Text Generation
Transformers
Safetensors
llada2_moe
dllm
diffusion
llm
text_generation
conversational
custom_code
Instructions to use inclusionAI/LLaDA2.2-flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use inclusionAI/LLaDA2.2-flash with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="inclusionAI/LLaDA2.2-flash", trust_remote_code=True) messages = [ {"role": "user", "content": "Who are you?"}, ] pipe(messages)# Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("inclusionAI/LLaDA2.2-flash", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use inclusionAI/LLaDA2.2-flash with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "inclusionAI/LLaDA2.2-flash" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/chat/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "inclusionAI/LLaDA2.2-flash", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/inclusionAI/LLaDA2.2-flash
- SGLang
How to use inclusionAI/LLaDA2.2-flash 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 "inclusionAI/LLaDA2.2-flash" \ --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": "inclusionAI/LLaDA2.2-flash", "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 "inclusionAI/LLaDA2.2-flash" \ --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": "inclusionAI/LLaDA2.2-flash", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use inclusionAI/LLaDA2.2-flash with Docker Model Runner:
docker model run hf.co/inclusionAI/LLaDA2.2-flash
Create chat_template.jinja
Browse files- chat_template.jinja +159 -0
chat_template.jinja
ADDED
|
@@ -0,0 +1,159 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{#-
|
| 2 |
+
LLaDA22 chat template — NO-THINK variant.
|
| 3 |
+
Usage:
|
| 4 |
+
start_header: "<role>assistant</role>"
|
| 5 |
+
end_header: "<|role_end|>"
|
| 6 |
+
end_learnable_inclusive: true
|
| 7 |
+
#}
|
| 8 |
+
|
| 9 |
+
{#- Message Content Rendering ============================================== #}
|
| 10 |
+
{%- macro render_content(content) -%}
|
| 11 |
+
{%- if content is string -%}
|
| 12 |
+
{{- content -}}
|
| 13 |
+
{%- elif content is iterable and content is not mapping -%}
|
| 14 |
+
{%- for item in content -%}
|
| 15 |
+
{%- if item is mapping -%}
|
| 16 |
+
{%- if 'text' in item -%}
|
| 17 |
+
{{- item.text -}}
|
| 18 |
+
{%- elif 'image' in item or 'image_url' in item or item.get('type') == 'image' -%}
|
| 19 |
+
{{- raise_exception('Constraint Violation: Image data detected.') -}}
|
| 20 |
+
{%- elif 'video' in item or item.get('type') == 'video' -%}
|
| 21 |
+
{{- raise_exception('Constraint Violation: Video data detected.') -}}
|
| 22 |
+
{%- else -%}
|
| 23 |
+
{{- raise_exception('Invalid mapping structure: Missing "text" key.') -}}
|
| 24 |
+
{%- endif -%}
|
| 25 |
+
{%- elif item is string -%}
|
| 26 |
+
{{- item -}}
|
| 27 |
+
{%- else -%}
|
| 28 |
+
{{- raise_exception('Invalid item type: Must be string or mapping.') -}}
|
| 29 |
+
{%- endif -%}
|
| 30 |
+
{%- endfor -%}
|
| 31 |
+
{%- elif content is none or content is undefined -%}
|
| 32 |
+
{{- '' -}}
|
| 33 |
+
{%- else -%}
|
| 34 |
+
{{- raise_exception('Fatal error: Content must be a string, iterable, or None.') -}}
|
| 35 |
+
{%- endif -%}
|
| 36 |
+
{%- endmacro -%}
|
| 37 |
+
|
| 38 |
+
{%- macro render_toolcalls(message, ns_tool) -%}
|
| 39 |
+
{{- '<|tool_calls_section_begin|>' -}}
|
| 40 |
+
{%- for tool_call in message.get('tool_calls', []) -%}
|
| 41 |
+
{%- set original_id = tool_call.get('id', '') -%}
|
| 42 |
+
{%- set func_name = tool_call.get('function', {}).get('name', 'unknown') -%}
|
| 43 |
+
|
| 44 |
+
{%- set tool_id = 'functions.' + func_name + ':' + ns_tool.index|string -%}
|
| 45 |
+
|
| 46 |
+
{%- set ns_tool.id_map = ns_tool.id_map + [[original_id, tool_id]] -%}
|
| 47 |
+
{%- set ns_tool.index = ns_tool.index + 1 -%}
|
| 48 |
+
|
| 49 |
+
{%- set args = tool_call.get('function', {}).get('arguments', '') -%}
|
| 50 |
+
{{- '<|tool_call_begin|>' + tool_id + '<|tool_call_argument_begin|>' -}}
|
| 51 |
+
{%- if args is string -%}
|
| 52 |
+
{{- args -}}
|
| 53 |
+
{%- else -%}
|
| 54 |
+
{{- args | tojson -}}
|
| 55 |
+
