diff --git a/config.json b/config.json
new file mode 100644
index 0000000000000000000000000000000000000000..43ec3420667962f974c7ceb840c4524f6bb0bd27
--- /dev/null
+++ b/config.json
@@ -0,0 +1 @@
+{"architectures": ["LLaMAForCausalLM"], "bos_token_id": 0, "eos_token_id": 1, "hidden_act": "silu", "hidden_size": 6656, "intermediate_size": 17920, "initializer_range": 0.02, "max_sequence_length": 2048, "model_type": "llama", "num_attention_heads": 52, "num_hidden_layers": 60, "pad_token_id": -1, "rms_norm_eps": 1e-06, "torch_dtype": "float16", "transformers_version": "4.27.0.dev0", "use_cache": true, "vocab_size": 32000}
\ No newline at end of file
diff --git a/generation_config.json b/generation_config.json
new file mode 100644
index 0000000000000000000000000000000000000000..8fdfba7b10dbe7d959d961b28f9e67433c0c0e9c
--- /dev/null
+++ b/generation_config.json
@@ -0,0 +1 @@
+{"_from_model_config": true, "bos_token_id": 0, "eos_token_id": 1, "pad_token_id": -1, "transformers_version": "4.27.0.dev0"}
\ No newline at end of file
diff --git a/pytorch_model-00000-of-00061.bin b/pytorch_model-00000-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..f34b8bae0ca1d8a4c990a85dc138e6699166db23
--- /dev/null
+++ b/pytorch_model-00000-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4973b21d44e4583fdc95522437ee15df16cef280797d713316e83fb10dc70e0e
+size 1337620210
diff --git a/pytorch_model-00001-of-00061.bin b/pytorch_model-00001-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0a3d4bc87f38fc17d1c32a9819e4e14f71a74602
--- /dev/null
+++ b/pytorch_model-00001-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e26dcd72e7967812038ef957ba58d2884678db1e5e9336d8335619ed0c4970b1
+size 1337620210
diff --git a/pytorch_model-00002-of-00061.bin b/pytorch_model-00002-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..199161236c484273109d89f2f6fdb1016450f922
--- /dev/null
+++ b/pytorch_model-00002-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12e65f293c8da08e96183c91b31f13d1c552406abc26f8f526d5bb06d8bd2fac
+size 1337620210
diff --git a/pytorch_model-00003-of-00061.bin b/pytorch_model-00003-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..7f4fc40ca0d7b473616309329e39460ba8c00556
--- /dev/null
+++ b/pytorch_model-00003-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44a082646bea5726c1805760c8f34c1b6996996f3c50146c3a1c801a7413f25f
+size 1337620210
diff --git a/pytorch_model-00004-of-00061.bin b/pytorch_model-00004-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..92f6e04458199b09addf7e4ae47e0653bc44ad49
--- /dev/null
+++ b/pytorch_model-00004-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf2540cad38b58f4a2e073eb01d84e75309af2861174142342634ca6562e7f27
+size 1337620210
diff --git a/pytorch_model-00005-of-00061.bin b/pytorch_model-00005-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a17a84dede43b5e0eb7bb8414c3b2626b64c946e
--- /dev/null
+++ b/pytorch_model-00005-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8cd18d38d5157f4c749ed1211c883c41cce4934b0f428a156b31622ab8ea2358
+size 1337620210
diff --git a/pytorch_model-00006-of-00061.bin b/pytorch_model-00006-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..71f664d4c0e34fcd76cec49e6d456816cf2efcf9
--- /dev/null
+++ b/pytorch_model-00006-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8e6d265ae64af0c0e186dedb4b315a7a6ad59ec5a40ee1bfd46753d574c11470
+size 1337620210
diff --git a/pytorch_model-00007-of-00061.bin b/pytorch_model-00007-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..31682d5a6c13b7741eb35f0579259ae9f5393c18
--- /dev/null
+++ b/pytorch_model-00007-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:131b3fab3c020445a72184d17a50cf1cdd77899719efeb7692625c8c4b4098d8
+size 1337620210
diff --git a/pytorch_model-00008-of-00061.bin b/pytorch_model-00008-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..de57fa4278dd9822011dca13fe298a7ef67d2116
--- /dev/null
+++ b/pytorch_model-00008-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d6359a4f6c102b229ada093a009547707627d1bb9af175c9b389960749c4396
+size 1337620210
diff --git a/pytorch_model-00009-of-00061.bin b/pytorch_model-00009-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..fcde3946fde84306285224caf76a304d7279ec7f
--- /dev/null
+++ b/pytorch_model-00009-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:44ea0e16e60983bfac93f85b38bb3cfecc588b772a20e1b50fda25179dd60940
+size 1337620210
diff --git a/pytorch_model-00010-of-00061.bin b/pytorch_model-00010-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..9cb0a88d0eeed45fa07306fa2e0c0c204ec095b3
--- /dev/null
+++ b/pytorch_model-00010-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b67de5d47023ad413905a9de9cc472fd344e5081795588e8eca199c9407c97eb
+size 1337620210
diff --git a/pytorch_model-00011-of-00061.bin b/pytorch_model-00011-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..62dfea56263317bf27ebb4ee18ade35f0fe543e6
--- /dev/null
+++ b/pytorch_model-00011-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9079204376feed571a2de1e1ae12981a683addfad5c00e4b6144a41d2694a614
+size 1337620210
diff --git a/pytorch_model-00012-of-00061.bin b/pytorch_model-00012-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..43820148aedf369477a81d71841c4f2dc8cd1155
--- /dev/null
+++ b/pytorch_model-00012-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ba7b4327d1de943347101a582e7c3aaedceb8d0846d3b962990f08400066aedf
+size 1337620210
