Reasoning Parser#
SGLang supports parsing reasoning content our from “normal” content for reasoning models such as DeepSeek R1.
Supported Models & Parsers#
Model |
Reasoning tags |
Parser |
|---|---|---|
|
|
|
|
|
Usage#
Launching the Server#
Specify the --reasoning-parser option.
[1]:
import requests
from openai import OpenAI
from sglang.test.test_utils import is_in_ci
if is_in_ci():
from patch import launch_server_cmd
else:
from sglang.utils import launch_server_cmd
from sglang.utils import wait_for_server, print_highlight, terminate_process
server_process, port = launch_server_cmd(
"python3 -m sglang.launch_server --model-path deepseek-ai/DeepSeek-R1-Distill-Qwen-7B --host 0.0.0.0 --reasoning-parser deepseek-r1"
)
wait_for_server(f"http://localhost:{port}")
[2025-05-19 21:25:19] server_args=ServerArgs(model_path='deepseek-ai/DeepSeek-R1-Distill-Qwen-7B', tokenizer_path='deepseek-ai/DeepSeek-R1-Distill-Qwen-7B', tokenizer_mode='auto', skip_tokenizer_init=False, load_format='auto', trust_remote_code=False, dtype='auto', kv_cache_dtype='auto', quantization=None, quantization_param_path=None, context_length=None, device='cuda', served_model_name='deepseek-ai/DeepSeek-R1-Distill-Qwen-7B', chat_template=None, completion_template=None, is_embedding=False, enable_multimodal=None, revision=None, host='0.0.0.0', port=32335, mem_fraction_static=0.88, max_running_requests=200, max_total_tokens=20480, chunked_prefill_size=8192, max_prefill_tokens=16384, schedule_policy='fcfs', schedule_conservativeness=1.0, cpu_offload_gb=0, page_size=1, tp_size=1, pp_size=1, max_micro_batch_size=None, stream_interval=1, stream_output=False, random_seed=684384841, constrained_json_whitespace_pattern=None, watchdog_timeout=300, dist_timeout=None, download_dir=None, base_gpu_id=0, gpu_id_step=1, log_level='info', log_level_http=None, log_requests=False, log_requests_level=0, show_time_cost=False, enable_metrics=False, bucket_time_to_first_token=None, bucket_e2e_request_latency=None, bucket_inter_token_latency=None, collect_tokens_histogram=False, decode_log_interval=40, enable_request_time_stats_logging=False, kv_events_config=None, api_key=None, file_storage_path='sglang_storage', enable_cache_report=False, reasoning_parser='deepseek-r1', dp_size=1, load_balance_method='round_robin', ep_size=1, dist_init_addr=None, nnodes=1, node_rank=0, json_model_override_args='{}', preferred_sampling_params=None, lora_paths=None, max_loras_per_batch=8, lora_backend='triton', attention_backend=None, sampling_backend='flashinfer', grammar_backend='xgrammar', speculative_algorithm=None, speculative_draft_model_path=None, speculative_num_steps=None, speculative_eagle_topk=None, speculative_num_draft_tokens=None, speculative_accept_threshold_single=1.0, speculative_accept_threshold_acc=1.0, speculative_token_map=None, enable_double_sparsity=False, ds_channel_config_path=None, ds_heavy_channel_num=32, ds_heavy_token_num=256, ds_heavy_channel_type='qk', ds_sparse_decode_threshold=4096, disable_radix_cache=False, disable_cuda_graph=True, disable_cuda_graph_padding=False, enable_nccl_nvls=False, enable_tokenizer_batch_encode=False, disable_outlines_disk_cache=False, disable_custom_all_reduce=False, disable_overlap_schedule=False, enable_mixed_chunk=False, enable_dp_attention=False, enable_dp_lm_head=False, enable_ep_moe=False, enable_deepep_moe=False, deepep_mode='auto', deepep_config=None, enable_torch_compile=False, torch_compile_max_bs=32, cuda_graph_max_bs=None, cuda_graph_bs=None, torchao_config='', enable_nan_detection=False, enable_p2p_check=False, triton_attention_reduce_in_fp32=False, triton_attention_num_kv_splits=8, num_continuous_decode_steps=1, delete_ckpt_after_loading=False, enable_memory_saver=False, allow_auto_truncate=False, enable_custom_logit_processor=False, tool_call_parser=None, enable_hierarchical_cache=False, hicache_ratio=2.0, hicache_size=0, hicache_write_policy='write_through_selective', flashinfer_mla_disable_ragged=False, warmups=None, moe_dense_tp_size=None, n_share_experts_fusion=0, disable_chunked_prefix_cache=False, disable_fast_image_processor=False, mm_attention_backend=None, debug_tensor_dump_output_folder=None, debug_tensor_dump_input_file=None, debug_tensor_dump_inject=False, disaggregation_mode='null', disaggregation_bootstrap_port=8998, disaggregation_transfer_backend='mooncake', disaggregation_ib_device=None, pdlb_url=None)
[2025-05-19 21:25:32] Attention backend not set. Use fa3 backend by default.
