8000 Update README.md · ggml-org/llama.cpp@df25e6b · GitHub
[go: up one dir, main page]
Skip to content

Commit df25e6b

Browse files
author
ochafik
committed
Update README.md
1 parent b457f89 commit df25e6b

File tree

2 files changed

+3
-4
lines changed

2 files changed

+3
-4
lines changed

‎common/arg.cpp

Lines changed: 2 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -2848,9 +2848,8 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
28482848
).set_examples({LLAMA_EXAMPLE_SERVER, LLAMA_EXAMPLE_MAIN}).set_env("LLAMA_ARG_JINJA"));
28492849
add_opt(common_arg(
28502850
{"--reasoning-format"}, "FORMAT",
2851-
"reasoning format (default: deepseek; allowed values: deepseek, none)\n"
2852-
"controls whether thought tags are extracted from the response, and in which format they're returned. 'none' leaves thoughts unparsed in `message.content`, 'deepseek' puts them in `message.reasoning_content` (for DeepSeek R1 & Command R7B only).\n"
2853-
"only supported for non-streamed responses",
2851+
"reasoning format (default: deepseek; allowed values: deepseek, none, disabled)\n"
2852+
"controls whether thought tags are allowed and/or extracted from the response, and in which format they're returned. 'none' leaves thoughts unparsed in `message.content`, 'deepseek' puts them in `message.reasoning_content` (for DeepSeek R1 & Command R7B only), 'disabled' prevents generation of thoughts (by closing any thoughts tag or setting template-specific variables such as `enable_thinking: false` for Qwen3).",
28542853
[](common_params & params, const std::string & value) {
28552854
/**/ if (value == "deepseek") { params.reasoning_format = COMMON_REASONING_FORMAT_DEEPSEEK; }
28562855
else if (value == "none") { params.reasoning_format = COMMON_REASONING_FORMAT_NONE; }

‎tools/server/README.md

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -173,7 +173,7 @@ The project is under active development, and we are [looking for feedback and co
173173
| `--no-slots` | disables slots monitoring endpoint<br/>(env: LLAMA_ARG_NO_ENDPOINT_SLOTS) |
174174
| `--slot-save-path PATH` | path to save slot kv cache (default: disabled) |
175175
| `--jinja` | use jinja template for chat (default: disabled)<br/>(env: LLAMA_ARG_JINJA) |
176-
| `--reasoning-format FORMAT` | reasoning format (default: deepseek; allowed values: deepseek, none)<br/>controls whether thought tags are extracted from the response, and in which format they're returned. 'none' leaves thoughts unparsed in `message.content`, 'deepseek' puts them in `message.reasoning_content` (for DeepSeek R1 & Command R7B only).<br/>only supported for non-streamed responses<br/>(env: LLAMA_ARG_THINK) |
176+
| `--reasoning-format FORMAT` | reasoning format (default: deepseek; allowed values: deepseek, none, disabled)<br/>controls whether thought tags are allowed and/or extracted from the response, and in which format they're returned. 'none' leaves thoughts unparsed in `message.content`, 'deepseek' puts them in `message.reasoning_content` (for DeepSeek R1 & Command R7B only), 'disabled' prevents generation of thoughts (by closing any thoughts tag or setting template-specific variables such as `enable_thinking: false` for Qwen3).<br/>(env: LLAMA_ARG_THINK) |
177177
| `--chat-template JINJA_TEMPLATE` | set custom jinja chat template (default: template taken from model's metadata)<br/>if suffix/prefix are specified, template will be disabled<br/>only commonly used templates are accepted (unless --jinja is set before this flag):<br/>list of built-in templates:<br/>bailing, chatglm3, chatglm4, chatml, command-r, deepseek, deepseek2, deepseek3, exaone3, falcon3, gemma, gigachat, glmedge, granite, llama2, llama2-sys, llama2-sys-bos, llama2-sys-strip, llama3, llama4, megrez, minicpm, mistral-v1, mistral-v3, mistral-v3-tekken, mistral-v7, mistral-v7-tekken, monarch, openchat, orion, phi3, phi4, rwkv-world, smolvlm, vicuna, vicuna-orca, yandex, zephyr<br/>(env: LLAMA_ARG_CHAT_TEMPLATE) |
178178
| `--chat-template-file JINJA_TEMPLATE_FILE` | set custom jinja chat template file (default: template taken from model's metadata)<br/>if suffix/prefix are specified, template will be disabled<br/>only commonly used templates are accepted (unless --jinja is set before this flag):<br/>list of built-in templates:<br/>bailing, chatglm3, chatglm4, chatml, command-r, deepseek, deepseek2, deepseek3, exaone3, falcon3, gemma, gigachat, glmedge, granite, llama2, llama2-sys, llama2-sys-bos, llama2-sys-strip, llama3, llama4, megrez, minicpm, mistral-v1, mistral-v3, mistral-v3-tekken, mistral-v7, mistral-v7-tekken, monarch, openchat, orion, phi3, phi4, rwkv-world, smolvlm, vicuna, vicuna-orca, yandex, zephyr<br/>(env: LLAMA_ARG_CHAT_TEMPLATE_FILE) |
179179
| `--no-prefill-assistant` | whether to prefill the assistant's response if the last message is an assistant message (default: prefill enabled)<br/>when this flag is set, if the last message is an assistant message then it will be treated as a full message and not prefilled<br/>(env: LLAMA_ARG_NO_PREFILL_ASSISTANT) |

0 commit comments

Comments
 (0)
0