sglang 0.2.10__py3-none-any.whl → 0.2.12__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- sglang/__init__.py +8 -0
- sglang/api.py +10 -2
- sglang/bench_latency.py +151 -40
- sglang/bench_serving.py +46 -22
- sglang/check_env.py +24 -2
- sglang/global_config.py +0 -1
- sglang/lang/backend/base_backend.py +3 -1
- sglang/lang/backend/openai.py +8 -3
- sglang/lang/backend/runtime_endpoint.py +46 -29
- sglang/lang/choices.py +164 -0
- sglang/lang/compiler.py +2 -2
- sglang/lang/interpreter.py +6 -13
- sglang/lang/ir.py +14 -5
- sglang/srt/constrained/base_tool_cache.py +1 -1
- sglang/srt/constrained/fsm_cache.py +12 -2
- sglang/srt/layers/activation.py +33 -0
- sglang/srt/layers/{token_attention.py → decode_attention.py} +9 -5
- sglang/srt/layers/extend_attention.py +6 -1
- sglang/srt/layers/layernorm.py +65 -0
- sglang/srt/layers/logits_processor.py +6 -1
- sglang/srt/layers/pooler.py +50 -0
- sglang/srt/layers/{context_flashattention_nopad.py → prefill_attention.py} +5 -0
- sglang/srt/layers/radix_attention.py +4 -7
- sglang/srt/managers/detokenizer_manager.py +31 -9
- sglang/srt/managers/io_struct.py +63 -0
- sglang/srt/managers/policy_scheduler.py +173 -25
- sglang/srt/managers/schedule_batch.py +174 -380
- sglang/srt/managers/tokenizer_manager.py +197 -112
- sglang/srt/managers/tp_worker.py +299 -364
- sglang/srt/mem_cache/{base_cache.py → base_prefix_cache.py} +9 -4
- sglang/srt/mem_cache/chunk_cache.py +43 -20
- sglang/srt/mem_cache/memory_pool.py +10 -15
- sglang/srt/mem_cache/radix_cache.py +74 -40
- sglang/srt/model_executor/cuda_graph_runner.py +27 -12
- sglang/srt/model_executor/forward_batch_info.py +319 -0
- sglang/srt/model_executor/model_runner.py +30 -47
- sglang/srt/models/chatglm.py +1 -1
- sglang/srt/models/commandr.py +1 -1
- sglang/srt/models/dbrx.py +1 -1
- sglang/srt/models/deepseek.py +1 -1
- sglang/srt/models/deepseek_v2.py +1 -1
- sglang/srt/models/gemma.py +1 -1
- sglang/srt/models/gemma2.py +1 -2
- sglang/srt/models/gpt_bigcode.py +1 -1
- sglang/srt/models/grok.py +1 -1
- sglang/srt/models/internlm2.py +3 -8
- sglang/srt/models/llama2.py +5 -5
- sglang/srt/models/llama_classification.py +1 -1
- sglang/srt/models/llama_embedding.py +88 -0
- sglang/srt/models/llava.py +1 -2
- sglang/srt/models/llavavid.py +1 -2