[2025-05-19 21:25:32] Init torch distributed begin.
[2025-05-19 21:25:32] Init torch distributed ends. mem usage=0.00 GB
[2025-05-19 21:25:32] Load weight begin. avail mem=78.58 GB
[2025-05-19 21:25:33] Using model weights format ['*.safetensors']
Loading safetensors checkpoint shards: 0% Completed | 0/2 [00:00<?, ?it/s]
Loading safetensors checkpoint shards: 50% Completed | 1/2 [00:01<00:01, 1.36s/it]
Loading safetensors checkpoint shards: 100% Completed | 2/2 [00:02<00:00, 1.27s/it]
Loading safetensors checkpoint shards: 100% Completed | 2/2 [00:02<00:00, 1.28s/it]
[2025-05-19 21:25:39] Load weight end. type=Qwen2ForCausalLM, dtype=torch.bfloat16, avail mem=51.88 GB, mem usage=26.70 GB.
[2025-05-19 21:25:39] KV Cache is allocated. #tokens: 20480, K size: 0.55 GB, V size: 0.55 GB
[2025-05-19 21:25:39] Memory pool end. avail mem=49.61 GB
[2025-05-19 21:25:39] max_total_num_tokens=20480, chunked_prefill_size=8192, max_prefill_tokens=16384, max_running_requests=200, context_len=131072
[2025-05-19 21:25:40] INFO: Started server process [957010]
[2025-05-19 21:25:40] INFO: Waiting for application startup.
[2025-05-19 21:25:40] INFO: Application startup complete.
[2025-05-19 21:25:40] INFO: Uvicorn running on http://0.0.0.0:32335 (Press CTRL+C to quit)
[2025-05-19 21:25:41] INFO: 127.0.0.1:50508 - "GET /v1/models HTTP/1.1" 200 OK
[2025-05-19 21:25:41] INFO: 127.0.0.1:50514 - "GET /get_model_info HTTP/1.1" 200 OK
[2025-05-19 21:25:41] Prefill batch. #new-seq: 1, #new-token: 7, #cached-token: 0, token usage: 0.00, #running-req: 0, #queue-req: 0
[2025-05-19 21:25:43] INFO: 127.0.0.1:50528 - "POST /generate HTTP/1.1" 200 OK
[2025-05-19 21:25:43] The server is fired up and ready to roll!
NOTE: Typically, the server runs in a separate terminal.
In this notebook, we run the server and notebook code together, so their outputs are combined.
To improve clarity, the server logs are displayed in the original black color, while the notebook outputs are highlighted in blue.
We are running those notebooks in a CI parallel environment, so the throughput is not representative of the actual performance.
Note that --reasoning-parser defines the parser used to interpret responses.
OpenAI Compatible API#
Using the OpenAI compatible API, the contract follows the DeepSeek API design established with the release of DeepSeek-R1:
reasoning_content: The content of the CoT.content: The content of the final answer.
[2]:
# Initialize OpenAI-like client
client = OpenAI(api_key="None", base_url=f"http://0.0.0.0:{port}/v1")
model_name = client.models.list().data[0].id
messages = [
{
"role": "user",
"content": "What is 1+3?",
}
]
[2025-05-19 21:25:46] INFO: 127.0.0.1:46128 - "GET /v1/models HTTP/1.1" 200 OK
Non-Streaming Request#
[3]:
response_non_stream = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.6,
top_p=0.95,
stream=False, # Non-streaming
extra_body={"separate_reasoning": True},
)
print_highlight("==== Reasoning ====")
print_highlight(response_non_stream.choices[0].message.reasoning_content)
print_highlight("==== Text ====")
print_highlight(response_non_stream.choices[0].message.content)
[2025-05-19 21:25:46] Prefill batch. #new-seq: 1, #new-token: 11, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0
[2025-05-19 21:25:46] Decode batch. #running-req: 1, #token: 45, token usage: 0.00, cuda graph: False, gen throughput (token/s): 6.06, #queue-req: 0
[2025-05-19 21:25:46] Decode batch. #running-req: 1, #token: 85, token usage: 0.00, cuda graph: False, gen throughput (token/s): 111.12, #queue-req: 0
[2025-05-19 21:25:47] INFO: 127.0.0.1:46128 - "POST /v1/chat/completions HTTP/1.1" 200 OK
==== Reasoning ====
First, I recognize that the problem is asking for the sum of the numbers 1 and 3.
