Publish verified raw NVIDIA PersonaPlex v1 mirror with provenance

.gitattributes

@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+figures/results_conversation_dynamics.png filter=lfs diff=lfs merge=lfs -text
+figures/results_latency.png filter=lfs diff=lfs merge=lfs -text
+figures/results_task_adherence.png filter=lfs diff=lfs merge=lfs -text

LICENSES/CC-BY-4.0.txt

@@ -0,0 +1,396 @@
+Attribution 4.0 International
+
+=======================================================================
+
+Creative Commons Corporation ("Creative Commons") is not a law firm and
+does not provide legal services or legal advice. Distribution of
+Creative Commons public licenses does not create a lawyer-client or
+other relationship. Creative Commons makes its licenses and related
+information available on an "as-is" basis. Creative Commons gives no
+warranties regarding its licenses, any material licensed under their
+terms and conditions, or any related information. Creative Commons
+disclaims all liability for damages resulting from their use to the
+fullest extent possible.
+
+Using Creative Commons Public Licenses
+
+Creative Commons public licenses provide a standard set of terms and
+conditions that creators and other rights holders may use to share
+original works of authorship and other material subject to copyright
+and certain other rights specified in the public license below. The
+following considerations are for informational purposes only, are not
+exhaustive, and do not form part of our licenses.
+
+     Considerations for licensors: Our public licenses are
+     intended for use by those authorized to give the public
+     permission to use material in ways otherwise restricted by
+     copyright and certain other rights. Our licenses are
+     irrevocable. Licensors should read and understand the terms
+     and conditions of the license they choose before applying it.
+     Licensors should also secure all rights necessary before
+     applying our licenses so that the public can reuse the
+     material as expected. Licensors should clearly mark any
+     material not subject to the license. This includes other CC-
+     licensed material, or material used under an exception or
+     limitation to copyright. More considerations for licensors:
+    wiki.creativecommons.org/Considerations_for_licensors
+
+     Considerations for the public: By using one of our public
+     licenses, a licensor grants the public permission to use the
+     licensed material under specified terms and conditions. If
+     the licensor's permission is not necessary for any reason--for
+     example, because of any applicable exception or limitation to
+     copyright--then that use is not regulated by the license. Our
+     licenses grant only permissions under copyright and certain
+     other rights that a licensor has authority to grant. Use of
+     the licensed material may still be restricted for other
+     reasons, including because others have copyright or other
+     rights in the material. A licensor may make special requests,
+     such as asking that all changes be marked or described.
+     Although not required by our licenses, you are encouraged to
+     respect those requests where reasonable. More considerations
+     for the public:
+    wiki.creativecommons.org/Considerations_for_licensees
+
+=======================================================================
+
+Creative Commons Attribution 4.0 International Public License
+
+By exercising the Licensed Rights (defined below), You accept and agree
+to be bound by the terms and conditions of this Creative Commons
+Attribution 4.0 International Public License ("Public License"). To the
+extent this Public License may be interpreted as a contract, You are
+granted the Licensed Rights in consideration of Your acceptance of
+these terms and conditions, and the Licensor grants You such rights in
+consideration of benefits the Licensor receives from making the
+Licensed Material available under these terms and conditions.
+
+
+Section 1 -- Definitions.
+
+  a. Adapted Material means material subject to Copyright and Similar
+     Rights that is derived from or based upon the Licensed Material
+     and in which the Licensed Material is translated, altered,
+     arranged, transformed, or otherwise modified in a manner requiring
+     permission under the Copyright and Similar Rights held by the
+     Licensor. For purposes of this Public License, where the Licensed
+     Material is a musical work, performance, or sound recording,
+     Adapted Material is always produced where the Licensed Material is
+     synched in timed relation with a moving image.
+
+  b. Adapter's License means the license You apply to Your Copyright
+     and Similar Rights in Your contributions to Adapted Material in
+     accordance with the terms and conditions of this Public License.
+
+  c. Copyright and Similar Rights means copyright and/or similar rights
+     closely related to copyright including, without limitation,
+     performance, broadcast, sound recording, and Sui Generis Database
+     Rights, without regard to how the rights are labeled or
+     categorized. For purposes of this Public License, the rights
+     specified in Section 2(b)(1)-(2) are not Copyright and Similar
+     Rights.
+
+  d. Effective Technological Measures means those measures that, in the
+     absence of proper authority, may not be circumvented under laws
+     fulfilling obligations under Article 11 of the WIPO Copyright
+     Treaty adopted on December 20, 1996, and/or similar international
+     agreements.
+
+  e. Exceptions and Limitations means fair use, fair dealing, and/or
+     any other exception or limitation to Copyright and Similar Rights
+     that applies to Your use of the Licensed Material.
+
+  f. Licensed Material means the artistic or literary work, database,
+     or other material to which the Licensor applied this Public
+     License.
+
+  g. Licensed Rights means the rights granted to You subject to the
+     terms and conditions of this Public License, which are limited to
+     all Copyright and Similar Rights that apply to Your use of the
+     Licensed Material and that the Licensor has authority to license.
+
+  h. Licensor means the individual(s) or entity(ies) granting rights
+     under this Public License.
+
+  i. Share means to provide material to the public by any means or
