Instructions to use onnx-internal-testing/tiny-random-NemotronHForCausalLM with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use onnx-internal-testing/tiny-random-NemotronHForCausalLM with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="onnx-internal-testing/tiny-random-NemotronHForCausalLM", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("onnx-internal-testing/tiny-random-NemotronHForCausalLM", trust_remote_code=True) model = AutoModelForCausalLM.from_pretrained("onnx-internal-testing/tiny-random-NemotronHForCausalLM", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] inputs = tokenizer.apply_chat_template( messages, add_generation_prompt=True, tokenize=True, return_dict=True, return_tensors="pt", ).to(model.device) outputs = model.generate(**inputs, max_new_tokens=40) print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use onnx-internal-testing/tiny-random-NemotronHForCausalLM with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "onnx-internal-testing/tiny-random-NemotronHForCausalLM" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "onnx-internal-testing/tiny-random-NemotronHForCausalLM", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/onnx-internal-testing/tiny-random-NemotronHForCausalLM
- SGLang
How to use onnx-internal-testing/tiny-random-NemotronHForCausalLM with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "onnx-internal-testing/tiny-random-NemotronHForCausalLM" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "onnx-internal-testing/tiny-random-NemotronHForCausalLM", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "onnx-internal-testing/tiny-random-NemotronHForCausalLM" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "onnx-internal-testing/tiny-random-NemotronHForCausalLM", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use onnx-internal-testing/tiny-random-NemotronHForCausalLM with Docker Model Runner:
docker model run hf.co/onnx-internal-testing/tiny-random-NemotronHForCausalLM
File size: 11,233 Bytes
17befda c6c9c3d | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 | {
"add_prefix_space": false,
"backend": "tokenizers",
"bos_token": "<s>",
"clean_up_tokenization_spaces": false,
"eos_token": "<|im_end|>",
"is_local": false,
"model_input_names": [
"input_ids",
"attention_mask"
],
"model_max_length": 262144,
"pad_token": "<|im_end|>",
"tokenizer_class": "TokenizersBackend",
"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 | safe) ~ '</' ~ 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 truncate_history_thinking = truncate_history_thinking if truncate_history_thinking is defined else True %}\n\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\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{# Recompute last_user_idx relative to loop_messages after handling system #}\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 | safe) ~ '</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 | safe) ~ '</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 {{- '\\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\n\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\n{%- for message in loop_messages %}\n {%- if message.role == \"assistant\" %}\n {# Add reasoning content in to content field for unified processing below. #}\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 ~ \"\\n</think>\\n\" ~ (message.content | default('', true)) %}\n {%- else %}\n {%- set content = message.content | default('', true) %}\n {%- if content is string -%}\n {# Allow downstream logic to to take care of broken thought, only handle coherent reasoning here. #}\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 {# Assistant message has tool calls. #}\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 {# Keep only content after the last closing think. Also generation prompt causes this. #}\n {%- set c = c.split('</think>')[-1] %}\n {%- elif '<think>' in c %}\n {# If <think> was opened but never closed, drop the trailing think segment #}\n {%- set c = c.split('<think>')[0] %}\n {%- endif %}\n {%- set c = \"<think></think>\" ~ c | trim %}\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 {# Assistant message doesn't have tool calls. #}\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 {{- content }}\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\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 %}\n"
} |