Next, I perform the addition: 1 plus 3 equals 4.
Therefore, the final answer is 4.
Next, I perform the addition: 1 plus 3 equals 4.
Therefore, the final answer is 4.
==== Text ====
**Solution:**
We are asked to find the sum of the numbers 1 and 3.
\[
1 + 3 = 4
\]
Therefore, the final answer is \(\boxed{4}\).
We are asked to find the sum of the numbers 1 and 3.
\[
1 + 3 = 4
\]
Therefore, the final answer is \(\boxed{4}\).
Streaming Request#
[4]:
response_stream = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.6,
top_p=0.95,
stream=True, # Non-streaming
extra_body={"separate_reasoning": True},
)
reasoning_content = ""
content = ""
for chunk in response_stream:
if chunk.choices[0].delta.content:
content += chunk.choices[0].delta.content
if chunk.choices[0].delta.reasoning_content:
reasoning_content += chunk.choices[0].delta.reasoning_content
print_highlight("==== Reasoning ====")
print_highlight(reasoning_content)
print_highlight("==== Text ====")
print_highlight(content)
[2025-05-19 21:25:47] INFO: 127.0.0.1:46128 - "POST /v1/chat/completions HTTP/1.1" 200 OK
[2025-05-19 21:25:47] Prefill batch. #new-seq: 1, #new-token: 1, #cached-token: 11, token usage: 0.00, #running-req: 0, #queue-req: 0
[2025-05-19 21:25:47] Decode batch. #running-req: 1, #token: 31, token usage: 0.00, cuda graph: False, gen throughput (token/s): 100.47, #queue-req: 0
[2025-05-19 21:25:47] Decode batch. #running-req: 1, #token: 71, token usage: 0.00, cuda graph: False, gen throughput (token/s): 106.32, #queue-req: 0
[2025-05-19 21:25:48] Decode batch. #running-req: 1, #token: 111, token usage: 0.01, cuda graph: False, gen throughput (token/s): 85.47, #queue-req: 0
[2025-05-19 21:25:48] Decode batch. #running-req: 1, #token: 1, token usage: 0.00, cuda graph: False, gen throughput (token/s): 102.98, #queue-req: 0
==== Reasoning ====
First, I recognize that the problem is asking for the sum of 1 and 3.
Next, I perform the addition of these two numbers.
Finally, I arrive at the result, which is 4.
Next, I perform the addition of these two numbers.
Finally, I arrive at the result, which is 4.
==== Text ====
Certainly! Let's solve the problem step by step.
**Question:** What is \(1 + 3\)?
**Solution:**
1. **Identify the numbers to add:**
\[
1 \quad \text{and} \quad 3
\]
2. **Perform the addition:**
\[
1 + 3 = 4
\]
**Answer:** \(\boxed{4}\)
Optionally, you can buffer the reasoning content to the last reasoning chunk (or the first chunk after the reasoning content).
[5]:
response_stream = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.6,
top_p=0.95,
stream=True, # Non-streaming
extra_body={"separate_reasoning": True, "stream_reasoning": False},
)
reasoning_content = ""
content = ""
for chunk in response_stream:
if chunk.choices[0].delta.content:
content += chunk.choices[0].delta.content
if chunk.choices[0].delta.reasoning_content:
reasoning_content = chunk.choices[0].delta.reasoning_content
print_highlight("==== Reasoning ====")
print_highlight(reasoning_content)
print_highlight("==== Text ====")
print_highlight(content)
[2025-05-19 21:25:48] INFO: 127.0.0.1:46128 - "POST /v1/chat/completions HTTP/1.1" 200 OK
[2025-05-19 21:25:48] Prefill batch. #new-seq: 1, #new-token: 1, #cached-token: 11, token usage: 0.00, #running-req: 0, #queue-req: 0
[2025-05-19 21:25:48] Decode batch. #running-req: 1, #token: 52, token usage: 0.00, cuda graph: False, gen throughput (token/s): 90.53, #queue-req: 0
[2025-05-19 21:25:49] Decode batch. #running-req: 1, #token: 92, token usage: 0.00, cuda graph: False, gen throughput (token/s): 110.58, #queue-req: 0
[2025-05-19 21:25:49] Decode batch. #running-req: 1, #token: 132, token usage: 0.01, cuda graph: False, gen throughput (token/s): 110.94, #queue-req: 0
==== Reasoning ====
First, I identify the numbers to be added, which are 1 and 3.
