Add files using upload-large-folder tool

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text
+accuracy_plot.png filter=lfs diff=lfs merge=lfs -text

bias.md

@@ -0,0 +1,10 @@
+| Field | Response |
+| :---- | :---- |
+| Participation considerations from adversely impacted groups [protected classes](https://www.senate.ca.gov/content/protected-classes) in model design and testing: | None |
+| Bias Metric (If Measured): | [BBQ Accuracy Scores in Ambiguous Contexts](https://github.com/nyu-mll/BBQ/) |
+| Which characteristic (feature) show(s) the greatest difference in performance?: | The model shows high variance in the characteristics when it is used with a high temperature. |
+| Measures taken to mitigate against unwanted bias: | Supervised Fine-Tuning (SFT) and Reinforcement Learning from Human Feedback (RLHF) employed to calibrate the model’s reasoning capabilities to maintain logical consistency and appropriate complexity when interacting with or interpreting data from diverse age demographics. |
+| If using internal data, description of methods implemented in data acquisition or processing, if any, to address the prevalence of identifiable biases in the training, testing, and validation data: | The training datasets contain a large amount of synthetic data generated by LLMs. We manually curated prompts. |
+| Tools used to assess statistical imbalances and highlight patterns that may introduce bias into AI models: | [BBQ](https://github.com/nyu-mll/BBQ/) |
+| Tools used to assess statistical imbalances and highlight patterns that may introduce bias into AI models: | These datasets, such as web-scraped finance reasoning data derived from SEC EDGAR filings, science and math problem datasets, OpenResearcher/source-document datasets, Common Crawl, CC-News, Wikimedia, and long-context document datasets, do not collectively or exhaustively represent all demographic groups (and proportionally therein). For instance, these datasets do not contain explicit mentions of demographic classes such as age, gender, or ethnicity in approximately 97% to 99.9% of finance reasoning samples and in over 85% of samples across the broader assessed datasets. In the subset where such terms are present, these datasets contain notable representational skews. For example, ethnicity mentions are often dominated by Middle Eastern contexts (found in finance documents) or "White," "Two or more," and "Black or African American" as the most frequent ethnic identifiers, while references categorized as male-only significantly outnumber those categorized as female-only. Furthermore, gender is explicitly mentioned in approximately 12% of samples across the broader dataset assessment, yet in only 0.9% of finance-specific samples. Dataset-level results vary by source type, with long-context/source-document datasets containing higher explicit demographic mention rates compared to certain web-scraped sources. To mitigate these imbalances, we recommend considering evaluation techniques such as bias audits, fine-tuning with demographically balanced datasets, and mitigation strategies such as counterfactual data augmentation to align with the desired model behavior. This evaluation used a 3,000-sample subset per dataset, identified as the optimal threshold for maximizing embedder accuracy. |
+| Unwanted Bias Testing: | Constrained to English-language inputs. Multi-lingual parity is not currently claimed or guaranteed. |

chat_template.jinja

@@ -2,7 +2,7 @@
     {%- if json_dict is mapping %}
         {%- for json_key in json_dict if json_key not in handled_keys %}
             {%- if json_dict[json_key] is mapping or (json_dict[json_key] is sequence and json_dict[json_key] is not string) %}
-                {{- '\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | tojson ) ~ '</' ~ json_key ~ '>' }}
+                {{- '\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | tojson | safe) ~ '</' ~ json_key ~ '>' }}
             {%- else %}
                 {{-'\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | string) ~ '</' ~ json_key ~ '>' }}
             {%- endif %}
@@ -68,7 +68,7 @@
                     {{- '\n<description>' ~ (param_fields.description | trim) ~ '</description>' }}
                 {%- endif %}
                 {%- if param_fields.enum is defined %}
-                    {{- '\n<enum>' ~ (param_fields.enum | tojson ) ~ '</enum>' }}
+                    {{- '\n<enum>' ~ (param_fields.enum | tojson | safe) ~ '</enum>' }}
                 {%- endif %}
                 {%- set handled_keys = ['name', 'type', 'description', 'enum'] %}
                 {{- render_extra_keys(param_fields, handled_keys) }}
@@ -78,7 +78,7 @@
         {% set handled_keys = ['type', 'properties', 'required'] %}
         {{- render_extra_keys(tool.parameters, handled_keys) }}
         {%- if tool.parameters is defined and tool.parameters.required is defined %}
-            {{- '\n<required>' ~ (tool.parameters.required | tojson ) ~ '</required>' }}
+            {{- '\n<required>' ~ (tool.parameters.required | tojson | safe) ~ '</required>' }}
         {%- endif %}
         {{- '\n</parameters>' }}
         {%- set handled_keys = ['type', 'name', 'description', 'parameters'] %}

