Text Generation
Transformers
Safetensors
English
Chinese
glm5_next
image-text-to-text
conversational
Eval Results
fp8
Instructions to use zai-org/GLM-5.3-Flash with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use zai-org/GLM-5.3-Flash with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="zai-org/GLM-5.3-Flash") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] pipe(text=messages)# Load model directly from transformers import AutoProcessor, AutoModelForMultimodalLM processor = AutoProcessor.from_pretrained("zai-org/GLM-5.3-Flash") model = AutoModelForMultimodalLM.from_pretrained("zai-org/GLM-5.3-Flash", device_map="auto") messages = [ { "role": "user", "content": [ {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"}, {"type": "text", "text": "What animal is on the candy?"} ] }, ] inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:])) - Inference
- HuggingChat
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use zai-org/GLM-5.3-Flash with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "zai-org/GLM-5.3-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": "zai-org/GLM-5.3-Flash", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }'Use Docker
docker model run hf.co/zai-org/GLM-5.3-Flash
- SGLang
How to use zai-org/GLM-5.3-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 "zai-org/GLM-5.3-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": "zai-org/GLM-5.3-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 "zai-org/GLM-5.3-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": "zai-org/GLM-5.3-Flash", "messages": [ { "role": "user", "content": "What is the capital of France?" } ] }' - Docker Model Runner
How to use zai-org/GLM-5.3-Flash with Docker Model Runner:
docker model run hf.co/zai-org/GLM-5.3-Flash
zRzRzRzRzRzRzR commited on
Commit ·
c5b82b6
1
Parent(s): dbc994b
update template
Browse files- .ms_upload_cache +0 -1
- chat_template.jinja +12 -59
.ms_upload_cache
DELETED
|
@@ -1 +0,0 @@
|
|
| 1 |
-
{"version": 3, "repo_id": "ZhipuAI/GLM-5.3-Flash", "files": {"chat_template.jinja|1787740514.0|10644": {"hash": "34d5ee66b12fa6446cdae131c352b8f68cd85369e0e6fda115583805fada3891", "size": 10644}, "config.json|1787740514.0|68794": {"hash": "00c3ce40c92fae7f248c72a8fff3163e74bda2fe0a82d190806ad2284bcddde8", "size": 68794}, "generation_config.json|1787740514.0|194": {"hash": "230c30609ecbbb9e6583bedde8e7bdda0c6eb8fe5fad0eaeb3d1b293d751cb4f", "size": 194}, "model-00003-of-00062.safetensors|1787739040.0|5363448800": {"hash": "e9806d89bb028a51393414a5817d104ca811def3459fcff82dc37616adcdafe2", "size": 5363448800}, "model-00006-of-00062.safetensors|1787739068.0|5361790216": {"hash": "20e2cd306152f5de1e1d4ffa474e90a466d1b2ffd01c21a4182a99e263e9eaa0", "size": 5361790216}, "model-00004-of-00062.safetensors|1787739049.0|5364323736": {"hash": "faaeaf32f8b9219a3a13437bf3b5436928d39d619a82c9c375c21f995c1607c0", "size": 5364323736}, "model-00007-of-00062.safetensors|1787739078.0|5364323968": {"hash": "00b63fd168956e79228899d1a0663f045f13e6d6466c01ef93cc067aad7803da", "size": 5364323968}, "model-00005-of-00062.safetensors|1787739058.0|5363896872": {"hash": "d2124352d6cb564db9144310fff9df6ca73c7d848514ce54d0ce627bc92d4b8f", "size": 5363896872}, "model-00008-of-00062.safetensors|1787739087.0|5364323752": {"hash": "6605c4a52d76579574edfe3b7084c46d91ded3f33b678be25ee36ef724175d87", "size": 5364323752}, "model-00002-of-00062.safetensors|1787739031.0|5320647024": {"hash": "d05313aa3d972efc49076d3b7fc4ed3e827cd5648fce6fd83e8e6e4caed68a1b", "size": 5320647024}, "model-00001-of-00062.safetensors|1787739025.0|5365306688": {"hash": "fd5202d28efdd01b47557c3f59111de3b4c4c375b2709c167b9b61834663c7da", "size": 5365306688}}}