Next, I perform the addition of these two numbers.
Finally, I arrive at the sum, which is 4.
Next, I perform the addition of these two numbers.
Finally, I arrive at the sum, which is 4.
==== Text ====
**Solution:**
To find the sum of 1 and 3, follow these simple steps:
1. **Identify the numbers to add:**
\[
1 \quad \text{and} \quad 3
\]
2. **Perform the addition:**
\[
1 + 3 = 4
\]
3. **Present the final answer:**
\[
\boxed{4}
\]
**Answer:** \(\boxed{4}\)
The reasoning separation is enable by default when specify . To disable it, set the ``separate_reasoning`` option to ``False`` in request.
[6]:
response_non_stream = client.chat.completions.create(
model=model_name,
messages=messages,
temperature=0.6,
top_p=0.95,
stream=False, # Non-streaming
extra_body={"separate_reasoning": False},
)
print_highlight("==== Original Output ====")
print_highlight(response_non_stream.choices[0].message.content)
[2025-05-19 21:25:50] Prefill batch. #new-seq: 1, #new-token: 1, #cached-token: 11, token usage: 0.00, #running-req: 0, #queue-req: 0
[2025-05-19 21:25:50] Decode batch. #running-req: 1, #token: 20, token usage: 0.00, cuda graph: False, gen throughput (token/s): 90.45, #queue-req: 0
[2025-05-19 21:25:50] Decode batch. #running-req: 1, #token: 60, token usage: 0.00, cuda graph: False, gen throughput (token/s): 107.54, #queue-req: 0
[2025-05-19 21:25:50] Decode batch. #running-req: 1, #token: 100, token usage: 0.00, cuda graph: False, gen throughput (token/s): 111.51, #queue-req: 0
[2025-05-19 21:25:51] INFO: 127.0.0.1:46128 - "POST /v1/chat/completions HTTP/1.1" 200 OK
==== Original Output ====
I need to calculate the sum of 1 and 3.
First, I'll identify the two numbers involved in the addition: 1 and 3.
Next, I'll add these two numbers together.
Finally, the result of adding 1 and 3 is 4.
**Solution:**
We are asked to find the sum of \(1\) and \(3\).
\[
1 + 3 = 4
\]
**Answer:** \(\boxed{4}\)
First, I'll identify the two numbers involved in the addition: 1 and 3.
Next, I'll add these two numbers together.
Finally, the result of adding 1 and 3 is 4.
**Solution:**
We are asked to find the sum of \(1\) and \(3\).
\[
1 + 3 = 4
\]
**Answer:** \(\boxed{4}\)
SGLang Native API#
[7]:
from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-R1-Distill-Qwen-7B")
input = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
)
gen_url = f"http://localhost:{port}/generate"
gen_data = {
"text": input,
"sampling_params": {
"skip_special_tokens": False,
"max_new_tokens": 1024,
"temperature": 0.6,
"top_p": 0.95,
},
}
gen_response = requests.post(gen_url, json=gen_data).json()["text"]
print_highlight("==== Original Output ====")
print_highlight(gen_response)
parse_url = f"http://localhost:{port}/separate_reasoning"
separate_reasoning_data = {
"text": gen_response,
"reasoning_parser": "deepseek-r1",
}
separate_reasoning_response_json = requests.post(
parse_url, json=separate_reasoning_data
).json()
print_highlight("==== Reasoning ====")
print_highlight(separate_reasoning_response_json["reasoning_text"])
print_highlight("==== Text ====")
print_highlight(separate_reasoning_response_json["text"])
[2025-05-19 21:25:55] Prefill batch. #new-seq: 1, #new-token: 12, #cached-token: 1, token usage: 0.00, #running-req: 0, #queue-req: 0
[2025-05-19 21:25:55] Decode batch. #running-req: 1, #token: 38, token usage: 0.00, cuda graph: False, gen throughput (token/s): 9.07, #queue-req: 0
[2025-05-19 21:25:55] Decode batch. #running-req: 1, #token: 78, token usage: 0.00, cuda graph: False, gen throughput (token/s): 113.12, #queue-req: 0
[2025-05-19 21:25:55] Decode batch. #running-req: 1, #token: 118, token usage: 0.01, cuda graph: False, gen throughput (token/s): 109.11, #queue-req: 0
[2025-05-19 21:25:56] Decode batch. #running-req: 1, #token: 158, token usage: 0.01, cuda graph: False, gen throughput (token/s): 109.57, #queue-req: 0
[2025-05-19 21:25:56] INFO: 127.0.0.1:56216 - "POST /generate HTTP/1.1" 200 OK
==== Original Output ====
I need to add the numbers 1 and 3 together.