config.json

@@ -7,7 +7,7 @@
   "bos_token_id": 1,
   "chunk_size": 128,
   "conv_kernel": 4,
-  "torch_dtype": "bfloat16",
+  "dtype": "bfloat16",
   "eos_token_id": 2,
   "expand": 2,
   "head_dim": 128,
@@ -164,10 +164,6 @@
   "ssm_state_size": 128,
   "tie_word_embeddings": false,
   "time_step_floor": 0.0001,
-  "time_step_limit": [
-    0.0,
-    Infinity
-  ],
   "time_step_max": 0.1,
   "time_step_min": 0.001,
   "topk_group": 1,
@@ -176,7 +172,5 @@
   "use_cache": true,
   "use_conv_bias": true,
   "use_mamba_kernels": true,
-  "vocab_size": 131072,
-  "mtp_num_hidden_layers": 1,
-  "unsloth_fixed_mtp": true
-}
+  "vocab_size": 131072
+}

explainability.md

@@ -0,0 +1,14 @@
+| Field | Response |
+| :---- | :---- |
+| Intended Task/Domain: | Text generation, reasoning, and chat |
+| Model Type: | Text-to-text Mamba2-Transformer Hybrid |
+| Intended Users: | Generative AI creators working with conversational AI models and image content. |
+| Output: | Text |
+| Tools used to evaluate datasets to identify synthetic data and ensure data authenticity. | We used a Gemma-3 4B-based filtering model fine-tuned on [Nemotron Content Safety Dataset v2](https://huggingface.co/datasets/nvidia/Aegis-AI-Content-Safety-Dataset-2.0) to ensure the quality of synthetic data. |
+| Describe how the model works: | Generates text by predicting the next word or token based on the context provided in the input sequence using multiple self-attention layers. |
+| Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of: | Age, Disability Status, Gender Identity, Nationality, Physical Appearance, Ethnicity, Socioeconomic Status, Sexual Orientation, Religion |
+| Technical Limitations & Mitigation: | This model performs particularly well in instruction following regimes, as such may be strongly influenced by untrusted inputs and should be paired with appropriate guardrails and data filtering to better align use-case behaviors when exposed to such data. |
+| Verified to have met prescribed NVIDIA quality standards: | Yes |
+| Performance Metrics: | Accuracy, Throughput, and User-side throughput |
+| Potential Known Risks: | The model was optimized explicitly for instruction following and as such is more susceptible to prompt injection and jailbreaking in various forms as a result of its instruction tuning. This means that the model should be paired with additional rails or system filtering to limit exposure to instructions from malicious sources -- either directly or indirectly by retrieval (e.g. via visiting a website) -- as they may yield outputs that can lead to harmful, system-level outcomes up to and including remote code execution in agentic systems when effective security controls including guardrails are not in place. The model may generate answers that may be inaccurate, omit key information, include irrelevant or redundant text, or produce socially unacceptable or undesirable text, even if the prompt itself does not include anything explicitly offensive. |
+| Licensing: | Use of this model is governed by the [OpenMDW License Agreement, version 1.1](https://raw.githubusercontent.com/OpenMDW/OpenMDW/refs/heads/main/1.1/LICENSE.OpenMDW-1.1) (OpenMDW-1.1). |

privacy.md

@@ -0,0 +1,5 @@
+| Privacy Information |
+| :--- |
+| Nemotron 3 Ultra was trained on large-scale publicly available data that may contain images, audio-video, and text relating to people. NVIDIA collected and used this data in compliance with applicable data protection and privacy laws. This model was not designed to derive insights or otherwise learn from any personal data contained in the datasets. |
+| NVIDIA uses a combination of filters, data minimization techniques, and other guardrails to help prevent personal data from being recited by our models. We employ automated tools and data processing techniques during pre-training or training to identify and filter certain categories of personal data. |
+| Please review NVIDIA's [Privacy Policy](https://www.nvidia.com/en-us/about-nvidia/privacy-policy/) for more information. |