{%- endif -%}
|
| 56 |
+
{{- '<|tool_call_end|>' -}}
|
| 57 |
+
{%- endfor -%}
|
| 58 |
+
{{- '<|tool_calls_section_end|>' -}}
|
| 59 |
+
{%- endmacro -%}
|
| 60 |
+
|
| 61 |
+
{%- if not messages -%}
|
| 62 |
+
{{- raise_exception('No messages provided.') -}}
|
| 63 |
+
{%- endif -%}
|
| 64 |
+
|
| 65 |
+
{# 1. System Instruction, Model Identity & Tools Injection #}
|
| 66 |
+
{%- set ns_sys = namespace(has_system=false, content='') -%}
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
{%- if messages[0].role == 'system' -%}
|
| 70 |
+
{%- set ns_sys.has_system = true -%}
|
| 71 |
+
{%- if ns_sys.content -%}
|
| 72 |
+
{%- set ns_sys.content = ns_sys.content + '\n\n' -%}
|
| 73 |
+
{%- endif -%}
|
| 74 |
+
{%- set ns_sys.content = ns_sys.content + render_content(messages[0].content) -%}
|
| 75 |
+
{%- endif -%}
|
| 76 |
+
|
| 77 |
+
{%- if tools or ns_sys.content -%}
|
| 78 |
+
{{- '<role>system</role>' -}}
|
| 79 |
+
|
| 80 |
+
{%- if ns_sys.content -%}
|
| 81 |
+
{{- ns_sys.content -}}
|
| 82 |
+
{%- if tools -%}{{ '\n\n' }}{%- endif -%}
|
| 83 |
+
{%- endif -%}
|
| 84 |
+
|
| 85 |
+
{#- Tool Definition Rendering -#}
|
| 86 |
+
{%- if tools and tools is iterable and tools is not mapping -%}
|
| 87 |
+
{{- "<tools>\n" -}}
|
| 88 |
+
{%- if tools_ts_str -%}
|
| 89 |
+
{{- tools_ts_str -}}
|
| 90 |
+
{%- else -%}
|
| 91 |
+
{%- for tool in tools -%}
|
| 92 |
+
{{- tool | tojson(ensure_ascii=False) -}}{{ '\n' -}}
|
| 93 |
+
{%- endfor -%}
|
| 94 |
+
{%- endif -%}
|
| 95 |
+
{{- "</tools>\n" -}}
|
| 96 |
+
{%- endif -%}
|
| 97 |
+
|
| 98 |
+
{{- '<|role_end|>' -}}
|
| 99 |
+
{%- endif -%}
|
| 100 |
+
|
| 101 |
+
{# 2. Main Message Rendering Loop #}
|
| 102 |
+
{%- set ns_tool = namespace(index=0, id_map=[]) -%}
|
| 103 |
+
|
| 104 |
+
{%- for message in messages -%}
|
| 105 |
+
{# Skip the first System message as it is pre-rendered above #}
|
| 106 |
+
{%- if not (loop.first and message.role == 'system') -%}
|
| 107 |
+
|
| 108 |
+
{%- set content_text = render_content(message.content) -%}
|
| 109 |
+
{%- set role_name = message.get('name') or message.role -%}
|
| 110 |
+
|
| 111 |
+
{%- if message.role == 'user' or message.role == 'system' -%}
|
| 112 |
+
{{- '<role>' + role_name + '</role>' -}}
|
| 113 |
+
{{- content_text + '<|role_end|>' -}}
|
| 114 |
+
|
| 115 |
+
{%- elif message.role == 'assistant' -%}
|
| 116 |
+
{{- '<role>' + role_name + '</role>' -}}
|
| 117 |
+
|
| 118 |
+
{#- NO-THINK: completely ignore reasoning_content, strip <think> from content -#}
|
| 119 |
+
{%- set final_content = content_text -%}
|
| 120 |
+
|
| 121 |
+
{#- Strip embedded <think>...</think> from content if present -#}
|
| 122 |
+
{%- if '<think>' in final_content and '</think>' in final_content -%}
|
| 123 |
+
{%- set parts = final_content.split('</think>', 1) -%}
|
| 124 |
+
{%- set final_content = parts[1].strip() -%}
|
| 125 |
+
{%- endif -%}
|
| 126 |
+
|
| 127 |
+
{#- Output content directly, no think tags -#}
|
| 128 |
+
{%- if final_content -%}
|
| 129 |
+
{{- final_content -}}
|
| 130 |
+
{%- endif -%}
|
| 131 |
+
|
| 132 |
+
{%- if message.tool_calls -%}
|
| 133 |
+
{{- render_toolcalls(message, ns_tool) -}}
|
| 134 |
+
{%- endif -%}
|
| 135 |
+
|
| 136 |
+
{{- '<|role_end|>' -}}
|
| 137 |
+
|
| 138 |
+
{%- elif message.role == 'tool' -%}
|
| 139 |
+
{{- '<role>tool</role>' -}}
|
| 140 |
+
{%- set original_id = message.get('tool_call_id', '') -%}
|
| 141 |
+
|
| 142 |
+
{%- set ns_lookup = namespace(mapped_id=original_id) -%}
|
| 143 |
+
{%- for pair in ns_tool.id_map -%}
|
| 144 |
+
{%- if pair[0] == original_id -%}
|
| 145 |
+
{%- set ns_lookup.mapped_id = pair[1] -%}
|
| 146 |
+
{%- endif -%}
|
| 147 |
+
{%- endfor -%}
|
| 148 |
+
|
| 149 |
+
{{- '## Return of ' + ns_lookup.mapped_id + '\n' -}}
|
| 150 |
+
{{- content_text + '\n<|role_end|>' -}}
|
| 151 |
+
{%- endif -%}
|
| 152 |
+
|
| 153 |
+
{%- endif -%}
|
| 154 |
+
{%- endfor -%}
|
| 155 |
+
|
| 156 |
+
{# 3. Generation Prompt Suffix — no think tag #}
|
| 157 |
+
{%- if add_generation_prompt -%}
|
| 158 |
+
{{- '<role>assistant</role>' -}}
|
| 159 |
+
{%- endif -%}
|