,216,111,221,204,216,190,180,189,206,108,204,101,224,74,223,65,181,142,211,203,122,220,214,66,179,185,222,116,120,209,222,101,172,209,232,56,223,177,212,107,191,168,193,193,105,212,105,163,209,191,117,194,170,211,190,121,221,90,161,207,67,226,90,216,60,171,219,89,220,184,214,194,144,182,214,208,196,133,226,214,77,190,215,63,115,208,92,227,204,68,197,77,178,188,204,216,202,194,152,175,181,208,131,196,223,128,177,116,207,102,220,121,216,199,202,207,74,211,196,217,177,214,218,221,93,219,192,215,143,139,183,226,83,102,219,211,100,194,47,119,204,160,143,180,149,183,226,96,172,212,186,198,207,182,207,221,164,146,224,61,200,140,147,196,214,60,226,215,111,216,186,180,161,234,196,124,208,214,221,189,178,112,168,208,136,169,181,202,124,201,74,211,143,192,175,119,163,202,67,191,172,148,179,223,122,217,117,180,190,186,209,72,193,182,218,80,156,138,201,202,218,208,149,190,109,224,64,192,114,219,199,210,208,97,232,205,197,188,189,87,213,152,196,164,131,225,94,219,205,163,191,172,196,189,206,110,201,73,191,122,210,173,208,73,221,136,222,185,224,74,213,86,185,221,170,210,69,165,156,210,102,211,210,221,197,209,154,127,212,180,208,139,231,207,50,166,96,150,158,206,228,214,202,95,162,220,191,136,217,54,155,201,140,179,191,91,213,220,74,145,216,232,45,208,217,209,182,160,182,100,221,155,219,227,160,180,209,147,174,212,67,209,82,213,148,208,51,176,88,210,71,117,174,218,82,211,188,170,210,186,136,176,220,157,189,167,190,212,160,212,135,201,219,49,162,165,209,117,175,213,152,176,220,221,124,150,204,220,73,228,194,218,57,195,173,159,173,175,206,176,229,79,164,201,203,152,214,116,137,219,66,222,214,80,175,202,90,225,113,219,206,78,190,89,214,50,209,154,188,227,194,157,195,74,186,206,130,198,73,212,60,204,122,222,99,205,196,229,213,83,230,108,171,126,192,104,216,207,117,217,197,214,79,207,110,221,79,217,144,206,160,206,172,197,183,207,217,207,113,210,221,71,161,221,164,227,214,142,177,185,180,103,130,198,123,205,74,216,102,219,160,217,75,204,114,192,213,166,188,118,222,227,92,195,219,161,200,221,69,203,143,198,198,217,198,66,212,50,208,116,199,125,210,207,167,225,116,207,97,184,99,220,184,203,184,219,177,167,202,214,55,207,161,197,122,212,226,187,96,216,201,188,135,224,207,139,225,230,220,121,221,107,212,66,170,169,210,199,102,220,94,159,184,207,92,207,231,214,125,227,220,205,58,193,203,215,223,229,78,196,170,185,196,162,234,56,201,123,171,231,196,86,162,199,213,220,68,200,68,205,88,225,135,220,82,182,215,222,79,152,230,62,162,218,184,224,67,206,99,189,124,214,197,73,204,105,221,179,102,218,232,80,214,181,170,204,165,216,207,217,212,195,176,215,106,192,160,221,182,217,57,211,88,198,233,113,171,204,138,193,209,225,59,176,184,134,223,151,193,200,217,100,225,79,180,142,190,123,222,80,232,216,133,216,148,211,110,198,96,187,224,95,208,112,178,227,94,171,96,181,209,170,225,196,206,94,216,87,217,171,191,82,218,127,227,176,219,207,230,79,214,203,105,213,143,174,188,125,193,220,60,215,172,214,101,211,110,161,117,187,180,125,218,220,62,208,203,217,87,198,156,216,226,161,161,223,224,72,178,198,213,195,219,208,140,175,217,74,201,201,66,186,154,229,89,226,169,204,87,184,85,161,133,201,80,176,188,114,224,77,207,126,202,83,219,200,125,172,169,190,216,80,88,221,68,218,133,216,117,217,157,217,170,190,124,214,210,156,231,84,207,204,113,200,70,222,162,208,227,92,223,136,167,195,221,221,77,173,213,109,214,117,211,217,89,217,91,210,152,194,206,202,110,216,177,190,207,227,185,172,230,172,207,171,199,234,207,149,194,192,179,212,209,210,101,198,225,85,164,211,110,194,182,211,224,65,228,218,79,224,81,122,208,154,129,206,92,193,171,148,188,221,80,220,161,165,166,161,214,99,210,64,174,224,221,105,200,122,230,216,94,223,128,225,161,219,126,187,137,191,222,214,148,151,198,218,210,110,208,228,184,211,35,202,218,195,216,115,212,95,177,199,101,184,208,202,212,134,193,129,192,81,182,223,70,226,230,134,167,183,198,222,227,227,226,63,213,109,187,177,219,223,203,144,179,209,103,177,181,158,221,90,222,166,207,175,230,207,99,205,234,210,210,168,223,143,210,187,209,204,150,209,213,208,193,221,214,77,215,199,81,197,82,177,190,210,231,79,179,221,64,182,199,82,204,204,95,172,187,178,209,86,222,220,118,192,223,88,220,77,174,104,224,137,182,186,96,207,198,74,152,196,217,206,79,214,208,204,180,94,215,81,177,160,201,164,173,205,76,199,220,228,91,215,155,226,79,133,181,136,182,226,96,221,109,209,223,71,202,95,217,87,202,204,183,210,187,212,81,226,184,224,88,170,214,198,226,142,212,81,209,189,172,192,221,216,123,221,126,204,218,222,76,205,73,225,221,73,204,108]