+     process that requires permission under the Licensed Rights, such
+     as reproduction, public display, public performance, distribution,
+     dissemination, communication, or importation, and to make material
+     available to the public including in ways that members of the
+     public may access the material from a place and at a time
+     individually chosen by them.
+
+  j. Sui Generis Database Rights means rights other than copyright
+     resulting from Directive 96/9/EC of the European Parliament and of
+     the Council of 11 March 1996 on the legal protection of databases,
+     as amended and/or succeeded, as well as other essentially
+     equivalent rights anywhere in the world.
+
+  k. You means the individual or entity exercising the Licensed Rights
+     under this Public License. Your has a corresponding meaning.
+
+
+Section 2 -- Scope.
+
+  a. License grant.
+
+       1. Subject to the terms and conditions of this Public License,
+          the Licensor hereby grants You a worldwide, royalty-free,
+          non-sublicensable, non-exclusive, irrevocable license to
+          exercise the Licensed Rights in the Licensed Material to:
+
+            a. reproduce and Share the Licensed Material, in whole or
+               in part; and
+
+            b. produce, reproduce, and Share Adapted Material.
+
+       2. Exceptions and Limitations. For the avoidance of doubt, where
+          Exceptions and Limitations apply to Your use, this Public
+          License does not apply, and You do not need to comply with
+          its terms and conditions.
+
+       3. Term. The term of this Public License is specified in Section
+          6(a).
+
+       4. Media and formats; technical modifications allowed. The
+          Licensor authorizes You to exercise the Licensed Rights in
+          all media and formats whether now known or hereafter created,
+          and to make technical modifications necessary to do so. The
+          Licensor waives and/or agrees not to assert any right or
+          authority to forbid You from making technical modifications
+          necessary to exercise the Licensed Rights, including
+          technical modifications necessary to circumvent Effective
+          Technological Measures. For purposes of this Public License,
+          simply making modifications authorized by this Section 2(a)
+          (4) never produces Adapted Material.
+
+       5. Downstream recipients.
+
+            a. Offer from the Licensor -- Licensed Material. Every
+               recipient of the Licensed Material automatically
+               receives an offer from the Licensor to exercise the
+               Licensed Rights under the terms and conditions of this
+               Public License.
+
+            b. No downstream restrictions. You may not offer or impose
+               any additional or different terms or conditions on, or
+               apply any Effective Technological Measures to, the
+               Licensed Material if doing so restricts exercise of the
+               Licensed Rights by any recipient of the Licensed
+               Material.
+
+       6. No endorsement. Nothing in this Public License constitutes or
+          may be construed as permission to assert or imply that You
+          are, or that Your use of the Licensed Material is, connected
+          with, or sponsored, endorsed, or granted official status by,
+          the Licensor or others designated to receive attribution as
+          provided in Section 3(a)(1)(A)(i).
+
+  b. Other rights.
+
+       1. Moral rights, such as the right of integrity, are not
+          licensed under this Public License, nor are publicity,
+          privacy, and/or other similar personality rights; however, to
+          the extent possible, the Licensor waives and/or agrees not to
+          assert any such rights held by the Licensor to the limited
+          extent necessary to allow You to exercise the Licensed
+          Rights, but not otherwise.
+
+       2. Patent and trademark rights are not licensed under this
+          Public License.
+
+       3. To the extent possible, the Licensor waives any right to
+          collect royalties from You for the exercise of the Licensed
+          Rights, whether directly or through a collecting society
+          under any voluntary or waivable statutory or compulsory
+          licensing scheme. In all other cases the Licensor expressly
+          reserves any right to collect such royalties.
+
+
+Section 3 -- License Conditions.
+
+Your exercise of the Licensed Rights is expressly made subject to the
+following conditions.
+
+  a. Attribution.
+
+       1. If You Share the Licensed Material (including in modified
+          form), You must:
+
+            a. retain the following if it is supplied by the Licensor
+               with the Licensed Material:
+
+                 i. identification of the creator(s) of the Licensed
+                    Material and any others designated to receive
+                    attribution, in any reasonable manner requested by
+                    the Licensor (including by pseudonym if
+                    designated);
+
+                ii. a copyright notice;
+
+               iii. a notice that refers to this Public License;
+
+                iv. a notice that refers to the disclaimer of
+                    warranties;
+
+                 v. a URI or hyperlink to the Licensed Material to the
+                    extent reasonably practicable;
+
+            b. indicate if You modified the Licensed Material and
+               retain an indication of any previous modifications; and
+
+            c. indicate the Licensed Material is licensed under this
+               Public License, and include the text of, or the URI or
+               hyperlink to, this Public License.
+
+       2. You may satisfy the conditions in Section 3(a)(1) in any
+          reasonable manner based on the medium, means, and context in
+          which You Share the Licensed Material. For example, it may be
+          reasonable to satisfy the conditions by providing a URI or
+          hyperlink to a resource that includes the required
+          information.
+
+       3. If requested by the Licensor, You must remove any of the
+          information required by Section 3(a)(1)(A) to the extent
+          reasonably practicable.
+
+       4. If You Share Adapted Material You produce, the Adapter's
