{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2019:QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7","short_pith_number":"pith:QQZ22ZJ6","schema_version":"1.0","canonical_sha256":"8433ad653ef6712627a4913095a74657fb68e60ea4798de4147773dd1bd37fea","source":{"kind":"arxiv","id":"1908.08681","version":3},"attestation_state":"computed","paper":{"title":"Mish: A Self Regularized Non-Monotonic Activation Function","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.NE","stat.ML"],"primary_cat":"cs.LG","authors_text":"Diganta Misra","submitted_at":"2019-08-23T06:22:06Z","abstract_excerpt":"We propose $\\textit{Mish}$, a novel self-regularized non-monotonic activation function which can be mathematically defined as: $f(x)=x\\tanh(softplus(x))$. As activation functions play a crucial role in the performance and training dynamics in neural networks, we validated experimentally on several well-known benchmarks against the best combinations of architectures and activation functions. We also observe that data augmentation techniques have a favorable effect on benchmarks like ImageNet-1k and MS-COCO across multiple architectures. For example, Mish outperformed Leaky ReLU on YOLOv4 with a"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"1908.08681","kind":"arxiv","version":3},"metadata":{"license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","primary_cat":"cs.LG","submitted_at":"2019-08-23T06:22:06Z","cross_cats_sorted":["cs.CV","cs.NE","stat.ML"],"title_canon_sha256":"02b9d41646dfae615ec88218085fef84cbab71a172e7c6865de22283113bf012","abstract_canon_sha256":"ff0741f41de7dfe4b55eaaae8d3fd6d4d5b0d7275beaeb9b74a6bc412a440b81"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T01:26:48.857843Z","signature_b64":"9/s0kK4wZwRdeIw62NELOp/zHJa+nvBEdzfcr3/DDO/AfrC35nJJklslBVC5BSYCsmIj7fF01OFoJ6+CQVKJAA==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8433ad653ef6712627a4913095a74657fb68e60ea4798de4147773dd1bd37fea","last_reissued_at":"2026-07-05T01:26:48.857374Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T01:26:48.857374Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Mish: A Self Regularized Non-Monotonic Activation Function","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","headline":"","cross_cats":["cs.CV","cs.NE","stat.ML"],"primary_cat":"cs.LG","authors_text":"Diganta Misra","submitted_at":"2019-08-23T06:22:06Z","abstract_excerpt":"We propose $\\textit{Mish}$, a novel self-regularized non-monotonic activation function which can be mathematically defined as: $f(x)=x\\tanh(softplus(x))$. As activation functions play a crucial role in the performance and training dynamics in neural networks, we validated experimentally on several well-known benchmarks against the best combinations of architectures and activation functions. We also observe that data augmentation techniques have a favorable effect on benchmarks like ImageNet-1k and MS-COCO across multiple architectures. For example, Mish outperformed Leaky ReLU on YOLOv4 with a"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"1908.08681","kind":"arxiv","version":3},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/1908.08681/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"1908.08681","created_at":"2026-07-05T01:26:48.857445+00:00"},{"alias_kind":"arxiv_version","alias_value":"1908.08681v3","created_at":"2026-07-05T01:26:48.857445+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.1908.08681","created_at":"2026-07-05T01:26:48.857445+00:00"},{"alias_kind":"pith_short_12","alias_value":"QQZ22ZJ66ZYS","created_at":"2026-07-05T01:26:48.857445+00:00"},{"alias_kind":"pith_short_16","alias_value":"QQZ22ZJ66ZYSMJ5E","created_at":"2026-07-05T01:26:48.857445+00:00"},{"alias_kind":"pith_short_8","alias_value":"QQZ22ZJ6","created_at":"2026-07-05T01:26:48.857445+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":22,"internal_anchor_count":1,"sample":[{"citing_arxiv_id":"2607.06274","citing_title":"Learning-based Physics-Constrained Neural Kernel for Sound Field Estimation With Source-Position-Dependent Directional Weighting","ref_index":35,"is_internal_anchor":true},{"citing_arxiv_id":"2606.21789","citing_title":"Bayesian three-dimensional seismic travel-time tomography for active- and passive-source seismic data using physics-informed neural network","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2606.19397","citing_title":"DiffusionVS: A Generative Framework for Robust Visual Servoing Based on Diffusion Policy","ref_index":29,"is_internal_anchor":false},{"citing_arxiv_id":"2606.10234","citing_title":"Performance of the Eos