liger-kernel-nightly 0.6.3.dev20251105190428__py3-none-any.whl → 0.6.3.dev20251105235313__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.
Files changed (40) hide show
  1. liger_kernel/ops/cross_entropy.py +59 -9
  2. liger_kernel/ops/fused_linear_cross_entropy.py +27 -4
  3. liger_kernel/transformers/cross_entropy.py +8 -3
  4. liger_kernel/transformers/functional.py +24 -6
  5. liger_kernel/transformers/fused_linear_cross_entropy.py +8 -3
  6. liger_kernel/transformers/model/falcon_h1.py +19 -5
  7. liger_kernel/transformers/model/gemma.py +17 -6
  8. liger_kernel/transformers/model/gemma2.py +14 -5
  9. liger_kernel/transformers/model/gemma3.py +25 -12
  10. liger_kernel/transformers/model/glm4.py +16 -4
  11. liger_kernel/transformers/model/glm4v.py +16 -4
  12. liger_kernel/transformers/model/glm4v_moe.py +19 -4
  13. liger_kernel/transformers/model/internvl.py +12 -5
  14. liger_kernel/transformers/model/llama.py +14 -5
  15. liger_kernel/transformers/model/llama4.py +16 -4
  16. liger_kernel/transformers/model/llava.py +12 -4
  17. liger_kernel/transformers/model/loss_utils.py +31 -3
  18. liger_kernel/transformers/model/mistral.py +15 -6
  19. liger_kernel/transformers/model/mixtral.py +16 -7
  20. liger_kernel/transformers/model/mllama.py +12 -4
  21. liger_kernel/transformers/model/olmo2.py +16 -4
  22. liger_kernel/transformers/model/output_classes.py +147 -0
  23. liger_kernel/transformers/model/paligemma.py +22 -5
  24. liger_kernel/transformers/model/phi3.py +14 -7
  25. liger_kernel/transformers/model/qwen2.py +16 -3
  26. liger_kernel/transformers/model/qwen2_5_vl.py +14 -6
  27. liger_kernel/transformers/model/qwen2_vl.py +16 -4
  28. liger_kernel/transformers/model/qwen3.py +18 -5
  29. liger_kernel/transformers/model/qwen3_moe.py +19 -5
  30. liger_kernel/transformers/model/qwen3_next.py +17 -5
  31. liger_kernel/transformers/model/qwen3_vl.py +11 -5
  32. liger_kernel/transformers/model/qwen3_vl_moe.py +12 -5
  33. liger_kernel/transformers/model/smollm3.py +15 -6
  34. liger_kernel/transformers/monkey_patch.py +4 -2
  35. {liger_kernel_nightly-0.6.3.dev20251105190428.dist-info → liger_kernel_nightly-0.6.3.dev20251105235313.dist-info}/METADATA +1 -1
  36. {liger_kernel_nightly-0.6.3.dev20251105190428.dist-info → liger_kernel_nightly-0.6.3.dev20251105235313.dist-info}/RECORD +40 -39
  37. {liger_kernel_nightly-0.6.3.dev20251105190428.dist-info → liger_kernel_nightly-0.6.3.dev20251105235313.dist-info}/LICENSE +0 -0