- sglang/srt/models/minicpm.py +1 -1
- sglang/srt/models/mixtral.py +1 -1
- sglang/srt/models/mixtral_quant.py +1 -1
- sglang/srt/models/qwen.py +1 -1
- sglang/srt/models/qwen2.py +1 -1
- sglang/srt/models/qwen2_moe.py +1 -12
- sglang/srt/models/stablelm.py +1 -1
- sglang/srt/openai_api/adapter.py +189 -39
- sglang/srt/openai_api/protocol.py +43 -1
- sglang/srt/sampling/penaltylib/__init__.py +13 -0
- sglang/srt/sampling/penaltylib/orchestrator.py +357 -0
- sglang/srt/sampling/penaltylib/penalizers/frequency_penalty.py +80 -0
- sglang/srt/sampling/penaltylib/penalizers/min_new_tokens.py +105 -0
- sglang/srt/sampling/penaltylib/penalizers/presence_penalty.py +79 -0
- sglang/srt/sampling/penaltylib/penalizers/repetition_penalty.py +83 -0
- sglang/srt/sampling_params.py +31 -4
- sglang/srt/server.py +93 -21
- sglang/srt/server_args.py +30 -19
- sglang/srt/utils.py +31 -13
- sglang/test/run_eval.py +10 -1
- sglang/test/runners.py +63 -63
- sglang/test/simple_eval_humaneval.py +2 -8
- sglang/test/simple_eval_mgsm.py +203 -0
- sglang/test/srt/sampling/penaltylib/utils.py +337 -0
- sglang/test/test_layernorm.py +60 -0
- sglang/test/test_programs.py +4 -2
- sglang/test/test_utils.py +21 -3
- sglang/utils.py +0 -1
- sglang/version.py +1 -1
- {sglang-0.2.10.dist-info → sglang-0.2.12.dist-info}/METADATA +50 -31
- sglang-0.2.12.dist-info/RECORD +112 -0
- sglang/srt/layers/linear.py +0 -884
- sglang/srt/layers/quantization/__init__.py +0 -64
- sglang/srt/layers/quantization/fp8.py +0 -677
- sglang-0.2.10.dist-info/RECORD +0 -100
- {sglang-0.2.10.dist-info → sglang-0.2.12.dist-info}/LICENSE +0 -0
- {sglang-0.2.10.dist-info → sglang-0.2.12.dist-info}/WHEEL +0 -0
- {sglang-0.2.10.dist-info → sglang-0.2.12.dist-info}/top_level.txt +0 -0
sglang-0.2.10.dist-info/RECORD
DELETED
@@ -1,100 +0,0 @@
|
|
1
|
-
sglang/__init__.py,sha256=ECjvAWlxIwKtUIXGchfkoCIbF-iqLjH-Q0o8xHTlVNY,1352
|
2
|
-
sglang/api.py,sha256=s_P8BvGDCQ0PiqOapr2TLFge1NA7QmKqUx6bFQ8Q5GQ,5676
|
3
|
-
sglang/bench_latency.py,sha256=lHk9C3XM1e-UQd6HY2qn-njr2rG5AFQ_sNVD5hcF5Vc,12162
|
4
|
-
sglang/bench_serving.py,sha256=M0YQT6xElpkx-FtmyUe6lhX1DZfVLGh54qd6qfFYquc,34801
|
5
|
-
sglang/check_env.py,sha256=XlVou81XC20tPFVTuKDSKqDqLQJoO2QvlnReWMf-Ho4,4152
|
6
|
-
sglang/global_config.py,sha256=CyhGL7PE-KlMcg7IHWykzImU1y4NQlpeIlh9lHA77uo,1749