|
|
|
|
|
|
chat_template.jinja
CHANGED
|
@@ -1,7 +1,6 @@
|
|
| 1 |
[gMASK]<sop>
|
| 2 |
{%- set effective_reasoning_effort = reasoning_effort if reasoning_effort is defined and reasoning_effort in ['low', 'high'] else 'max' -%}
|
| 3 |
{%- if effective_reasoning_effort is not none -%}<|system|>Reasoning Effort: {{ effective_reasoning_effort | capitalize }}{%- endif -%}
|
| 4 |
-
{%- set clear_thinking = clear_thinking if clear_thinking is defined else false -%}
|
| 5 |
{%- if tools -%}
|
| 6 |
{%- macro tool_to_json(tool) -%}
|
| 7 |
{%- set ns_tool = namespace(first=true) -%}
|
|
@@ -48,24 +47,18 @@ You are provided with function signatures within <tools></tools> XML tags:
|
|
| 48 |
|
| 49 |
For each function call, output the function name and arguments within the following XML format:
|
| 50 |
<tool_call>{function-name}<arg_key>{arg-key-1}</arg_key><arg_value>{arg-value-1}</arg_value><arg_key>{arg-key-2}</arg_key><arg_value>{arg-value-2}</arg_value>...</tool_call>{%- endif -%}
|
| 51 |
-
{%- macro emit_image() -%}<|begin_of_image|><|image|><|end_of_image|>{%- endmacro -%}
|
| 52 |
-
{%- macro emit_video() -%}<|begin_of_video|><|video|><|end_of_video|>{%- endmacro -%}
|
| 53 |
-
{%- macro emit_audio() -%}<|begin_of_audio|><|end_of_audio|>{%- endmacro -%}
|
| 54 |
{%- macro visible_text(content) -%}
|
| 55 |
{%- if content is string -%}
|
| 56 |
-
{{- content
|
| 57 |
{%- elif content is iterable and content is not mapping -%}
|
| 58 |
{%- for item in content -%}
|
| 59 |
{%- if item is mapping and item.type == 'text' -%}
|
| 60 |
-
{{- item.text
|
| 61 |
{%- elif item is string -%}
|
| 62 |
-
{{- item
|
| 63 |
-
{%- elif item is mapping and item.type in ['image', 'image_url'] -%}
|
| 64 |
-
{
|
| 65 |
-
|
| 66 |
-
{{- emit_video() -}}
|
| 67 |
-
{%- elif item is mapping and item.type in ['audio', 'audio_url', 'input_audio'] -%}
|
| 68 |
-
{{- emit_audio() -}}
|
| 69 |
{%- endif -%}
|
| 70 |
{%- endfor -%}
|
| 71 |
{%- else -%}
|
|
@@ -102,33 +95,6 @@ For each function call, output the function name and arguments within the follow
|
|
| 102 |
{%- macro is_list_of_outputs(m) -%}
|
| 103 |
{%- if m.content and m.content.0.output is defined -%}1{%- endif -%}
|
| 104 |
{%- endmacro -%}
|
| 105 |
-
{%- macro has_dup_tool_result_id(lo, hi, target) -%}
|
| 106 |
-
{%- set ns_cnt = namespace(n=0) -%}
|
| 107 |
-
{%- for k in range(lo, hi + 1) -%}
|
| 108 |
-
{%- set m = messages[k] -%}
|
| 109 |
-
{%- if is_list_of_outputs(m) -%}
|
| 110 |
-
{%- for entry in m.content -%}
|
| 111 |
-
{%- if id_of(entry) == target -%}
|
| 112 |
-
{%- set ns_cnt.n = ns_cnt.n + 1 -%}
|
| 113 |
-
{%- endif -%}
|
| 114 |
-
{%- endfor -%}
|
| 115 |
-
{%- elif id_of(m) == target -%}
|
| 116 |
-
{%- set ns_cnt.n = ns_cnt.n + 1 -%}
|
| 117 |
-
{%- endif -%}
|
| 118 |
-