First, I'll identify the two numbers involved in the addition.
Next, I'll perform the addition operation to find the sum.
Finally, I'll state the result clearly.
Sure! Let's solve the addition step by step.
**Problem:**
What is \(1 + 3\)?
**Solution:**
To find the sum of 1 and 3, follow these simple steps:
1. **Start with the first number:**
\[
1
\]
2. **Add the second number:**
\[
1 + 3
\]
3. **Calculate the sum:**
\[
1 + 3 = 4
\]
**Answer:**
\[
\boxed{4}
\]
First, I'll identify the two numbers involved in the addition.
Next, I'll perform the addition operation to find the sum.
Finally, I'll state the result clearly.
Sure! Let's solve the addition step by step.
**Problem:**
What is \(1 + 3\)?
**Solution:**
To find the sum of 1 and 3, follow these simple steps:
1. **Start with the first number:**
\[
1
\]
2. **Add the second number:**
\[
1 + 3
\]
3. **Calculate the sum:**
\[
1 + 3 = 4
\]
**Answer:**
\[
\boxed{4}
\]
[2025-05-19 21:25:56] INFO: 127.0.0.1:56222 - "POST /separate_reasoning HTTP/1.1" 200 OK
==== Reasoning ====
I need to add the numbers 1 and 3 together.
First, I'll identify the two numbers involved in the addition.
Next, I'll perform the addition operation to find the sum.
Finally, I'll state the result clearly.
First, I'll identify the two numbers involved in the addition.
Next, I'll perform the addition operation to find the sum.
Finally, I'll state the result clearly.
==== Text ====
Sure! Let's solve the addition step by step.
**Problem:**
What is \(1 + 3\)?
**Solution:**
To find the sum of 1 and 3, follow these simple steps:
1. **Start with the first number:**
\[
1
\]
2. **Add the second number:**
\[
1 + 3
\]
3. **Calculate the sum:**
\[
1 + 3 = 4
\]
**Answer:**
\[
\boxed{4}
\]
**Problem:**
What is \(1 + 3\)?
**Solution:**
To find the sum of 1 and 3, follow these simple steps:
1. **Start with the first number:**
\[
1
\]
2. **Add the second number:**
\[
1 + 3
\]
3. **Calculate the sum:**
\[
1 + 3 = 4
\]
**Answer:**
\[
\boxed{4}
\]
[8]:
terminate_process(server_process)
[2025-05-19 21:25:56] Child process unexpectedly failed with an exit code 9. pid=957579
[2025-05-19 21:25:56] Child process unexpectedly failed with an exit code 9. pid=957449
Offline Engine API#
[9]:
import sglang as sgl
from sglang.srt.reasoning_parser import ReasoningParser
from sglang.utils import print_highlight
llm = sgl.Engine(model_path="deepseek-ai/DeepSeek-R1-Distill-Qwen-7B")
tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-R1-Distill-Qwen-7B")
input = tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True,
)
sampling_params = {
"max_new_tokens": 1024,
"skip_special_tokens": False,
"temperature": 0.6,
"top_p": 0.95,
}
result = llm.generate(prompt=input, sampling_params=sampling_params)
generated_text = result["text"] # Assume there is only one prompt
print_highlight("==== Original Output ====")
print_highlight(generated_text)
parser = ReasoningParser("deepseek-r1")
reasoning_text, text = parser.parse_non_stream(generated_text)
print_highlight("==== Reasoning ====")
print_highlight(reasoning_text)
print_highlight("==== Text ====")
print_highlight(text)
Loading safetensors checkpoint shards: 0% Completed | 0/2 [00:00<?, ?it/s]
Loading safetensors checkpoint shards: 50% Completed | 1/2 [00:01<00:01, 1.43s/it]
Loading safetensors checkpoint shards: 100% Completed | 2/2 [00:02<00:00, 1.29s/it]
Loading safetensors checkpoint shards: 100% Completed | 2/2 [00:02<00:00, 1.31s/it]
==== Original Output ====
First, I recognize that the problem is asking for the sum of the numbers 1 and 3.