|
7
|
-
sglang/launch_server.py,sha256=Gg8CwNlTCCfg1dF65ZT9ePLxOT9LKtY79GhIPG6PCrU,358
|
8
|
-
sglang/launch_server_llavavid.py,sha256=40uaazMsavKuk6YXFa5v37kdUpFGuealgJJeph1g8gU,1025
|
9
|
-
sglang/utils.py,sha256=C50xm06WWKpKB8kSNs9vO4egJ2QTk_OAA6M13S2cB_A,8369
|
10
|
-
sglang/version.py,sha256=waXgc7p-jgGCsUjdVfO_KjlVZblnCvrzf4A0dsBj_lg,23
|
11
|
-
sglang/lang/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
12
|
-
sglang/lang/chat_template.py,sha256=psIlhaDo70twgLrx5Lgln03metLEA3-FZuixeI0Y7Ao,13309
|
13
|
-
sglang/lang/compiler.py,sha256=UiXUmPR9wBAPtnORrLcyQX8Uh0ZL0nKeV8ZgBozAJPw,7531
|
14
|
-
sglang/lang/interpreter.py,sha256=_MbvYB0vweCgALklpM2DlofiCXuITCmX_fl8rPPcp5U,30340
|
15
|
-
sglang/lang/ir.py,sha256=0r-mhA4aO-uuS97Dvkw99ERTcJXfzuV6jJQMmuCwHEg,16615
|
16
|
-
sglang/lang/tracer.py,sha256=borJmlSJOhg1RUndGRnilnR60eEZz2Y9aU7BpftsOxU,8287
|
17
|
-
sglang/lang/backend/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
18
|
-
sglang/lang/backend/anthropic.py,sha256=EXRX7xJgA5KZszX7toSLVnKzFQ5EO0Loj-YjHFtxSxg,2081
|
19
|
-
sglang/lang/backend/base_backend.py,sha256=APiMht4WYECLCOGRPCEUF6lX-an1vjVe2dWoMSgymWY,1831
|
20
|
-
sglang/lang/backend/litellm.py,sha256=ugmL7sfUxkUHVbHtwNzHgdQAEd4UCjNQboFuE3KThcY,2450
|
21
|
-
sglang/lang/backend/openai.py,sha256=6ww2rwKouWgtmjaCf4hk-kXXJ6bY6n9Xnbm3UTFZvl4,14808
|
22
|
-
sglang/lang/backend/runtime_endpoint.py,sha256=n78pyBWTCMYmDAS-0yZVFvzQYCiACz8Usj7FTDfdVKE,8763
|
23
|
-
sglang/lang/backend/vertexai.py,sha256=O-iBLD-y3vq80UxnrAoJri7bxpgd-_eakZ88Cf8bEGA,4855
|
24
|
-
sglang/srt/conversation.py,sha256=V5YuoeO6-aLqGv0p3J2qx8TnBJbN1oTopYFutNul3GQ,16491
|
25
|
-
sglang/srt/hf_transformers_utils.py,sha256=Tf_RplcW7llVXsigRvSGqmeAUxBeAL8rPCkzuqWfZ8U,11925
|
26
|
-
sglang/srt/mm_utils.py,sha256=n7_GmbOM_0IWVXovpM34rKIBw0Py9yb_NXSQw27u4OA,9454
|
27
|
-
sglang/srt/model_config.py,sha256=k4OfRV-szWkFaJMIC40JoJGJ75AfYQ2hf4M1dS1aQ-o,6366
|
28
|
-
sglang/srt/sampling_params.py,sha256=uZFDlTUPnNR5_3IDH-INDeN-tm6LlRkC2KT-B3njxJs,3687
|
29
|
-
sglang/srt/server.py,sha256=ur_fDb-nEmlzz1mSKwWa87XFJdQM1gxFz4cahMcMatA,16028
|
30
|
-
sglang/srt/server_args.py,sha256=oUMzSSBrJ5_g0yeBapABUv2MlhDNWEfWLdLVROgqZOU,16305
|
31
|
-
sglang/srt/utils.py,sha256=GcRFf3pb5l-Q5TJU4gF-Wp7Ct46l3BO0aMpjlyHXp3I,23766