{%- if ns_cnt.n > 1 -%}{%- break -%}{%- endif -%}
|
| 119 |
-
{%- endfor -%}
|
| 120 |
-
{%- if ns_cnt.n > 1 -%}1{%- endif -%}
|
| 121 |
-
{%- endmacro -%}
|
| 122 |
-
{%- macro tc_id_exists(tcs, target) -%}
|
| 123 |
-
{%- set ns_f = namespace(found=false) -%}
|
| 124 |
-
{%- for tc in tcs -%}
|
| 125 |
-
{%- if id_of(tc) == target -%}
|
| 126 |
-
{%- set ns_f.found = true -%}
|
| 127 |
-
{%- break -%}
|
| 128 |
-
{%- endif -%}
|
| 129 |
-
{%- endfor -%}
|
| 130 |
-
{%- if ns_f.found -%}1{%- endif -%}
|
| 131 |
-
{%- endmacro -%}
|
| 132 |
{%- set ns = namespace(last_user_index=-1) -%}
|
| 133 |
{%- for m in messages %}
|
| 134 |
{%- if m.role == 'user' %}
|
|
@@ -146,7 +112,7 @@ For each function call, output the function name and arguments within the follow
|
|
| 146 |
{%- set reasoning_content = content.split('</think>')[0].split('<think>')[-1] %}
|
| 147 |
{%- set content = content.split('</think>')[-1] %}
|
| 148 |
{%- endif %}
|
| 149 |
-
{%- if (not clear_thinking or loop.index0 > ns.last_user_index) and reasoning_content is defined -%}
|
| 150 |
{{ '<think>' + reasoning_content + '</think>'}}
|
| 151 |
{%- else -%}
|
| 152 |
{{ '<think></think>' }}
|
|
@@ -189,34 +155,21 @@ For each function call, output the function name and arguments within the follow
|
|
| 189 |
{%- set eid = id_of(entry) -%}
|
| 190 |
{%- if not eid -%}
|
| 191 |
{%- set ns_chk.can_sort = false -%}
|
| 192 |
-
{%- elif has_dup_tool_result_id(block_start, ns_blk.end, eid) -%}
|
| 193 |
-
{%- set ns_chk.can_sort = false -%}
|
| 194 |
-
{%- elif not tc_id_exists(ns_a.tool_calls, eid) -%}
|
| 195 |
-
{%- set ns_chk.can_sort = false -%}
|
| 196 |
{%- endif -%}
|
| 197 |
{%- endfor -%}
|
| 198 |
{%- else -%}
|
| 199 |
{%- set tk_id = id_of(m) -%}
|
| 200 |
{%- if not tk_id -%}
|
| 201 |
{%- set ns_chk.can_sort = false -%}
|
| 202 |
-
{%- elif has_dup_tool_result_id(block_start, ns_blk.end, tk_id) -%}
|
| 203 |
-
{%- set ns_chk.can_sort = false -%}
|
| 204 |
-
{%- elif not tc_id_exists(ns_a.tool_calls, tk_id) -%}
|
| 205 |
-
{%- set ns_chk.can_sort = false -%}
|
| 206 |
{%- endif -%}
|
| 207 |
{%- endif -%}
|
| 208 |
{%- endfor -%}
|
| 209 |
-
{%- for
|
| 210 |
-
{%- set tc_id = id_of(
|
| 211 |
{%- if not tc_id -%}
|
| 212 |
{%- set ns_chk.can_sort = false -%}
|
| 213 |
{%- endif -%}
|
| 214 |
-
|
| 215 |
-
{%- if id_of(ns_a.tool_calls[j]) == tc_id -%}
|
| 216 |
-
{%- set ns_chk.can_sort = false -%}
|
| 217 |
-
{%- endif -%}
|
| 218 |
-
{%- endfor -%}
|
| 219 |
-
{%- endfor -%}
|
| 220 |
{%- endif -%}
|
| 221 |
{%- if ns_chk.can_sort -%}
|
| 222 |
{%- for tc in ns_a.tool_calls -%}
|
|
@@ -233,8 +186,8 @@ For each function call, output the function name and arguments within the follow
|
|
| 233 |
{{- tool_response(visible_text(entry.output)) -}}
|
| 234 |
{%- endif -%}
|
| 235 |
{%- endif -%}
|
| 236 |
-
|
| 237 |
-
{%- else -%}
|
| 238 |
{%- set tk_id = id_of(m) -%}
|
| 239 |
{%- if tk_id == tc_id -%}
|
| 240 |
{{- render_tool_response(m) -}}
|
|
|
|
| 1 |
[gMASK]<sop>
|
| 2 |
{%- set effective_reasoning_effort = reasoning_effort if reasoning_effort is defined and reasoning_effort in ['low', 'high'] else 'max' -%}
|
| 3 |
{%- if effective_reasoning_effort is not none -%}<|system|>Reasoning Effort: {{ effective_reasoning_effort | capitalize }}{%- endif -%}