safety.md

@@ -0,0 +1,9 @@
+| Field | Response |
+| :---- | :---- |
+| Model Application Field(s): | Chat, Instruction Following, Chatbot Development, Code Generation, Reasoning, Customer Service |
+| Describe the life critical impact (if present). | Not Applicable |
+| Description of methods implemented in data acquisition or processing, if any, to address other types of potentially harmful data in the training, testing, and validation data: | We used a guard model for content safety to exclude potentially harmful data from training.  |
+| Description of any methods implemented in data acquisition or processing, if any, to address illegal or harmful content in the training data, including, but not limited to, child sexual abuse material (CSAM) and non-consensual intimate imagery (NCII) | We used a Gemma-3 4B-based guard model trained on [Nemotron Content Safety Dataset v2](https://huggingface.co/datasets/nvidia/Aegis-AI-Content-Safety-Dataset-2.0) for content safety to exclude potentially illegal or harmful content from the training. |
+| Use Case Restrictions: | Use of this model is governed by the [OpenMDW License Agreement, version 1.1](https://raw.githubusercontent.com/OpenMDW/OpenMDW/refs/heads/main/1.1/LICENSE.OpenMDW-1.1) (OpenMDW-1.1).|
+| 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. |
+| This AI model was developed based on our policies to ensure responsible data handling and risk mitigation. The datasets used for training have been scanned for harmful content and illegal content, consistent with our policies including scanning for Child Sexual Abuse Material (CSAM). Ongoing review and monitoring mechanisms are in place based on our policies and to maintain data integrity. | True. We use [Nemotron Content Safety Dataset V2](https://huggingface.co/datasets/nvidia/Aegis-AI-Content-Safety-Dataset-2.0) and an internal safety dataset specialized for minority sexuality for content safety evaluation to ensure the safety of this model. |

special_tokens_map.json

@@ -14,7 +14,7 @@
     "single_word": false
   },
   "pad_token": {
-    "content": "<SPECIAL_999>",
+    "content": "<|im_end|>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,