|
32
|
-
sglang/srt/constrained/__init__.py,sha256=NLpZGj9RIx83ejDrM_pfaRtqGgaPq_ggJszPQENUJ2E,2037
|
33
|
-
sglang/srt/constrained/base_tool_cache.py,sha256=1_m-AivPtWRwUgGiEZBafCrSFUGahK4UM4vgAd8TkMg,2004
|
34
|
-
sglang/srt/constrained/fsm_cache.py,sha256=GoPBr_9ZdJizF2PKbYoQw2I4ckfrUYwCeMZxB9sY3TM,2639
|
35
|
-
sglang/srt/constrained/jump_forward.py,sha256=IgZ8D0woy5FLIQvXkE8wZRYejDsfVkjU0sqUlkiv_f4,6193
|
36
|
-
sglang/srt/layers/context_flashattention_nopad.py,sha256=r_TpHuYAVgq1pN81PiWe1bebtY-p9MBndBaoIE2VXrk,5180
|
37
|
-
sglang/srt/layers/extend_attention.py,sha256=V5pm7toSDlzByaV4lGRgXVGWFUPf68chvvahlT2h4mk,14092
|
38
|
-
sglang/srt/layers/fused_moe.py,sha256=KmyXwau2OOZpQimGIQrHptzGNs1trIud5AKEEKXdzPU,20823
|
39
|
-
sglang/srt/layers/linear.py,sha256=3Se2FRXyqXcd-uvNx2b7s-jolsUTEVeYBMYHmV82wPw,34518
|
40
|
-
sglang/srt/layers/logits_processor.py,sha256=5Cg3h5b4H0EUeOJRst3IOMWL5dniP63A5s15BRkAMmk,11091
|
41
|
-
sglang/srt/layers/radix_attention.py,sha256=cNSQWO74DcXgpAMKSMaHzfpy5IcLORUnWe5gOwATLrw,7466
|
42
|
-
sglang/srt/layers/token_attention.py,sha256=pdBORaWQGvDy_Aitcq0XDHk2Rravol-jZZkrsgkXeng,8849
|
43
|
-
sglang/srt/layers/quantization/__init__.py,sha256=JMlgE-FWS759lfQ9Uc6mGFqBbTFLlvKeVEFpZLATe14,2536
|
44
|
-
sglang/srt/layers/quantization/fp8.py,sha256=GQOLeGbrcUfwO-7oClzDda0RXGPHR70ZXUHArZsa174,25511
|
45
|
-
sglang/srt/managers/controller_multi.py,sha256=LYI-XE9h57DW8Uh4gpd8upsC3p2dd5weKzddEH274jg,6626
|
46
|
-
sglang/srt/managers/controller_single.py,sha256=CdQ9_XPZdcWF5jArDmVR8K-WZ9_8Gpgk4SwANKxTX-Y,5112
|
47
|
-
sglang/srt/managers/detokenizer_manager.py,sha256=GXWdW4n2N-otL3zcgdr0t1PcEe2EmQJA8AElntiNV1o,5606
|
48
|
-
sglang/srt/managers/io_struct.py,sha256=VK61d6zfnBz5a3IMmwYsa5PNa9jUXPPmED1TdDRQGDs,7345
|
49
|
-
sglang/srt/managers/policy_scheduler.py,sha256=ajSB-gCC6VJkXvnKU8FYU3Kgcigozp2pMTwF84Wp14o,3138
|
50
|
-
sglang/srt/managers/schedule_batch.py,sha256=yIjiiMcaYYN9iaEOGQZoPUpFviDptMVh9hMwRRnDAco,37896
|
51
|
-
sglang/srt/managers/tokenizer_manager.py,sha256=kxkoAa8VbQt9FJPX1fN-7IzAD8RIcIvz3AGR8uEMYjk,21202
|
52
|
-
sglang/srt/managers/tp_worker.py,sha256=JPLneFwcPlmPXZX1QxZHWgcdau8FC8wNuVqfCqsgOkU,35234
|
53
|
-
sglang/srt/mem_cache/base_cache.py,sha256=czyN8IumXcMQskYOZDV3DzjfD4kdR-qwLVxceDqnOmE,788