+          License You apply must not prevent recipients of the Adapted
+          Material from complying with this Public License.
+
+
+Section 4 -- Sui Generis Database Rights.
+
+Where the Licensed Rights include Sui Generis Database Rights that
+apply to Your use of the Licensed Material:
+
+  a. for the avoidance of doubt, Section 2(a)(1) grants You the right
+     to extract, reuse, reproduce, and Share all or a substantial
+     portion of the contents of the database;
+
+  b. if You include all or a substantial portion of the database
+     contents in a database in which You have Sui Generis Database
+     Rights, then the database in which You have Sui Generis Database
+     Rights (but not its individual contents) is Adapted Material; and
+
+  c. You must comply with the conditions in Section 3(a) if You Share
+     all or a substantial portion of the contents of the database.
+
+For the avoidance of doubt, this Section 4 supplements and does not
+replace Your obligations under this Public License where the Licensed
+Rights include other Copyright and Similar Rights.
+
+
+Section 5 -- Disclaimer of Warranties and Limitation of Liability.
+
+  a. UNLESS OTHERWISE SEPARATELY UNDERTAKEN BY THE LICENSOR, TO THE
+     EXTENT POSSIBLE, THE LICENSOR OFFERS THE LICENSED MATERIAL AS-IS
+     AND AS-AVAILABLE, AND MAKES NO REPRESENTATIONS OR WARRANTIES OF
+     ANY KIND CONCERNING THE LICENSED MATERIAL, WHETHER EXPRESS,
+     IMPLIED, STATUTORY, OR OTHER. THIS INCLUDES, WITHOUT LIMITATION,
+     WARRANTIES OF TITLE, MERCHANTABILITY, FITNESS FOR A PARTICULAR
+     PURPOSE, NON-INFRINGEMENT, ABSENCE OF LATENT OR OTHER DEFECTS,
+     ACCURACY, OR THE PRESENCE OR ABSENCE OF ERRORS, WHETHER OR NOT
+     KNOWN OR DISCOVERABLE. WHERE DISCLAIMERS OF WARRANTIES ARE NOT
+     ALLOWED IN FULL OR IN PART, THIS DISCLAIMER MAY NOT APPLY TO YOU.
+
+  b. TO THE EXTENT POSSIBLE, IN NO EVENT WILL THE LICENSOR BE LIABLE
+     TO YOU ON ANY LEGAL THEORY (INCLUDING, WITHOUT LIMITATION,
+     NEGLIGENCE) OR OTHERWISE FOR ANY DIRECT, SPECIAL, INDIRECT,
+     INCIDENTAL, CONSEQUENTIAL, PUNITIVE, EXEMPLARY, OR OTHER LOSSES,
+     COSTS, EXPENSES, OR DAMAGES ARISING OUT OF THIS PUBLIC LICENSE OR
+     USE OF THE LICENSED MATERIAL, EVEN IF THE LICENSOR HAS BEEN
+     ADVISED OF THE POSSIBILITY OF SUCH LOSSES, COSTS, EXPENSES, OR
+     DAMAGES. WHERE A LIMITATION OF LIABILITY IS NOT ALLOWED IN FULL OR
+     IN PART, THIS LIMITATION MAY NOT APPLY TO YOU.
+
+  c. The disclaimer of warranties and limitation of liability provided
+     above shall be interpreted in a manner that, to the extent
+     possible, most closely approximates an absolute disclaimer and
+     waiver of all liability.
+
+
+Section 6 -- Term and Termination.
+
+  a. This Public License applies for the term of the Copyright and
+     Similar Rights licensed here. However, if You fail to comply with
+     this Public License, then Your rights under this Public License
+     terminate automatically.
+
+  b. Where Your right to use the Licensed Material has terminated under
+     Section 6(a), it reinstates:
+
+       1. automatically as of the date the violation is cured, provided
+          it is cured within 30 days of Your discovery of the
+          violation; or
+
+       2. upon express reinstatement by the Licensor.
+
+     For the avoidance of doubt, this Section 6(b) does not affect any
+     right the Licensor may have to seek remedies for Your violations
+     of this Public License.
+
+  c. For the avoidance of doubt, the Licensor may also offer the
+     Licensed Material under separate terms or conditions or stop
+     distributing the Licensed Material at any time; however, doing so
+     will not terminate this Public License.
+
+  d. Sections 1, 5, 6, 7, and 8 survive termination of this Public
+     License.
+
+
+Section 7 -- Other Terms and Conditions.
+
+  a. The Licensor shall not be bound by any additional or different
+     terms or conditions communicated by You unless expressly agreed.
+
+  b. Any arrangements, understandings, or agreements regarding the
+     Licensed Material not stated herein are separate from and
+     independent of the terms and conditions of this Public License.
+
+
+Section 8 -- Interpretation.
+
+  a. For the avoidance of doubt, this Public License does not, and
+     shall not be interpreted to, reduce, limit, restrict, or impose
+     conditions on any use of the Licensed Material that could lawfully
+     be made without permission under this Public License.
+
+  b. To the extent possible, if any provision of this Public License is
+     deemed unenforceable, it shall be automatically reformed to the
+     minimum extent necessary to make it enforceable. If the provision
+     cannot be reformed, it shall be severed from this Public License
+     without affecting the enforceability of the remaining terms and
+     conditions.
+
+  c. No term or condition of this Public License will be waived and no
+     failure to comply consented to unless expressly agreed to by the
+     Licensor.
+
+  d. Nothing in this Public License constitutes or may be interpreted
+     as a limitation upon, or waiver of, any privileges and immunities
+     that apply to the Licensor or You, including from the legal
+     processes of any jurisdiction or authority.
+
+
+=======================================================================
+
+Creative Commons is not a party to its public
+licenses. Notwithstanding, Creative Commons may elect to apply one of
+its public licenses to material it publishes and in those instances
+will be considered the “Licensor.” The text of the Creative Commons
+public licenses is dedicated to the public domain under the CC0 Public
+Domain Dedication. Except for the limited purpose of indicating that
+material is shared under a Creative Commons public license or as
+otherwise permitted by the Creative Commons policies published at
+creativecommons.org/policies, Creative Commons does not authorize the
+use of the trademark "Creative Commons" or any other trademark or logo
+of Creative Commons without its prior written consent including,
+without limitation, in connection with any unauthorized modifications
+to any of its public licenses or any other arrangements,
+understandings, or agreements concerning use of licensed material. For
+the avoidance of doubt, this paragraph does not form part of the
+public licenses.
+
+Creative Commons may be contacted at creativecommons.org.
+