tokenizer_config.json

@@ -8013,9 +8013,7 @@
     "attention_mask"
   ],
   "model_max_length": 262144,
-  "pad_token": "<SPECIAL_999>",
-  "padding_side": "left",
+  "pad_token": "<|im_end|>",
   "tokenizer_class": "PreTrainedTokenizerFast",
-  "unk_token": "<unk>",
-  "chat_template": "{% macro render_extra_keys(json_dict, handled_keys) %}\n    {%- if json_dict is mapping %}\n        {%- for json_key in json_dict if json_key not in handled_keys %}\n            {%- if json_dict[json_key] is mapping or (json_dict[json_key] is sequence and json_dict[json_key] is not string) %}\n                {{- '\\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | tojson ) ~ '</' ~ json_key ~ '>' }}\n            {%- else %}\n                {{-'\\n<' ~ json_key ~ '>' ~ (json_dict[json_key] | string) ~ '</' ~ json_key ~ '>' }}\n            {%- endif %}\n        {%- endfor %}\n    {%- endif %}\n{% endmacro %}\n{%- set enable_thinking = enable_thinking if enable_thinking is defined else True %}\n{%- set medium_effort = medium_effort if medium_effort is defined else False %}\n{%- set truncate_history_thinking = truncate_history_thinking if truncate_history_thinking is defined else True %}\n{%- set ns = namespace(last_user_idx = -1) %}\n{%- set loop_messages = messages %}\n{%- for m in loop_messages %}\n  {%- if m[\"role\"] == \"user\" %}\n    {%- set ns.last_user_idx = loop.index0 %}\n  {%- endif %}\n{%- endfor %}\n{%- if messages[0][\"role\"] == \"system\" %}\n    {%- set system_message = messages[0][\"content\"] %}\n    {%- set loop_messages = messages[1:] %}\n{%- else %}\n    {%- set system_message = \"\" %}\n    {%- set loop_messages = messages %}\n{%- endif %}\n{%- if not tools is defined %}\n    {%- set tools = [] %}\n{%- endif %}\n{%- set ns = namespace(last_user_idx = -1) %}\n{%- for m in loop_messages %}\n  {%- if m[\"role\"] == \"user\" %}\n    {%- set ns.last_user_idx = loop.index0 %}\n  {%- endif %}\n{%- endfor %}\n{%- if system_message is defined %}\n    {{- \"<|im_start|>system\\n\" + system_message }}\n{%- else %}\n    {%- if tools is iterable and tools | length > 0 %}\n        {{- \"<|im_start|>system\\n\" }}\n    {%- endif %}\n{%- endif %}\n{%- if tools is iterable and tools | length > 0 %}\n    {%- if system_message is defined and system_message | length > 0 %}\n        {{- \"\\n\\n\" }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou have access to the following functions:\\n\\n\" }}\n    {{- \"<tools>\" }}\n    {%- for tool in tools %}\n        {%- if tool.function is defined %}\n            {%- set tool = tool.function %}\n        {%- endif %}\n        {{- \"\\n<function>\\n<name>\" ~ tool.name ~ \"</name>\" }}\n        {%- if tool.description is defined %}\n            {{- '\\n<description>' ~ (tool.description | trim) ~ '</description>' }}\n        {%- endif %}\n        {{- '\\n<parameters>' }}\n        {%- if tool.parameters is defined and tool.parameters is mapping and tool.parameters.properties is defined and tool.parameters.properties is mapping %}\n            {%- for param_name, param_fields in tool.parameters.properties|items %}\n                {{- '\\n<parameter>' }}\n                {{- '\\n<name>' ~ param_name ~ '</name>' }}\n                {%- if param_fields.type is defined %}\n                    {{- '\\n<type>' ~ (param_fields.type | string) ~ '</type>' }}\n                {%- endif %}\n                {%- if param_fields.description is defined %}\n                    {{- '\\n<description>' ~ (param_fields.description | trim) ~ '</description>' }}\n                {%- endif %}\n                {%- if param_fields.enum is defined %}\n                    {{- '\\n<enum>' ~ (param_fields.enum | tojson ) ~ '</enum>' }}\n                {%- endif %}\n                {%- set handled_keys = ['name', 'type', 'description', 'enum'] %}\n                {{- render_extra_keys(param_fields, handled_keys) }}\n                {{- '\\n</parameter>' }}\n            {%- endfor %}\n        {%- endif %}\n        {% set handled_keys = ['type', 'properties', 'required'] %}\n        {{- render_extra_keys(tool.parameters, handled_keys) }}\n        {%- if tool.parameters is defined and tool.parameters.required is defined %}\n            {{- '\\n<required>' ~ (tool.parameters.required | tojson ) ~ '</required>' }}\n        {%- endif %}\n        {{- '\\n</parameters>' }}\n        {%- set handled_keys = ['type', 'name', 'description', 'parameters'] %}\n        {{- render_extra_keys(tool, handled_keys) }}\n        {{- '\\n</function>' }}\n    {%- endfor %}\n    {{- \"\\n</tools>\" }}\n    {{- '\\n\\nIf you choose to call a function ONLY reply in the following format with NO suffix:\\n\\n<tool_call>\\n<function=example_function_name>\\n<parameter=example_parameter_1>\\nvalue_1\\n</parameter>\\n<parameter=example_parameter_2>\\nThis is the value for the second parameter\\nthat can span\\nmultiple lines\\n</parameter>\\n</function>\\n</tool_call>\\n\\n<IMPORTANT>\\nReminder:\\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\\n- Required parameters MUST be