|
54
|
-
sglang/srt/mem_cache/chunk_cache.py,sha256=u1mkGoTI7_31H0i0mhKT7S57StYSsdmsSPqyGubE7lY,1560
|
55
|
-
sglang/srt/mem_cache/flush_cache.py,sha256=pTLKPRB17U6vl5RFJJvuJ4jCL2SyomgkUBNlkDpGRqo,978
|
56
|
-
sglang/srt/mem_cache/memory_pool.py,sha256=8N4eHybhtBuwIwYyeNSvrZI90LGgMG8sA3OrXdXZAZs,5496
|
57
|
-
sglang/srt/mem_cache/radix_cache.py,sha256=pa5RD4xNKPSuvL55BnC4mimoca5oJRXr4Rg91-sbTcs,8881
|
58
|
-
sglang/srt/model_executor/cuda_graph_runner.py,sha256=OdmO6R7nHWrRJCtZOxYkt0KNdGoX7Md4knsypwPYjaQ,9365
|
59
|
-
sglang/srt/model_executor/model_runner.py,sha256=tJHlqk_JH3RJDaPAiSljaDI951LUu9AYW679eCKMJXs,17404
|
60
|
-
sglang/srt/model_loader/model_loader.py,sha256=QmZUhHh1nmWrfYlunfnxMcTsIvip1l6aMIlrXoCED4I,10697
|
61
|
-
sglang/srt/model_loader/utils.py,sha256=0AoWXX9uV5rKRYXJ4HduSnvdeerytI4ONCLCH6X4XFQ,10675
|
62
|
-
sglang/srt/models/chatglm.py,sha256=vYWooqyPmcSFZNjxj_g5I_FgHJlDytbEiz6vyv3JBNM,13856
|
63
|
-
sglang/srt/models/commandr.py,sha256=gaTI77hgitPlcUNyxMEdGu_XZQj2DuAMnh3KbZQ9HFg,14166
|
64
|
-
sglang/srt/models/dbrx.py,sha256=LQu7I2KH-XzY9iBlaK7IQsM1o3kzsuI1vTCspK2C19o,14655
|
65
|
-
sglang/srt/models/deepseek.py,sha256=adr57ZX6aPOBOpmvm7YIvoqo6u0jdrKJPZ8SGcVXAh8,16014
|
66
|
-
sglang/srt/models/deepseek_v2.py,sha256=jaVaQlL1aPCTu8nLcvtAW_rmtvHe6y2CviIOjXzh4q4,26962
|
67
|
-
sglang/srt/models/gemma.py,sha256=PMPI1-WLuLdk6e7u6I9d_LoCkauLkWY3aOP8MFEZ-sI,12279
|
68
|
-
sglang/srt/models/gemma2.py,sha256=kTjZcsptgtYaO8BL_NlygjVSMSloq2Mc4Rf3FKvEhbs,16420
|
69
|
-
sglang/srt/models/gpt_bigcode.py,sha256=U7GmHKywSu12D-EwvuWv3RwHkx6bPawaRIjlFIpQkfs,10194
|
70
|
-
sglang/srt/models/grok.py,sha256=NfZdsRVErDIUWFqjhtNf2pqC9G4cRdYHBFpgDq1IZ2A,27855
|
71
|
-
sglang/srt/models/internlm2.py,sha256=Ld2GUxZeqqqJ2vd4QiX2s1y2AceJLA1nVnUYY88GMQk,12219
|
72
|
-
sglang/srt/models/llama2.py,sha256=zfOk3OK1_B6s6yuXsZFmNCf07RsfytVD72GunLBt8Cc,14282
|
73
|
-
sglang/srt/models/llama_classification.py,sha256=4r_orFZqBR3U_yC4bus1K3Z3-ADscYGSzgA82_VDN0g,4926
|
74
|
-
sglang/srt/models/llava.py,sha256=BJphgyQGdo7uTpJcKGEfWwdpH9GTMDnyiznLSSgmvm8,18476
|
75
|
-
sglang/srt/models/llavavid.py,sha256=-7vaVqaIfukCvMkNakEPblpwjIHC6ezrAvmpE5RzlUY,13602
|
76
|
-
sglang/srt/models/minicpm.py,sha256=Mj-dbhfN7li7cTEP-0sV7i5PSYkMGIaYCqRU7eDc-BY,13837
|
77
|
-