detector with water","ref_index":47,"is_internal_anchor":false},{"citing_arxiv_id":"2606.08438","citing_title":"Improving Bayesian Optimization via Training-Aware Conditional Diffusion Models","ref_index":27,"is_internal_anchor":false},{"citing_arxiv_id":"2605.26647","citing_title":"More Expressive Feedforward Layers: Part I. Token-Adaptive Mixing of Activations","ref_index":30,"is_internal_anchor":false},{"citing_arxiv_id":"2605.28704","citing_title":"Expressive Power of Floating-Point Neural Networks with Arbitrary Reduction Orders and Inexact Activation Implementations","ref_index":19,"is_internal_anchor":false},{"citing_arxiv_id":"2501.05465","citing_title":"Small Language Models (SLMs) Can Still Pack a Punch: A survey (updated 2026)","ref_index":89,"is_internal_anchor":false},{"citing_arxiv_id":"2508.20614","citing_title":"Improving the Accuracy of Amortized Model Comparison with Self-Consistency","ref_index":33,"is_internal_anchor":false},{"citing_arxiv_id":"2509.13095","citing_title":"Empowering Multi-Robot Cooperation via Sequential World Models","ref_index":24,"is_internal_anchor":false},{"citing_arxiv_id":"2603.14267","citing_title":"DiFlowDubber: Discrete Flow Matching for Automated Video Dubbing via Cross-Modal Alignment and Synchronization","ref_index":38,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07060","citing_title":"Functional-prior-based approaches to Bayesian PDE-constrained inversion using physics-informed neural networks","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14200","citing_title":"How to Scale Mixture-of-Experts: From muP to the Maximally Scale-Stable Parameterization","ref_index":267,"is_internal_anchor":false},{"citing_arxiv_id":"2605.14493","citing_title":"Deep Learning for Solving and Estimating Dynamic Models in Economics and Finance","ref_index":9,"is_internal_anchor":false},{"citing_arxiv_id":"2310.16828","citing_title":"TD-MPC2: Scalable, Robust World Models for Continuous Control","ref_index":176,"is_internal_anchor":false},{"citing_arxiv_id":"2004.10934","citing_title":"YOLOv4: Optimal Speed and Accuracy of Object Detection","ref_index":55,"is_internal_anchor":false},{"citing_arxiv_id":"2605.09908","citing_title":"Voice Biomarkers for Depression and Anxiety","ref_index":16,"is_internal_anchor":false},{"citing_arxiv_id":"2604.12159","citing_title":"VidTAG: Temporally Aligned Video to GPS Geolocalization with Denoising Sequence Prediction at a Global Scale","ref_index":31,"is_internal_anchor":false},{"citing_arxiv_id":"2605.07060","citing_title":"Functional-prior-based approaches to Bayesian PDE-constrained inversion using physics-informed neural networks","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2604.21677","citing_title":"Geometric Monomial (GEM): a family of rational 2N-differentiable activation functions","ref_index":6,"is_internal_anchor":false},{"citing_arxiv_id":"2605.02591","citing_title":"Universal Smoothness via Bernstein Polynomials: A Constructive Approximation Approach for Activation Functions","ref_index":14,"is_internal_anchor":false},{"citing_arxiv_id":"2605.01702","citing_title":"Floating-Point Networks with Automatic Differentiation Can Represent Almost All Floating-Point Functions and Their Gradients","ref_index":34,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7","json":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7.json","graph_json":"https://pith.science/api/pith-number/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/graph.json","events_json":"https://pith.science/api/pith-number/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/events.json","paper":"https://pith.science/paper/QQZ22ZJ6"},"agent_actions":{"view_html":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7","download_json":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7.json","view_paper":"https://pith.science/paper/QQZ22ZJ6","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=1908.08681&json=true","fetch_graph":"https://pith.science/api/pith-number/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/graph.json","fetch_events":"https://pith.science/api/pith-number/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/action/storage_attestation","attest_author":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/action/author_attestation","sign_citation":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/action/citation_signature","submit_replication":"https://pith.science/pith/QQZ22ZJ66ZYSMJ5ESEYJLJ2GK7/action/replication_record"}},"created_at":"2026-07-05T01:26:48.857445+00:00","updated_at":"2026-07-05T01:26:48.857445+00:00"}