  38. {liger_kernel_nightly-0.6.3.dev20251105190428.dist-info → liger_kernel_nightly-0.6.3.dev20251105235313.dist-info}/NOTICE +0 -0
  39. {liger_kernel_nightly-0.6.3.dev20251105190428.dist-info → liger_kernel_nightly-0.6.3.dev20251105235313.dist-info}/WHEEL +0 -0
  40. {liger_kernel_nightly-0.6.3.dev20251105190428.dist-info → liger_kernel_nightly-0.6.3.dev20251105235313.dist-info}/top_level.txt +0 -0
@@ -17,10 +17,10 @@ liger_kernel/chunked_loss/kto_loss.py,sha256=llVCe6DkcpCo57seGWoMikaQVFApx764jsm
17
17
  liger_kernel/chunked_loss/orpo_loss.py,sha256=nu9UYG16dcMw93lvHi4_hYs3Q0FK1KnlmMRj7OpYU8s,4872
18
18
  liger_kernel/chunked_loss/simpo_loss.py,sha256=fy2w8KbhMrBv7b1jdIeH3bBFxY52bPQPZb3KwBvmurM,5385
19
19
  liger_kernel/ops/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
20
- liger_kernel/ops/cross_entropy.py,sha256=CEgAeX97ezIBRhK3dPQRKsEQiwgnBDOewtDoqKXzw_Q,19605
20
+ liger_kernel/ops/cross_entropy.py,sha256=-fd8qVxn_66MGSLs-Gs8yGmWlkET5YAoyb__Bolfz4c,22617
21
21
  liger_kernel/ops/dyt.py,sha256=gCLz4S8aul8SY9nvIGaoK67aGb7U9MJRQdo3ONqmQYs,5417
22
22
  liger_kernel/ops/fused_add_rms_norm.py,sha256=UBqmlqFCmhSAIpkNKd8rrfXatX7Z4J9bp2dX9A0lrJQ,14017
23
- liger_kernel/ops/fused_linear_cross_entropy.py,sha256=fnOYx1JyScQjsbb5r7Ua2S2bE-Abr0y8IH-oSWAvLGo,14583
23
+ liger_kernel/ops/fused_linear_cross_entropy.py,sha256=YepeWqX37gKc1-FUrzkDTzXYdOvmBmfv4KgL__KN_UI,16158
24
24
  liger_kernel/ops/fused_linear_jsd.py,sha256=CSoprxb-YcJy-YUKiTcYkxN8sb9h2kdk_iHuncvSV5c,9683
25
25
  liger_kernel/ops/fused_neighborhood_attention.py,sha256=vPi5xbnh6wxyZehaqo6Tuilqo2fN5SGDiONjnNmIKqs,35556
26
26
  liger_kernel/ops/geglu.py,sha256=r0WSq9E93zzynL44Wh8femzOWK07_SseBM_pJUyxT3s,4144
@@ -44,12 +44,12 @@ liger_kernel/ops/experimental/embedding.py,sha256=tolj3tItkzpSb30zWqDN2_yX4ectfl
44
44
  liger_kernel/ops/experimental/mm_int8int2.py,sha256=TrS9lpwekrik_w5qE7AhMJD1bcq-OidjtbsW80oZ6IM,13314
45
45
  liger_kernel/transformers/__init__.py,sha256=iV1X0gH1JXwgeb7AeY8Ryv7q3r44MLQvSvn79yIVDzw,9874
46
46
  liger_kernel/transformers/auto_model.py,sha256=0qCTRZt280Bj_LcFdzo9hlaR-BWNazawXOGgoCZjgEg,1545
47
- liger_kernel/transformers/cross_entropy.py,sha256=z3KTWQnFxr_IZaVjtYt0ZNEWQdDdYThN35xWkHlDGH0,1683
47
+ liger_kernel/transformers/cross_entropy.py,sha256=DMtHkKrVJDSsels7KgGQJqrXkEAd6Zopcdr-5oRmQgE,2010
48
48
  liger_kernel/transformers/dyt.py,sha256=i-4GPaMrl-jab9TVI5qN0-H9qycn_mCbV82ozU4nbmU,723
49
49
  liger_kernel/transformers/fsdp.py,sha256=CUiyjTmjkjY7pLXQv8ly9rnzgXw6529csd9pvtJNMYc,3096