NOTICE

@@ -0,0 +1,10 @@
+Licensed by NVIDIA Corporation under the NVIDIA Open Model License.
+
+Upstream work: PersonaPlex 7B v1 by NVIDIA Corporation.
+Upstream repository: https://huggingface.co/nvidia/personaplex-7b-v1
+Upstream source code: https://github.com/NVIDIA/personaplex
+Upstream base model: https://huggingface.co/kyutai/moshiko-pytorch-bf16
+
+This repository is an independently maintained, byte-identical mirror of the
+listed upstream revision. It is not an NVIDIA repository and NVIDIA does not
+endorse this mirror.

PROVENANCE.json

@@ -0,0 +1,20 @@
+{
+  "schema_version": 1,
+  "kind": "personaplex-upstream-raw-mirror-provenance",
+  "upstream": {
+    "repo_id": "nvidia/personaplex-7b-v1",
+    "revision": "fdaf4090a61cb315c138a1faee287ffd6c716309",
+    "acquired_at": "2026-07-20T01:49:36Z"
+  },
+  "mirror": {
+    "repo_id": "cudabenchmarktest/personaplex-7b-v1-raw-mirror",
+    "visibility": "public",
+    "gated": false
+  },
+  "transformations": [],
+  "weight_identity": {
+    "path": "model.safetensors",
+    "bytes": 16742874000,
+    "sha256": "db1290db583cdaa6cb4de444ed279e0b586ca2a372b41434b07a7461c8c0e2f4"
+  }
+}

README.md

@@ -0,0 +1,71 @@
+---
+license: other
+license_name: nvidia-open-model-license
+license_link: https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/
+language:
+- en
+base_model:
+- nvidia/personaplex-7b-v1
+- kyutai/moshiko-pytorch-bf16
+library_name: moshi
+pipeline_tag: audio-to-audio
+tags:
+- personaplex
+- moshi
+- speech-to-speech
+- full-duplex
+- raw-weights
+---
+
+# PersonaPlex 7B v1 raw-weight mirror
+
+This is a public, ungated, byte-identical mirror of the model artifacts from
+[`nvidia/personaplex-7b-v1`](https://huggingface.co/nvidia/personaplex-7b-v1)
+at immutable upstream revision
+`fdaf4090a61cb315c138a1faee287ffd6c716309`.
+
+This mirror exists to make reproducible deployment, quantization, adapter
+training, and research automation possible without silently substituting the
+historical `student_best.pt` derivative. It is independently maintained and is
+not an NVIDIA repository.
+
+## Integrity and provenance
+
+| Artifact | Bytes | SHA-256 |
+|---|---:|---|
+| `model.safetensors` | 16,742,874,000 | `db1290db583cdaa6cb4de444ed279e0b586ca2a372b41434b07a7461c8c0e2f4` |
+| `tokenizer-e351c8d8-checkpoint125.safetensors` | 384,644,900 | `09b782f0629851a271227fb9d36db65c041790365f11bbe5d3d59369cf863f50` |
+| `tokenizer_spm_32k_3.model` | 552,778 | `78d4336533ddc26f9acf7250d7fb83492152196c6ea4212c841df76933f18d2d` |
+| `voices.tgz` | 6,095,521 | `8564e9ca7a06ca723b07c3a77c623f0faa5937d04b2647b3a727b06c5ca0b7bb` |
+
+`PROVENANCE.json` records the upstream revision and transformation status.
+`UPSTREAM_SHA256SUMS` covers every mirrored upstream file. The original NVIDIA
+model card is preserved as `UPSTREAM_README.md`.
+
+## Use
+
+```bash
+pip install moshi
+python -m moshi.server --hf-repo cudabenchmarktest/personaplex-7b-v1-raw-mirror
+```
+
+For the maintained PersonaPlex runtime and source instructions, use
+[`NVIDIA/personaplex`](https://github.com/NVIDIA/personaplex).
+
+## What this repository is not
+
+- It is not the historical `cudabenchmarktest/personaplex-7b-nf4-distilled`
+  checkpoint. That file is a BF16-derived training artifact with a different
+  byte size and SHA-256.
+- It is not an NF4 conversion.
+- It contains no semantic-control adapter and makes no semantic-control quality
+  claim.
+- It has not modified, merged, pruned, quantized, or fine-tuned the raw weights.
+
+## License
+
+Use and redistribution are governed by the
+[NVIDIA Open Model License](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).
+The required NVIDIA attribution is in `NOTICE`, and copies of the governing
+license materials are under `LICENSES/`. Additional base-model attribution is
+in `THIRD_PARTY_NOTICES.md`.