ivf.pid.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:677fd54bf0a60aa840fe1e311c19f23d63a224ee09b61ea37336c735349aee53
-size 1597784
+oid sha256:b051bdc06d137304fa4a7e4c91ac0c006e50a06736f96c106a939af8bb0f5581
+size 1610584

metadata.json

@@ -37,7 +37,7 @@
     "checkpoint":"colbert-ir/colbertv2.0",
     "triples":"/future/u/okhattab/root/unit/experiments/2021.10/downstream.distillation.round2.2_score/round2.nway6.cosine.ib/examples.64.json",
     "collection":[
-      "list with 3543 elements starting with...",
+      "list with 3562 elements starting with...",
       [
         "Deep neural networks have demonstrated remarkable performance in supervised learning tasks but require large amounts of labeled data. Self-supervised learning offers an alternative paradigm, enabling the model to learn from data without explicit labels. Information theory has been instrumental in understanding and optimizing deep neural networks. Specifically, the information bottleneck principle has been applied to optimize the trade-off between compression and relevant information preservation in supervised settings. However, the optimal information objective in self-supervised learning remains unclear. In this paper, we review various approaches to self-supervised learning from an information-theoretic standpoint and present a unified framework that formalizes the self-supervised information-theoretic learning problem. We integrate existing research into a coherent framework, examine recent self-supervised methods, and identify research opportunities and challenges. Moreover, we discuss empirical measurement of information-theoretic quantities and their estimators. This paper offers a comprehensive review of the intersection between information theory, self-supervised learning, and deep neural networks.",
         "Pre-trained large language models (LLMs) capture procedural knowledge about the world. Recent work has leveraged LLM's ability to generate abstract plans to simplify challenging control tasks, either by action scoring, or action modeling (fine-tuning). However, the transformer architecture inherits several constraints that make it difficult for the LLM to directly serve as the agent: e.g. limited input lengths, fine-tuning inefficiency, bias from pre-training, and incompatibility with non-text environments. To maintain compatibility with a low-level trainable actor, we propose to instead use the knowledge in LLMs to simplify the control problem, rather than solving it. We propose the Plan, Eliminate, and Track (PET) framework. The Plan module translates a task description into a list of high-level sub-tasks. The Eliminate module masks out irrelevant objects and receptacles from the observation for the current sub-task. Finally, the Track module determines whether the agent has accomplished each sub-task. On the AlfWorld instruction following benchmark, the PET framework leads to a significant 15% improvement over SOTA for generalization to human goal specifications.",
@@ -50,7 +50,7 @@
     "root":".ragatouille/",
     "experiment":"colbert",
     "index_root":null,
-    "name":"2024-05/29/13.53.40",
+    "name":"2024-05/30/14.54.06",
     "rank":0,
     "nranks":1,
     "amp":true,
@@ -59,8 +59,8 @@
   },
   "num_chunks":1,
   "num_partitions":8192,
-  "num_embeddings":609827,
-  "avg_doclen":172.1216483206,
+  "num_embeddings":613116,
+  "avg_doclen":172.1268950028,
   "RAGatouille":{
     "index_config":{
       "index_type":"PLAID",