sglang/srt/models/mistral.py,sha256=jlrWBVNXbAUziAaIdHAjFcOJnKtn9Bl8rBd65ypJM-I,819
|
78
|
-
sglang/srt/models/mixtral.py,sha256=QiswCUdZ4VwMghtrr_vGP_dkzxSCrcUIcBgjlOZh_Ao,21391
|
79
|
-
sglang/srt/models/mixtral_quant.py,sha256=I1sIdistZHw7GO35qvlteA16DGVtME5rvEVV86v0-7Y,14216
|
80
|
-
sglang/srt/models/qwen.py,sha256=xAtlWyhMkcfwocRqzZoH01qKbkohXxAf4tnkPh0xtpM,10000
|
81
|
-
sglang/srt/models/qwen2.py,sha256=mXlVd6UTCXY3VdgodFpQnlaY-NYLIbA-SknxdA9R13w,12278
|
82
|
-
sglang/srt/models/qwen2_moe.py,sha256=YYdJEezic7GyW-_bXlNIaqBa0C4IHQpz_vuRBLxms4k,18141
|
83
|
-
sglang/srt/models/stablelm.py,sha256=b3d-ZwLQoLjZ6CupnkIq7d-z9tzGSxAyIcgSmZiZxZw,11362
|
84
|
-
sglang/srt/models/yivl.py,sha256=p4s_D_m4H2exP4b91Y-CTkq8T-eIG3DJsFy9pB0e7TM,4932
|
85
|
-
sglang/srt/openai_api/adapter.py,sha256=p2HeYO9Qgl7EERXutwpsQ659NvZhFnkQmTZX5s-x-oI,37444
|
86
|
-
sglang/srt/openai_api/protocol.py,sha256=q1MuDUhwSM-8G2uGnWUMeEk87aZxei8lCcaP6VuA8So,8200
|
87
|
-
sglang/test/run_eval.py,sha256=kbM6SiosfXj-1uYTFXPWMd7hZDvJZwV-AmdHi_WfP3A,3559
|
88
|
-
sglang/test/runners.py,sha256=APXXbrqmUGUqnX7T1Aq8X2NJQkIqtv6B42a2ybdlPjA,7459
|
89
|
-
sglang/test/simple_eval_common.py,sha256=HL1bfgkTAKP7sk-kShg73WTeADhuBD6xSsuLbV_9C3s,12359
|
90
|
-
sglang/test/simple_eval_gpqa.py,sha256=CaRAuHdZj0m4mRm4tH9k7cB0kQxe0LHwlz7Vn1qyKps,3189
|
91
|
-
sglang/test/simple_eval_humaneval.py,sha256=k50DKoAbXiw-ubrFXHet9B-7tboHU2dQJf5G3C-KKq4,5838
|
92
|
-
sglang/test/simple_eval_math.py,sha256=EQblQmtUt-kl558drzhP7c6KhpDNgr1EJhhKx5eeHM4,2519
|
93
|
-
sglang/test/simple_eval_mmlu.py,sha256=KqSSdSu2qfoKQ870ttxev1NJ7c90xv2mvKOQsSODtAw,4326
|
94
|
-
sglang/test/test_programs.py,sha256=e9_ifoIvuI1Ctkbkz3wfdZLBBSRikby8ywcodBIkf9M,13826
|
95
|
-
sglang/test/test_utils.py,sha256=p-G6iiT5-Vkg6LMYgvDheomLJ6IYMLsYHCp3tkatiy8,13983
|
96
|
-
sglang-0.2.10.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
|
97
|
-
sglang-0.2.10.dist-info/METADATA,sha256=Lt9wnP2134unvF88fDj2PfQIf2YaeYJ6xZdfmMAJkoM,33303
|
98
|
-
sglang-0.2.10.dist-info/WHEEL,sha256=R0nc6qTxuoLk7ShA2_Y-UWkN8ZdfDBG2B6Eqpz2WXbs,91
|
99
|
-
sglang-0.2.10.dist-info/top_level.txt,sha256=yxhh3pYQkcnA7v3Bg889C2jZhvtJdEincysO7PEB09M,7
|
100
|
-
sglang-0.2.10.dist-info/RECORD,,
|
File without changes
|
File without changes
|
File without changes
|