|
|
|
|
| 4 |
{%- if tools -%}
|
| 5 |
{%- macro tool_to_json(tool) -%}
|
| 6 |
{%- set ns_tool = namespace(first=true) -%}
|
|
|
|
| 47 |
|
| 48 |
For each function call, output the function name and arguments within the following XML format:
|
| 49 |
<tool_call>{function-name}<arg_key>{arg-key-1}</arg_key><arg_value>{arg-value-1}</arg_value><arg_key>{arg-key-2}</arg_key><arg_value>{arg-value-2}</arg_value>...</tool_call>{%- endif -%}
|
|
|
|
|
|
|
|
|
|
| 50 |
{%- macro visible_text(content) -%}
|
| 51 |
{%- if content is string -%}
|
| 52 |
+
{{- content }}
|
| 53 |
{%- elif content is iterable and content is not mapping -%}
|
| 54 |
{%- for item in content -%}
|
| 55 |
{%- if item is mapping and item.type == 'text' -%}
|
| 56 |
+
{{- item.text }}
|
| 57 |
{%- elif item is string -%}
|
| 58 |
+
{{- item }}
|
| 59 |
+
{%- elif item is mapping and item.type in ['image', 'image_url', 'video', 'video_url', 'audio', 'audio_url', 'input_audio'] -%}
|
| 60 |
+
{%- set media_type = item.type | replace('_url', '') | replace('input_', '') -%}
|
| 61 |
+
{{- "<reminder>You are unable to process this " ~ media_type ~ " because you don't have multi-modal input ability. Try different methods.</reminder>" }}
|
|
|
|
|
|
|
|
|
|
| 62 |
{%- endif -%}
|
| 63 |
{%- endfor -%}
|
| 64 |
{%- else -%}
|
|
|
|
| 95 |
{%- macro is_list_of_outputs(m) -%}
|
| 96 |
{%- if m.content and m.content.0.output is defined -%}1{%- endif -%}
|
| 97 |
{%- endmacro -%}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 98 |
{%- set ns = namespace(last_user_index=-1) -%}
|
| 99 |
{%- for m in messages %}
|
| 100 |
{%- if m.role == 'user' %}
|
|
|
|
| 112 |
{%- set reasoning_content = content.split('</think>')[0].split('<think>')[-1] %}
|
| 113 |
{%- set content = content.split('</think>')[-1] %}
|
| 114 |
{%- endif %}
|
| 115 |
+
{%- if ((clear_thinking is defined and not clear_thinking) or loop.index0 > ns.last_user_index) and reasoning_content is defined -%}
|
| 116 |
{{ '<think>' + reasoning_content + '</think>'}}
|
| 117 |
{%- else -%}
|
| 118 |
{{ '<think></think>' }}
|
|
|
|
| 155 |
{%- set eid = id_of(entry) -%}
|
| 156 |
{%- if not eid -%}
|
| 157 |
{%- set ns_chk.can_sort = false -%}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 158 |
{%- endif -%}
|
| 159 |
{%- endfor -%}
|
| 160 |
{%- else -%}
|
| 161 |
{%- set tk_id = id_of(m) -%}
|
| 162 |
{%- if not tk_id -%}
|
| 163 |
{%- set ns_chk.can_sort = false -%}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 164 |
{%- endif -%}
|
| 165 |
{%- endif -%}
|
| 166 |
{%- endfor -%}
|
| 167 |
+
{%- for tc in ns_a.tool_calls -%}
|
| 168 |
+
{%- set tc_id = id_of(tc) -%}
|
| 169 |
{%- if not tc_id -%}
|
| 170 |
{%- set ns_chk.can_sort = false -%}
|
| 171 |
{%- endif -%}
|
| 172 |
+
{%- endfor -%}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 173 |
{%- endif -%}
|
| 174 |
{%- if ns_chk.can_sort -%}
|
| 175 |
{%- for tc in ns_a.tool_calls -%}
|
|
|
|
| 186 |
{{- tool_response(visible_text(entry.output)) -}}
|
| 187 |
{%- endif -%}
|
| 188 |
{%- endif -%}
|
| 189 |
+
{%- endfor -%}
|
| 190 |
+
{%- else -%}
|
| 191 |
{%- set tk_id = id_of(m) -%}
|
| 192 |
{%- if tk_id == tc_id -%}
|
| 193 |
{{- render_tool_response(m) -}}
|