50
- liger_kernel/transformers/functional.py,sha256=a8EGYjHDg34rhnaD4JpU8I20XJ7xiqJvqqjoh4NcwYk,8022
50
+ liger_kernel/transformers/functional.py,sha256=OqEmsDkaV3YiXaw1zqjDvHcC9_tU5TBrmhCNPOdgHQY,8590
51
51
  liger_kernel/transformers/fused_add_rms_norm.py,sha256=7_Bzg-x6lLe6W1qG2DtjDALhEpNZlC6N5GppEs9cTYY,1199
52
- liger_kernel/transformers/fused_linear_cross_entropy.py,sha256=toa54dpmJduoZLhU3lJA-HPZ03MYcMKekDWPcdYjvYA,2020
52
+ liger_kernel/transformers/fused_linear_cross_entropy.py,sha256=Hhp9XGgMKZhvlkjHY5Jkl_T7fSyJoCL9m5c3z_9mflQ,2347
53
53
  liger_kernel/transformers/fused_linear_jsd.py,sha256=bZ4otCvWBuOnA5XdQL-FzZVItJlDt-ht9e_pG7PG93E,3999
54
54
  liger_kernel/transformers/fused_neighborhood_attention.py,sha256=TxYDUAt9B6WSP14aJP66C_2Mbds2sSIPGnamhUSTrC8,7957
55
55
  liger_kernel/transformers/geglu.py,sha256=mrgqzIUVd6lN7fkDKLkw5YaESDxDtFgbot430WwPVOQ,1107
@@ -59,7 +59,7 @@ liger_kernel/transformers/jsd.py,sha256=DGqRnxIZxsvxo0_tbbxX3b-sDbDjC_yKufyRIHCc
59
59
  liger_kernel/transformers/kl_div.py,sha256=WLffFbh1EExD2Eb1F7lN11fo9JJC-0751WJjZAF1Fj8,409
60
60
  liger_kernel/transformers/layer_norm.py,sha256=c9pk3PEasOKYR0rhe5e5nNrnYKVCEW4VC8S6LpCq9EQ,906
61
61
  liger_kernel/transformers/llama4_rope.py,sha256=kS6PSHEwf3dS7hD7C7p8S0geugx2EMCiP0h0F7LsUoY,3639
62
- liger_kernel/transformers/monkey_patch.py,sha256=O_kl0l56oHinVv-bwl1LU5nKPm6nA0YBjKTYmmwgRbk,124732
62
+ liger_kernel/transformers/monkey_patch.py,sha256=ZGnLygHuCiKGd6hT-C0pt1aY85f6GNFdV98oCDpxHHo,124742
63
63
  liger_kernel/transformers/multi_token_attention.py,sha256=K3NIY9_5TPgZ4_Rahn0xnkMXxD_fmlJHK4CWGYvGQp0,1752
64
64
  liger_kernel/transformers/poly_norm.py,sha256=g5tC75i3qy1_N26ZUP-jfpct7ivQAEdJfIfx8IXzeyE,1377
65
65
  liger_kernel/transformers/qwen2vl_mrope.py,sha256=5EwSqrMdsL9MYspeBMXBsNJKvH0MOmRrtJXAJlnnlOI,1047
@@ -73,41 +73,42 @@ liger_kernel/transformers/tvd.py,sha256=XrRfyJIqN6HFxXk8MYyFVZM1OLz3mtSbRZvWfZ_J
73
73
  liger_kernel/transformers/experimental/__init__.py,sha256=oQqk-f32JYgWEP9DJCj6ty6bbJSGrdXsFDQFwGeX6vI,127
74
74
  liger_kernel/transformers/experimental/embedding.py,sha256=2P0QYdlFyFrG5OqTzTa1wcRgDSyjBMv5i1a7BrDPDQw,881
75
75
  liger_kernel/transformers/model/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
76
- liger_kernel/transformers/model/falcon_h1.py,sha256=DTzfT-5OzQ6I-pU80Vn5e5ibd1EOEbJV5cMTJFhfwFg,4302
77
- liger_kernel/transformers/model/gemma.py,sha256=WryzpVmCm2H_XgLKNu3jJ6gVawjQDjapTetg4WHlbR4,10078