pid_docid_map.json

@@ -3541,5 +3541,24 @@
   "3539":"2405.18426",
   "3540":"2405.18386",
   "3541":"2405.18386",
-  "3542":"2405.18424"
+  "3542":"2405.18424",
+  "3543":"2405.19327",
+  "3544":"2405.19327",
+  "3545":"2405.18870",
+  "3546":"2405.18750",
+  "3547":"2405.19331",
+  "3548":"2405.19331",
+  "3549":"2405.18503",
+  "3550":"2405.18503",
+  "3551":"2405.18515",
+  "3552":"2405.18991",
+  "3553":"2405.19107",
+  "3554":"2405.19107",
+  "3555":"2405.18669",
+  "3556":"2405.18669",
+  "3557":"2405.19325",
+  "3558":"2405.19320",
+  "3559":"2405.19320",
+  "3560":"2405.19332",
+  "3561":"2405.19332"
 }

plan.json

@@ -37,7 +37,7 @@
         "checkpoint": "colbert-ir\/colbertv2.0",
         "triples": "\/future\/u\/okhattab\/root\/unit\/experiments\/2021.10\/downstream.distillation.round2.2_score\/round2.nway6.cosine.ib\/examples.64.json",
         "collection": [
-            "list with 3543 elements starting with...",
+            "list with 3562 elements starting with...",
             [
                 "Deep neural networks have demonstrated remarkable performance in supervised learning tasks but require large amounts of labeled data. Self-supervised learning offers an alternative paradigm, enabling the model to learn from data without explicit labels. Information theory has been instrumental in understanding and optimizing deep neural networks. Specifically, the information bottleneck principle has been applied to optimize the trade-off between compression and relevant information preservation in supervised settings. However, the optimal information objective in self-supervised learning remains unclear. In this paper, we review various approaches to self-supervised learning from an information-theoretic standpoint and present a unified framework that formalizes the self-supervised information-theoretic learning problem. We integrate existing research into a coherent framework, examine recent self-supervised methods, and identify research opportunities and challenges. Moreover, we discuss empirical measurement of information-theoretic quantities and their estimators. This paper offers a comprehensive review of the intersection between information theory, self-supervised learning, and deep neural networks.",
                 "Pre-trained large language models (LLMs) capture procedural knowledge about the world. Recent work has leveraged LLM's ability to generate abstract plans to simplify challenging control tasks, either by action scoring, or action modeling (fine-tuning). However, the transformer architecture inherits several constraints that make it difficult for the LLM to directly serve as the agent: e.g. limited input lengths, fine-tuning inefficiency, bias from pre-training, and incompatibility with non-text environments. To maintain compatibility with a low-level trainable actor, we propose to instead use the knowledge in LLMs to simplify the control problem, rather than solving it. We propose the Plan, Eliminate, and Track (PET) framework. The Plan module translates a task description into a list of high-level sub-tasks. The Eliminate module masks out irrelevant objects and receptacles from the observation for the current sub-task. Finally, the Track module determines whether the agent has accomplished each sub-task. On the AlfWorld instruction following benchmark, the PET framework leads to a significant 15% improvement over SOTA for generalization to human goal specifications.",
@@ -50,7 +50,7 @@
         "root": ".ragatouille\/",
         "experiment": "colbert",
         "index_root": null,
-        "name": "2024-05\/29\/13.53.40",
+        "name": "2024-05\/30\/14.54.06",
         "rank": 0,
         "nranks": 1,
         "amp": true,
@@ -59,6 +59,6 @@
     },
     "num_chunks": 1,
     "num_partitions": 8192,
-    "num_embeddings_est": 609826.9813842773,
-    "avg_doclen_est": 172.12164306640625
+    "num_embeddings_est": 613115.9896240234,
+    "avg_doclen_est": 172.12689208984375
 }