Next, I add the two numbers together.
Finally, I arrive at the result of 4.
**Solution:**
We are asked to find the sum of \(1\) and \(3\).
\[
1 + 3 = \boxed{4}
\]
**Step-by-Step Explanation:**
1. **Identify the numbers to add:** We have the numbers \(1\) and \(3\).
2. **Perform the addition:**
\[
1 + 3 = 4
\]
3. **Present the final answer:** The sum of \(1\) and \(3\) is \(\boxed{4}\).
\[
\boxed{4}
\]
Next, I add the two numbers together.
Finally, I arrive at the result of 4.
**Solution:**
We are asked to find the sum of \(1\) and \(3\).
\[
1 + 3 = \boxed{4}
\]
**Step-by-Step Explanation:**
1. **Identify the numbers to add:** We have the numbers \(1\) and \(3\).
2. **Perform the addition:**
\[
1 + 3 = 4
\]
3. **Present the final answer:** The sum of \(1\) and \(3\) is \(\boxed{4}\).
\[
\boxed{4}
\]
==== Reasoning ====
First, I recognize that the problem is asking for the sum of the numbers 1 and 3.
Next, I add the two numbers together.
Finally, I arrive at the result of 4.
Next, I add the two numbers together.
Finally, I arrive at the result of 4.
==== Text ====
**Solution:**
We are asked to find the sum of \(1\) and \(3\).
\[
1 + 3 = \boxed{4}
\]
**Step-by-Step Explanation:**
1. **Identify the numbers to add:** We have the numbers \(1\) and \(3\).
2. **Perform the addition:**
\[
1 + 3 = 4
\]
3. **Present the final answer:** The sum of \(1\) and \(3\) is \(\boxed{4}\).
\[
\boxed{4}
\]
We are asked to find the sum of \(1\) and \(3\).
\[
1 + 3 = \boxed{4}
\]
**Step-by-Step Explanation:**
1. **Identify the numbers to add:** We have the numbers \(1\) and \(3\).
2. **Perform the addition:**
\[
1 + 3 = 4
\]
3. **Present the final answer:** The sum of \(1\) and \(3\) is \(\boxed{4}\).
\[
\boxed{4}
\]
[10]:
llm.shutdown()
Supporting New Reasoning Model Schemas#
For future reasoning models, you can implement the reasoning parser as a subclass of BaseReasoningFormatDetector in python/sglang/srt/reasoning_parser.py and specify the reasoning parser for new reasoning model schemas accordingly.
class DeepSeekR1Detector(BaseReasoningFormatDetector):
"""
Detector for DeepSeek-R1 model.
Assumes reasoning format:
(<think>)*(.*)</think>
Returns all the text before the </think> tag as `reasoning_text`
and the rest of the text as `normal_text`.
Args:
stream_reasoning (bool): If False, accumulates reasoning content until the end tag.
If True, streams reasoning content as it arrives.
"""
def __init__(self, stream_reasoning: bool = False):
# DeepSeek-R1 is assumed to be reasoning until `</think>` token
super().__init__("<think>", "</think>", True, stream_reasoning=stream_reasoning)
# https://github.com/sgl-project/sglang/pull/3202#discussion_r1950153599
class ReasoningParser:
"""
Parser that handles both streaming and non-streaming scenarios for extracting
reasoning content from model outputs.
Args:
model_type (str): Type of model to parse reasoning from
stream_reasoning (bool): If False, accumulates reasoning content until complete.
If True, streams reasoning content as it arrives.
"""
DetectorMap: Dict[str, BaseReasoningFormatDetector] = {
"deepseek-r1": DeepSeekR1Detector
}
def __init__(self, model_type: str = None, stream_reasoning: bool = True):
if not model_type:
raise ValueError("Model type must be specified")
detector_class = self.DetectorMap.get(model_type.lower())
if not detector_class:
raise ValueError(f"Unsupported model type: {model_type}")
self.detector = detector_class(stream_reasoning=stream_reasoning)
def parse_non_stream(self, full_text: str) -> StreamingParseResult:
"""Non-streaming call: one-time parsing"""
ret = self.detector.detect_and_parse(full_text)
return ret.reasoning_text, ret.normal_text
def parse_stream_chunk(self, chunk_text: str) -> StreamingParseResult:
"""Streaming call: incremental parsing"""
ret = self.detector.parse_streaming_increment(chunk_text)
return ret.reasoning_text, ret.normal_text