78
- liger_kernel/transformers/model/gemma2.py,sha256=eOQEfJBKezJNNrirhkPSagGxr9qj_y4lENOZgjUZKpE,11471
79
- liger_kernel/transformers/model/gemma3.py,sha256=-tvZw88S-STqmvdim-xrZZRJ17KLWoge_73ilIvhpIU,14157
80
- liger_kernel/transformers/model/glm4.py,sha256=2TBM5-4URpj6uX96G1AZ_DrjAmQtgLwXGzBvaXtfwdk,5328
81
- liger_kernel/transformers/model/glm4v.py,sha256=nlgEMOBjFEOu7a-cwwp9mWhTFqIs3QrOvcxW-uaPq-s,6022
82
- liger_kernel/transformers/model/glm4v_moe.py,sha256=q3-R_FoQPayS85AriJWWebblXB6Ix9fvxhSrI3mHiz4,6237
83
- liger_kernel/transformers/model/internvl.py,sha256=Uv8KGXOz9NhiKVZDeRNzAJH5kRuMZikUbswWM9u5KM0,6069
84
- liger_kernel/transformers/model/llama.py,sha256=L_VuaxxFJpzEmpLnaqwBbI5-Q14Qgfj-ufhLydCWgdk,12903
85
- liger_kernel/transformers/model/llama4.py,sha256=epEO_VD1gJCDovabSIQLxxncoh-TQTBfj-UgIlR5c7U,4281
86
- liger_kernel/transformers/model/llava.py,sha256=t6kMiyBkteVam-ltiod2f1mevj8l8ZHxYDvfu9C_lEk,15196
87
- liger_kernel/transformers/model/loss_utils.py,sha256=02RVkPI7Qs4ZP4yU_udCAvD_2hgIaHmxremRKe3N7EE,1885
88
- liger_kernel/transformers/model/mistral.py,sha256=XmM4N21RIOkJ9PJ4PZ3DcRUhGUczn_lbx0plf1zeHb0,5571
89
- liger_kernel/transformers/model/mixtral.py,sha256=SLdLO81AZL7zror0LXLkn2PHqKzjwMMs4kALNqoaT00,11571
90
- liger_kernel/transformers/model/mllama.py,sha256=5q8q2BxQR_8hNZ83XrJIbndw-l6T7ZyFLM7OCv_uPK0,11593
91
- liger_kernel/transformers/model/olmo2.py,sha256=9O1Cze2B6ON-i1jgjQwjpS_WsDEK0PzL003s-MkevWA,5330
92
- liger_kernel/transformers/model/paligemma.py,sha256=mnTnSmEDla_bbVmPFmqhNVT__Cuf-TM-KLGFUa1sU-4,19967
93
- liger_kernel/transformers/model/phi3.py,sha256=L4gG8htOABmaxzcmHph0bBFCACRvL9r6wuDVFXi2o7Q,4117
94
- liger_kernel/transformers/model/qwen2.py,sha256=lgn0X6EzAZUhOv17ZDD9choIDdaPVIAsIrrdvwzWXqs,10033
95
- liger_kernel/transformers/model/qwen2_5_vl.py,sha256=Ea3zvL1FJfjlaerpeXCq-1zmorrajwNsR-XsgWr4fFQ,6465
96
- liger_kernel/transformers/model/qwen2_vl.py,sha256=ZeasFPGs-bxm2Y_E15mo0YNx5wwtKYDV-bjVKjkLPBk,6018
97
- liger_kernel/transformers/model/qwen3.py,sha256=Q2aOg5erPrgVgRcqJm8sefLSDtvU1AD5B7aJnP7mRMM,4956
98
- liger_kernel/transformers/model/qwen3_moe.py,sha256=1CwTMCNFDYsjGoa_aHFBagtC5HuJTV-s0__5UvcjD3A,5686
99
- liger_kernel/transformers/model/qwen3_next.py,sha256=7To7azriAogxeE7oEvByKztH9154dnDiDVNHHm7PZK4,5632
100
- liger_kernel/transformers/model/qwen3_vl.py,sha256=YU76HJ0A9kG5CUaZM4i9Bzci4eeXcNl_VSC2tsPWA3k,6301
101
- liger_kernel/transformers/model/qwen3_vl_moe.py,sha256=ykNIvGBtmcTkn236lhmJHzU1IHVR1Kq1YYYlJ5ynhw4,4445