THIRD_PARTY_NOTICES.md

@@ -0,0 +1,10 @@
+# Third-party notices
+
+PersonaPlex is based on Moshi/Moshiko. The upstream PersonaPlex model card lists
+`kyutai/moshiko-pytorch-bf16` as the base model and identifies CC BY 4.0 as
+additional governing information. A copy of CC BY 4.0 is included under
+`LICENSES/CC-BY-4.0.txt`.
+
+The authoritative upstream model card is preserved byte-for-byte as
+`UPSTREAM_README.md`. Components with separate notices remain governed by those
+notices.

UPSTREAM_README.md

@@ -0,0 +1,218 @@
+---
+license: other
+license_name: nvidia-open-model-license
+license_link: >-
+  https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/
+language:
+- en
+base_model:
+- kyutai/moshiko-pytorch-bf16
+library_name: moshi
+extra_gated_prompt: >-
+  GOVERNING TERMS: Use of this model is governed by the [NVIDIA Open Model
+  License
+  Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/).
+  ADDITIONAL INFORMATION:
+  [CC-BY-4.0](https://huggingface.co/kyutai/moshiko-pytorch-bf16).
+pipeline_tag: audio-to-audio
+tags:
+- speech-to-speech
+- agent
+---
+
+# PersonaPlex: Voice and role control for full duplex conversational speech models
+
+<style>
+h1, h2, h3, h4, h5, h6 {
+  color: #76b900; /* NVIDIA green */
+  font-weight: 700;
+}
+
+hr {
+  border: none;
+  border-top: 1px solid #e5e7eb;
+  margin: 2rem 0;
+}
+
+/* Improve list spacing */
+ul, ol {
+  margin-top: 0.5rem;
+  margin-bottom: 0.5rem;
+}
+
+/* Badge alignment consistency */
+img {
+  display: inline;
+  vertical-align: middle;
+}
+</style>
+
+➡️ **Code:** [nvidia/personaplex](https://github.com/NVIDIA/personaplex) <br>
+➡️ **Demo:** [PersonaPlex Project Page](https://research.nvidia.com/labs/adlr/personaplex/) <br>
+➡️ **Paper:** [PersonaPlex Preprint](https://arxiv.org/abs/2602.06053) <br>
+
+
+### Description:
+Personaplex is a real-time speech-to-speech conversational model that jointly performs streaming speech understanding and speech generation. The model operates on continuous audio encoded with a neural codec and predicts both text tokens and audio tokens autoregressively to produce its spoken responses. Incoming user audio is incrementally encoded and fed to the model while Personaplex simultaneously generates its own outgoing speech, enabling natural conversational dynamics such as interruptions, barge-ins, overlaps, and rapid turn-taking.
+Personaplex runs in a dual-stream configuration in which listening and speaking occur concurrently. This design allows the model to update its internal state based on the user’s ongoing speech while still producing fluent output audio, supporting highly interactive conversations.
+Before the conversation begins, Personaplex is conditioned on two prompts: a voice prompt and a text prompt. The voice prompt consists of a sequence of audio tokens that establish the target vocal characteristics and speaking style. The text prompt specifies persona attributes such as role, background, and scenario context. Together, these prompts define the model's conversational identity and guide its linguistic and acoustic behavior throughout the interaction.
+
+This model is ready for commercial use.
+
+## Explore more from NVIDIA:
+For documentation, deployment guides, enterprise-ready APIs, and the latest open models—including Nemotron and other cutting-edge speech, translation, and generative AI—visit the NVIDIA Developer Portal at [developer.nvidia.com](https://developer.nvidia.com/).
+Join the community to access tools, support, and resources to accelerate your development with NVIDIA's NeMo, Riva, NIM, and foundation models.<br>
+
+What is [Nemotron](https://www.nvidia.com/en-us/ai-data-science/foundation-models/nemotron/)?<br>
+NVIDIA Developer [Nemotron](https://developer.nvidia.com/nemotron)<br>
+[NVIDIA Riva Speech](https://developer.nvidia.com/riva?sortBy=developer_learning_library%2Fsort%2Ffeatured_in.riva%3Adesc%2Ctitle%3Aasc#demos)<br>
+[NeMo Documentation](https://docs.nvidia.com/nemo-framework/user-guide/latest/nemotoolkit/asr/models.html)<br>
+
+### License/Terms of Use:
+GOVERNING TERMS: Use of this model is governed by the [NVIDIA Open Model License Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/). ADDITIONAL INFORMATION: [CC-BY-4.0](https://huggingface.co/kyutai/moshiko-pytorch-bf16).
+
+### Use Case: <br>
+Wherever NVIDIA’s speech-to-speech conversational models are used, PersonaPlex can generate English speech response for English speech input.
+
+### Deployment Geography:
+Global
+
+### Release Date:  <br>
+Hugging Face [01/15/2026] via [[https://huggingface.co/nvidia/personaplex-7b-v1](https://huggingface.co/nvidia/personaplex-7b-v1)] <br> 
+Github [01/15/2026] via [[https://github.com/NVIDIA/personaplex](https://github.com/NVIDIA/personaplex)] <br> 
+
+
+## Model Architecture:
+**Architecture Type:** Transformer <br>
+
+**Network Architecture:** [Moshi](https://github.com/kyutai-labs/moshi) <br>
+
+Moshi uses:
+* Mimi Speech Encoder (ConvNet, Transformer)
+* Moshi Temporal Transformer + Depth Transformer
+* Mimi Speech Decoder (Transformer, ConvNet)
+
+** This model was developed based on [Moshi (Moshiko weights)](https://huggingface.co/kyutai/moshiko-pytorch-bf16) <br> 
+** Number of model parameters: 7B <br>
+
+
+## Input(s): <br>
+**Input Type(s):** Text (prompt), Audio (user speech) <br>
+**Input Format:** String, WAV/WebAudio <br>
+**Input Parameters:** One-Dimensional (1D) <br>