specified\\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\\n</IMPORTANT>' }}\n{%- endif %}\n{%- if system_message is defined %}\n    {{- '<|im_end|>\\n' }}\n{%- else %}\n    {%- if tools is iterable and tools | length > 0 %}\n        {{- '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- for message in loop_messages %}\n    {%- if message.role == \"assistant\" %}\n        {%- if message.reasoning_content is defined and message.reasoning_content is string and message.reasoning_content | trim | length > 0 %}\n            {%- set content = \"<think>\\n\" ~ message.reasoning_content ~ \"</think>\" ~ (message.content | default('', true)) %}\n        {%- else %}\n            {%- set content = message.content | default('', true) %}\n            {%- if content is string -%}\n                {%- if '<think>' not in content and '</think>' not in content -%}\n                    {%- set content = \"<think></think>\" ~ content -%}\n                {%- endif -%}\n            {%- else -%}\n                {%- set content = content -%}\n            {%- endif -%}\n        {%- endif %}\n        {%- if message.tool_calls is defined and message.tool_calls is iterable and message.tool_calls | length > 0 %}\n            {{- '<|im_start|>assistant\\n' }}\n                {%- set include_content = not (truncate_history_thinking and loop.index0 < ns.last_user_idx) %}\n                {%- if content is string and content | trim | length > 0 %}\n                    {%- if include_content %}\n                        {{- (content | trim) ~ '\\n' -}}\n                    {%- else %}\n                        {%- set c = (content | string) %}\n                        {%- if '</think>' in c %}\n                            {%- set c = c.split('</think>')[-1] %}\n                        {%- elif '<think>' in c %}\n                            {%- set c = c.split('<think>')[0] %}\n                        {%- endif %}\n                        {%- set c = \"<think></think>\" ~ c %}\n                        {%- if c | length > 0 %}\n                            {{- c ~ '\\n' -}}\n                        {%- endif %}\n                    {%- endif %}\n                {%- else %}\n                    {{- \"<think></think>\" -}}\n                {%- endif %}\n                {%- for tool_call in message.tool_calls %}\n                    {%- if tool_call.function is defined %}\n                        {%- set tool_call = tool_call.function %}\n                    {%- endif %}\n                    {{- '<tool_call>\\n<function=' ~ tool_call.name ~ '>\\n' -}}\n                        {%- if tool_call.arguments is defined %}\n                            {%- for args_name, args_value in tool_call.arguments|items %}\n                                {{- '<parameter=' ~ args_name ~ '>\\n' -}}\n                                    {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}\n                                {{- args_value ~ '\\n</parameter>\\n' -}}\n                            {%- endfor %}\n                        {%- endif %}\n                    {{- '</function>\\n</tool_call>\\n' -}}\n                {%- endfor %}\n                {{- '<|im_end|>\\n' }}\n        {%- else %}\n            {%- if not (truncate_history_thinking and loop.index0 < ns.last_user_idx) %}\n                {{- '<|im_start|>assistant\\n' ~ (content | default('', true) | string | trim) ~ '<|im_end|>\\n' }}\n            {%- else %}\n                {%- set c = (content | default('', true) | string) %}\n                {%- if '<think>' in c and '</think>' in c %}\n                    {%- set c = \"<think></think>\" ~ c.split('</think>')[-1] %}\n                {%- endif %}\n                {%- set c = c | trim %}\n                {%- if c | length > 0 %}\n                    {{- '<|im_start|>assistant\\n' ~ c ~ '<|im_end|>\\n' }}\n                {%- else %}\n                    {{- '<|im_start|>assistant\\n<|im_end|>\\n' }}\n                {%- endif %}\n            {%- endif %}\n        {%- endif %}\n    {%- elif message.role == \"user\" or message.role == \"system\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- set content = message.content | string %}\n        {%- if message.role == \"user\" and loop.index0 == ns.last_user_idx and medium_effort %}\n            {{- content + '\\n\\n{reasoning effort: efficient}' }}\n        {%- else %}\n            {{- content }}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.previtem and loop.previtem.role != \"tool\" %}\n            {{- '<|im_start|>user\\n' }}\n        {%- endif %}\n        {{- '<tool_response>\\n' }}\n        {{- message.content }}\n        {{- '\\n</tool_response>\\n' }}\n        {%- if not loop.last and loop.nextitem.role != \"tool\" %}\n            {{- '<|im_end|>\\n' }}\n        {%- elif loop.last %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- else %}\n        {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {%- if enable_thinking %}\n        {{- '<|im_start|>assistant\\n<think>\\n' }}\n    {%- else %}\n        {{- '<|im_start|>assistant\\n<think></think>' }}\n    {%- endif %}\n{%- endif %}"
-}
+  "unk_token": "<unk>"
+}