102
- liger_kernel/transformers/model/smollm3.py,sha256=0KWVkDtXbjsBKhJnaquV6vUUYyLtfmNwYH0sxJt-qTk,7667
76
+ liger_kernel/transformers/model/falcon_h1.py,sha256=heUZ4wUt2ATmtBtmv8Rcro3pQl6fV9T0pburjTTW7os,5004
77
+ liger_kernel/transformers/model/gemma.py,sha256=pAri4PYpknsFfkvyo8Ez2NNlqrUDW-KkExUXTGZAcH4,10621
78
+ liger_kernel/transformers/model/gemma2.py,sha256=qa9Ok42vFojVGNmASTH3Ek566Vu507kjd--ZpZDKX9M,12024
79
+ liger_kernel/transformers/model/gemma3.py,sha256=mEV3Kuy-dqfTk_b899Vb-InuD4_DvwH0nm5xgbG-0MM,14911
80
+ liger_kernel/transformers/model/glm4.py,sha256=bSp22iPIjsli4-c_usUOsyh1Bs2gIK8X6ynS0azseUs,5900
81
+ liger_kernel/transformers/model/glm4v.py,sha256=dd-BQpccDCp1SbIxcJ5rG8xcwYQK3KOv1Tgm9TGnZc4,6594
82
+ liger_kernel/transformers/model/glm4v_moe.py,sha256=4UH3HGOWFZejfwLzC-bikwD7y2qQYDstYQWSCsWj2mo,6979
83
+ liger_kernel/transformers/model/internvl.py,sha256=OOutracs9qrPHSU7FVYar08yinvGrHQVPvo39JEws6w,6473
84
+ liger_kernel/transformers/model/llama.py,sha256=kqZeONzwTBzudoChlKMzq1w23BtYGbxWZC1l1V__JTw,13410
85
+ liger_kernel/transformers/model/llama4.py,sha256=PfkynGVI0xxMs3EtyYpCgaALI6stu25OIrTIymE-pvg,4853
86
+ liger_kernel/transformers/model/llava.py,sha256=yoADM_BuIEummtTDiwWqjfUjXUMZD78VJzS0TRj5GJ4,15687
87
+ liger_kernel/transformers/model/loss_utils.py,sha256=mAV6NsE1xR2smQMlr_n9afh4ek3BhIfieZdTn1Z-9Fw,2836
88
+ liger_kernel/transformers/model/mistral.py,sha256=OcwOzVDMwwDbVccVPv-AaocznzWwzLT3aRaKK5SMaAg,6030
89
+ liger_kernel/transformers/model/mixtral.py,sha256=YcBDoTEJDgLFJ_RTo180DYGxR8D5Ad9-idumif7kCPE,12130
90
+ liger_kernel/transformers/model/mllama.py,sha256=vAHwCm63sn4kpAY0rDGf_N0HR7KRTBVpBYDVTPOaZTg,12079
91
+ liger_kernel/transformers/model/olmo2.py,sha256=-h2bUOeuPfY1MdShdRvq5_wFDHKP4PEimgIl0fL-BT4,5902
92
+ liger_kernel/transformers/model/output_classes.py,sha256=0BGXVR4dYQpSHLkSqpRoXuHMryrceGSlTYRu6pvd8ZY,4542
93
+ liger_kernel/transformers/model/paligemma.py,sha256=r0smHLADkEwfLS6d6ArWoSWEeLt2d_8pmgOO5F04b1o,20793
94
+ liger_kernel/transformers/model/phi3.py,sha256=PT7Kw6yySg-7TsssWfi82eVMN3SWujCqzCqHigAdfeQ,4574
95
+ liger_kernel/transformers/model/qwen2.py,sha256=ojqdJpD3A9A5uCS0N_rSq8gyNYWSsHfuvx3Z3ObC7ss,10686
96
+ liger_kernel/transformers/model/qwen2_5_vl.py,sha256=FbIZDcg9cOr4PtBLNN8yVubN-gu2clndjSIzfi8NMos,6894
97
+ liger_kernel/transformers/model/qwen2_vl.py,sha256=967Ex4Scm0ehhiVxOtjwfj396nD9xkAwFwHcoURH6-o,6578
98