+**Other Properties Related to Input:** 24kHz sample rate for audio. <br>
+
+## Output(s)
+**Output Type(s):** Text (agent text), Audio (agent speech) <br>
+**Output Format:** String, WAV/WebAudio <br>
+**Output Parameters:** One-Dimensional (1D) <br>
+**Other Properties Related to Output:** 24kHz sample rate for audio. <br>
+
+Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA’s hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions. <br> 
+
+## Software Integration:
+**Runtime Engine:** PyTorch <br> 
+
+**Supported Hardware Microarchitecture Compatibility:** <br>
+* NVIDIA Ampere (A100)
+* NVIDIA Hopper (H100)
+
+**Preferred/Supported Operating System(s):**
+* Linux
+
+The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment. <br>
+
+## Model Version(s):
+* v1.0
+
+## Training, Testing, and Evaluation Datasets:
+
+### Training Dataset:
+**Link:** Fisher English: [Part1](https://catalog.ldc.upenn.edu/LDC2004S13), [Part2](https://catalog.ldc.upenn.edu/LDC2005S13) <br>
+**Data Modality:** Audio (speech) <br>
+**Audio Training Data Size:** Less than 10,000 Hours <br>
+**Data Collection Method by dataset:** Human <br>
+**Labeling Method by dataset:** Automated <br>
+**Properties:** 7303 conversations (upto 10 minutes each).
+
+
+### Testing/Evaluation Dataset:
+**Link:** [FullDuplexBench](https://arxiv.org/abs/2503.04721) <br>
+**Data Collection Method by dataset:** Hybrid: Human, Synthetic, Automated. <br>
+**Labeling Method by dataset:** Automated. <br>
+**Properties:** The [FullDuplexBench](https://arxiv.org/abs/2503.04721) public benchmark aggregates various synthetic and real datasets. <br>
+Additionally speaker similarity (SSIM) between voice prompts and model outputs on the User Interruption portion of the FullDuplexBench benchmark were measured using [WavLM-TDNN](https://arxiv.org/pdf/2110.13900) embedding cosine similarity.
+
+**FullDuplexBench Benchmark Scores:** <br>
+| Metric                                   | Value |
+|------------------------------------------|-------|
+| Pause Handling(Synthetic): TOR↓          | 0.358 |
+| Pause Handling(Candor): TOR↓             | 0.431 |
+| Backchannel: TOR↓                        | 0.273 |
+| Backchannel: Freq↑                       | 0.042 |
+| Backchannel: JSD↓                        | 0.662 |
+| Smooth Turn Taking: TOR↑                 | 0.908 |
+| Smooth Turn Taking: Latency↓             | 0.170 |
+| User Interruption: TOR↑                  | 0.950 |
+| User Interruption: GPT-4o↑               | 4.290 |
+| User Interruption: Latency↓              | 0.240 |
+| User Interruption: SSIM(WavLM)↑          | 0.650 |
+
+
+**Comparison With Other Conversational AI Systems:**
+PersonaPlex outperforms other open-source and commercial systems on conversational dynamics, response and interruption latency, and task adherence in both question-answering assistant and customer service roles.
+
+<figure align="center">
+  <img src="figures/results_conversation_dynamics.png" width="1000" />
+  <figcaption>
+    FullDuplexBench Conversational Dynamics Evaluation. Success rate uses the Takeover Rate (TOR) metric for Smooth Turn-Taking and User Interruption, and 1-TOR for Pause Handling.
+  </figcaption>
+</figure>
+
+<figure align="center">
+  <img src="figures/results_latency.png" width="1000" />
+  <figcaption>
+    FullDuplexBench Latency Evaluation. Smooth turn-taking latency is measured as the duration from when the user stops speaking to when the agent starts responding. User interruption latency is measured as the duration from when the user interrupts the agent while it is speaking to when the agent stops speaking.
+  </figcaption>
+</figure>
+
+<figure align="center">
+  <img src="figures/results_task_adherence.png" width="1000" />
+  <figcaption>
+    Task Adherence Evaluation. FullDuplexBench scores are based on general knowledge question-answering in the "User Interruption" category. ServiceDuplexBench (to be released soon) scores are based on varied customer service scenarios. GPT-4o is used to judge the content of agent responses.
+  </figcaption>
+</figure>
+
+# Inference:
+**Acceleration Engine:** PyTorch <br>
+**Test Hardware:** NVIDIA A100 80 GB <br>
+
+
+## Ethical Considerations:
+NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications.  When downloaded or used in accordance with our terms of service, developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. <br> 
+
+For more detailed information on ethical considerations for this model, please see the Model Card++ 
+[Bias](bias.md), 
+[Explainability](explainability.md), 
+[Safety & Security](safety.md), 
+and [Privacy](privacy.md) Subcards. <br>
+
+Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns [here](https://www.nvidia.com/en-us/support/submit-security-vulnerability/).
+
+## Citation
+If you use PersonaPlex in your research, please cite our paper:
+```bibtex
+@misc{roy2026personaplexvoicerolecontrol,
+      title={PersonaPlex: Voice and Role Control for Full Duplex Conversational Speech Models}, 
+      author={Rajarshi Roy and Jonathan Raiman and Sang-gil Lee and Teodor-Dumitru Ene and Robert Kirby and Sungwon Kim and Jaehyeon Kim and Bryan Catanzaro},
+      year={2026},
+      eprint={2602.06053},
+      archivePrefix={arXiv},
+      primaryClass={cs.CL},
+      url={https://arxiv.org/abs/2602.06053}, 
+}
+```
+
+## References(s):
+1. [Moshi and Mimi](https://arxiv.org/pdf/2410.00037) <br>
+2. [FullDuplexBench](https://arxiv.org/abs/2503.04721) <br>