+ liger_kernel/transformers/model/qwen3.py,sha256=UkPx5uGuWcE_6qHLRfhZNFlf34eCuDQBPawyH04BrSc,5651
99
+ liger_kernel/transformers/model/qwen3_moe.py,sha256=yljJO4kyeM5U2Q4pXH3Mmq71ZFEC_Z73qgBx1-an-o8,6457
100
+ liger_kernel/transformers/model/qwen3_next.py,sha256=TayfD91GVLA1-fJwtVl6vMZgkUTYLQYURMRGBdCtnFc,6331
101
+ liger_kernel/transformers/model/qwen3_vl.py,sha256=sUIdJ-32IlFm_4pHv6PpLgVafqBS0QeJm_91tY67NdY,6646
102
+ liger_kernel/transformers/model/qwen3_vl_moe.py,sha256=CJEFcwBqItSEw9NA0mhEozlDTgIuJQ6VTjgkh5iLZ78,4856
103
+ liger_kernel/transformers/model/smollm3.py,sha256=1ewDY-99UAFJEfoeqfZxDcxjkqKYUSr5b7X-E_2BLLs,8126
103
104
  liger_kernel/transformers/model/smolvlm.py,sha256=yFpPKawLVo3zXzLjM7Y_T8FyRrPxVyp-YPFMM8m3k0c,6734
104
105
  liger_kernel/transformers/trainer/__init__.py,sha256=p7yQfklV8-467qSz_ZMimkbDF7HHWHwku25A-GYL0WU,193
105
106
  liger_kernel/transformers/trainer/orpo_trainer.py,sha256=tX0h63aOFe3rNqTmk6JpMf75UPo981yzEa6TghnjS0Q,5370
106
107
  liger_kernel/triton/__init__.py,sha256=qCiCamzCRv6lpV8IqpAc9YMdNKC7GKurClWceQPnlis,92
107
108
  liger_kernel/triton/monkey_patch.py,sha256=Rd0hUHAzDkFfHvnX7-PBaNK5EKnZhtfM_h-fgQH9HPY,1568
108
- liger_kernel_nightly-0.6.3.dev20251105190428.dist-info/LICENSE,sha256=OhzLDHJ0to4a8sodVLELZiCFylZ1NAAYLs-HrjPy0ag,1312
109
- liger_kernel_nightly-0.6.3.dev20251105190428.dist-info/METADATA,sha256=-9EYPsEnYUeI2OzXLaeZT-XyOBerw0KUkawlBL-jUcM,24777
110
- liger_kernel_nightly-0.6.3.dev20251105190428.dist-info/NOTICE,sha256=njwnoPZLh9AN8SJQzxvCGLHi-8X__AvWRze6joNXIY8,2066
111
- liger_kernel_nightly-0.6.3.dev20251105190428.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
112
- liger_kernel_nightly-0.6.3.dev20251105190428.dist-info/top_level.txt,sha256=2eghu4hA3LnkM7ElW92tQ8zegWKgSbeo-k-aGe1YnvY,13
113
- liger_kernel_nightly-0.6.3.dev20251105190428.dist-info/RECORD,,
109
+ liger_kernel_nightly-0.6.3.dev20251105235313.dist-info/LICENSE,sha256=OhzLDHJ0to4a8sodVLELZiCFylZ1NAAYLs-HrjPy0ag,1312
110
+ liger_kernel_nightly-0.6.3.dev20251105235313.dist-info/METADATA,sha256=6KnnAgqnY1JVClGWaYcFr7Q37aHNnKEHRS6RoDD5_ys,24777
111
+ liger_kernel_nightly-0.6.3.dev20251105235313.dist-info/NOTICE,sha256=njwnoPZLh9AN8SJQzxvCGLHi-8X__AvWRze6joNXIY8,2066
112
+ liger_kernel_nightly-0.6.3.dev20251105235313.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
113
+ liger_kernel_nightly-0.6.3.dev20251105235313.dist-info/top_level.txt,sha256=2eghu4hA3LnkM7ElW92tQ8zegWKgSbeo-k-aGe1YnvY,13
114
+ liger_kernel_nightly-0.6.3.dev20251105235313.dist-info/RECORD,,