UPSTREAM_SHA256SUMS

@@ -0,0 +1,16 @@
+f75ce98ed382f1d6c04774da0789866e18eba9175ae0db00d35923b4839925a3  ./.gitattributes
+9af15fd1599e7231b112304a532f625008430d6219e7ac6d523a2807e9c08376  ./README.md
+aea3cf1e5fadc65a07cdf3c4ef6a729679a6283cd8960e9a9aead67cffe870fd  ./bias.md
+2d1cfef43c2cec307f2a3c7f7835b69a87c685261e785549e2c6558f811e1b08  ./config.json
+8de47fe2477491fac3dca404185d0430c8d39f9e0daf4c90ada5f03fdf830f45  ./dist.tgz
+97f9027e2d25d0534f06bad1dacc895eabbc16a4dea3b571893c8be028d63d08  ./explainability.md
+1ae3468de4cc8a886bfaddae28c6994075a75127b95c8892acb0ea46b7d6ff96  ./figures/.DS_Store
+93cc850f6bd2c0d07c43d40ac2166b6ba20ee585d4c856d09fb4eff706fd0204  ./figures/results_conversation_dynamics.png
+ffaed271517d2355566f1da635d3bbdc02fbee82077f210e8806ff057bb98850  ./figures/results_latency.png
+5a7e19c9f045fa87b3cb5edec87b5e854901dcb589f00c291cb11fbd5d0adfba  ./figures/results_task_adherence.png
+db1290db583cdaa6cb4de444ed279e0b586ca2a372b41434b07a7461c8c0e2f4  ./model.safetensors
+37eb20a609fa901bd86301a4d58082bf4e11daeee344943518c8500467b9f102  ./privacy.md
+d2a00bad1c1a60fff35656fe7facfa914657522bd268c185dffec7cf2d8595da  ./safety.md
+09b782f0629851a271227fb9d36db65c041790365f11bbe5d3d59369cf863f50  ./tokenizer-e351c8d8-checkpoint125.safetensors
+78d4336533ddc26f9acf7250d7fb83492152196c6ea4212c841df76933f18d2d  ./tokenizer_spm_32k_3.model
+8564e9ca7a06ca723b07c3a77c623f0faa5937d04b2647b3a727b06c5ca0b7bb  ./voices.tgz

bias.md

@@ -0,0 +1,5 @@
+Field                                                                                               |  Response
+:---------------------------------------------------------------------------------------------------|:---------------
+Participation considerations from adversely impacted groups [protected classes](https://www.senate.ca.gov/content/protected-classes) in model design and testing:  |  None
+Measures taken to mitigate against unwanted bias:                                                   |  None
+Bias Metric (If Measured):                                                   |  None

config.json

@@ -0,0 +1,4 @@
+{
+  "model_type": "personaplex",
+  "version": "7b-v1"
+}

dist.tgz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8de47fe2477491fac3dca404185d0430c8d39f9e0daf4c90ada5f03fdf830f45
+size 598195

explainability.md

@@ -0,0 +1,13 @@
+Field                                                                                                  |  Response
+:------------------------------------------------------------------------------------------------------|:---------------------------------------------------------------------------------
+Intended Task/Domain:                                                                   |  Speech To Speech Conversational Chat Agent 
+Model Type:                                                                                            |  Speech Encoder (ConvNet, Transformer), Temporal Transformer, Depth Transfomer, Speech Decoder (ConvNet, Transformer)
+Intended Users:                                                                                        |  People working with conversational AI systems that respond to user speech with generated speech (and optionally accompanying text).
+Output:                                                                                                |  Speech, Text
+Describe how the model works:                                                                          |  Personaplex is a real-time speech-to-speech conversational model that jointly performs streaming speech understanding and speech generation. The model operates on continuous audio encoded with a neural codec and predicts both text tokens and audio tokens autoregressively to produce its spoken responses. Incoming user audio is incrementally encoded and fed to the model while Personaplex simultaneously generates its own outgoing speech, enabling natural conversational dynamics such as interruptions, barge-ins, overlaps, and rapid turn-taking.<br> Personaplex runs in a dual-stream configuration in which listening and speaking occur concurrently. This design allows the model to update its internal state based on the user’s ongoing speech while still producing fluent output audio, supporting highly interactive conversations. <br> Prior to conversation, Personaplex is conditioned on two prompts: a voice prompt and a text prompt. The voice prompt consists of a sequence of audio tokens that establish the target vocal characteristics and speaking style. The text prompt specifies persona attributes such as role, background, and scenario context. Together, these prompts define the model's conversational identity and guide its linguistic and acoustic behavior throughout the interaction.
+Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of:  |  Not Applicable
+Technical Limitations & Mitigation:                                                                    |  Personaplex is trained with a 2048-token context window, corresponding to roughly 160 seconds of audio. Conversational context beyond this window may not be retained reliably.<br> The model’s knowledge and linguistic competence derive from its underlying Moshi base model. As a result, Personaplex may produce inaccurate or outdated responses and does not have access to recent events or comprehensive world knowledge. <br>Personaplex was not explicitly trained for reasoning or alignment. Its performance on tasks requiring multi-step reasoning, arithmetic, or safety-aligned behavior may therefore be limited.
+Verified to have met prescribed NVIDIA quality standards:  |  Yes
+Performance Metrics:                                                                                   |  Conversational dynamics metrics include takeover rate (TOR); latency (seconds) for interruptions, pause handling, and turn-taking; frequency and Jensen–Shannon divergence (JSD) for backchannels; question-answering accuracy (score); and speaker similarity (SSIM).
+Potential Known Risks:                                                                                 |  The model may present errors in speech understanding and/or pronunciation. Additionally, the training data consists exclusively of English speech. The model is not expected to generalize well to non-English languages. Personaplex was trained primarily on assistant-style, customer service, and general conversational interactions. Its behavior may degrade when applied outside these domains. For security and privacy reasons, Personaplex does not support generating voice prompts from real user voice recordings.
+Licensing:                                                                                             |  GOVERNING TERMS: Use of this model is governed by the [NVIDIA Open Model License Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/). ADDITIONAL INFORMATION: [CC-BY-4.0](https://huggingface.co/kyutai/moshiko-pytorch-bf16).

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:db1290db583cdaa6cb4de444ed279e0b586ca2a372b41434b07a7461c8c0e2f4
+size 16742874000

privacy.md

@@ -0,0 +1,15 @@
+Field                                                                                                                              |  Response
+:----------------------------------------------------------------------------------------------------------------------------------|:-----------------------------------------------
+Generatable or reverse engineerable personal data?                                                     |  No
+Personal data used to create this model?                                                                                       |  Yes - Voice
+Was consent obtained for any personal data used?                                                                                             |  Yes
+How often is dataset reviewed?                                                                                                     |  During dataset creation, model training, evaluation and before release
+Is a mechanism in place to honor data subject right of access or deletion of personal data?                                        |  Yes
+If personal data was collected for the development of the model, was it collected directly by NVIDIA?                                            |  Yes
+If personal data was collected for the development of the model by NVIDIA, do you maintain or have access to disclosures made to data subjects?  |  Yes
+If personal data was collected for the development of this AI model, was it minimized to only what was required?                                 |  Yes
+Was data from user interactions with the AI model (e.g. user input and prompts) used to train the model? |  No
+Is there provenance for all datasets used in training?                                                                                |  Yes
+Does data labeling (annotation, metadata) comply with privacy laws?                                                                |  Yes
+Is data compliant with data subject requests for data correction or removal, if such a request was made?                           |  Yes
+Applicable Privacy Policy        | https://www.nvidia.com/en-us/about-nvidia/privacy-policy/ 

safety.md

@@ -0,0 +1,6 @@
+Field                                               |  Response
+:---------------------------------------------------|:----------------------------------
+Model Application Field(s):                               |  Speech To Speech Conversational Systems 
+Describe the life critical impact (if present).   |  Not Applicable 
+Use Case Restrictions:                              |  Abide by [NVIDIA Open Model License Agreement](https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/). ADDITIONAL INFORMATION: [CC-BY-4.0](https://huggingface.co/kyutai/moshiko-pytorch-bf16).
+Model and dataset restrictions:            |  The Principle of least privilege (PoLP) is applied limiting access for dataset generation and model development.  Restrictions enforce dataset access during training, and dataset license constraints adhered to.

tokenizer-e351c8d8-checkpoint125.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09b782f0629851a271227fb9d36db65c041790365f11bbe5d3d59369cf863f50
+size 384644900

tokenizer_spm_32k_3.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78d4336533ddc26f9acf7250d7fb83492152196c6ea4212c841df76933f18d2d
+size 552778

voices.tgz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8564e9ca7a06ca723b07c3a77c623f0faa5937d04b2647b3a727b06c5ca0b7bb
+size 6095521