diff --git a/pytorch_model-00013-of-00061.bin b/pytorch_model-00013-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..683b6d594774b8e8faf8b21bda5308d50bab7390
--- /dev/null
+++ b/pytorch_model-00013-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdd48972a3108e2b61e159311540523e13e22249552e64b8e806417cafcd24a2
+size 1337620210
diff --git a/pytorch_model-00014-of-00061.bin b/pytorch_model-00014-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..d081031aa934a221628b6bb35b2576518d2915d7
--- /dev/null
+++ b/pytorch_model-00014-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef8f104efce6d490d1677e7a6717d43494b41eb6be83aaa6a7454a297fc9c7b6
+size 1337620210
diff --git a/pytorch_model-00015-of-00061.bin b/pytorch_model-00015-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..5765d709578d180b0a484ae750ed67f564fb3b2a
--- /dev/null
+++ b/pytorch_model-00015-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5f5961cd4d6b44eaf6540ae6d3c7160a96fdc754738c482a4f84f2ca39e6d571
+size 1337620210
diff --git a/pytorch_model-00016-of-00061.bin b/pytorch_model-00016-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..baceec24e30e3d69c1de7ecdfb5492671fce9170
--- /dev/null
+++ b/pytorch_model-00016-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd4c065c013ea5857153d09aa99df93b4fe63e69e6b7d2a3132d7c3a5041a35a
+size 1337620210
diff --git a/pytorch_model-00017-of-00061.bin b/pytorch_model-00017-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..3b8359d95494040b80ecd1635a3fc3970f3eaa8b
--- /dev/null
+++ b/pytorch_model-00017-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2473a2bb09eabb5236aa7f831df684e977d9da3134b44a0c3f2a49f72ca8fa2
+size 1337620210
diff --git a/pytorch_model-00018-of-00061.bin b/pytorch_model-00018-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8216f6bfbb7d50d8a3136c00003f79603ac5cec7
--- /dev/null
+++ b/pytorch_model-00018-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ef02b9cfc02fdc3cfbc812a1d49a9221a9fde9032e84b91d21faf7415c1acbd9
+size 1337620210
diff --git a/pytorch_model-00019-of-00061.bin b/pytorch_model-00019-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0d9cd4a772a6b7487530b22893a5d5e2cdd1f899
--- /dev/null
+++ b/pytorch_model-00019-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:31b1d06ccb074168f0ccf7cb2ebc6078d1cdae6a1c6ab97ad5d97f7600fde0d0
+size 1337620210
diff --git a/pytorch_model-00020-of-00061.bin b/pytorch_model-00020-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0fde24d6526141c883d696300ce1b9009128a951
--- /dev/null
+++ b/pytorch_model-00020-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45d800d8dee1986a34153c396b2d5cd501affa4ac2986082028b5f619058ef52
+size 1337620210
diff --git a/pytorch_model-00021-of-00061.bin b/pytorch_model-00021-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..2d3e970406db0aa4cb9591d3748d0ddeacc7d65e
--- /dev/null
+++ b/pytorch_model-00021-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d9721f07a51ee49279428edac3c6644a1c6bde44d713a4d0175e7733dbdc9c0c
+size 1337620210
diff --git a/pytorch_model-00022-of-00061.bin b/pytorch_model-00022-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..899705fb6228128db8e6a7897a03a6a853911a23
--- /dev/null
+++ b/pytorch_model-00022-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77c93f28cc7c39e908e3f692d4ad4238d71de9cb82be51d9f200bf3c873747dd
+size 1337620210
diff --git a/pytorch_model-00023-of-00061.bin b/pytorch_model-00023-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0b3b2590d26f0657737477bb2db3ad06d6c7df9e
--- /dev/null
+++ b/pytorch_model-00023-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d46f1dee410916317d907d54e81771654bbab78aff17abbcde256ddb0c258752
+size 1337620210
diff --git a/pytorch_model-00024-of-00061.bin b/pytorch_model-00024-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..402dc8f86d7fb4171d96f990a25032c08aa6810c
--- /dev/null
+++ b/pytorch_model-00024-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ff77cde942aeab61126a8a0c8c36dac8e8a93e9240a5508e4ba2c69bf749b31
+size 1337620210
diff --git a/pytorch_model-00025-of-00061.bin b/pytorch_model-00025-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..fcdf4f874102d8d0b6211de626d78e8a652205b5
--- /dev/null
+++ b/pytorch_model-00025-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6fa48f916cd163028082776dbeebd9bc9173e3adc9116f913df0a2b174c35c4c
+size 1337620210
diff --git a/pytorch_model-00026-of-00061.bin b/pytorch_model-00026-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..4fc900762a18377a4fe0023a205b15b7c2b3b1a9
--- /dev/null
+++ b/pytorch_model-00026-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5dcda4e8049bb3dcd344dcdf8e669f276606a2aec9528a4f1454f9a2409aaa54
+size 1337620210
diff --git a/pytorch_model-00027-of-00061.bin b/pytorch_model-00027-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..f634e9eb14c29be99d95cfbec2432bc0fcafcdea
--- /dev/null
+++ b/pytorch_model-00027-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1cdeacbb55277d3c3c5b213eb5ce3509eca91cef0cb22bf23b6e22bc7b867e30
+size 1337620210
diff --git a/pytorch_model-00028-of-00061.bin b/pytorch_model-00028-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ae14ad7dddd57227298353bf6201743f64eff472
--- /dev/null
+++ b/pytorch_model-00028-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:507a0fa36bbeac4c5fec340cb000139d76c01ada14f04de004027aa8d7bd4ac3
+size 1337620210
diff --git a/pytorch_model-00029-of-00061.bin b/pytorch_model-00029-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..41451ff59f0ba1af62440a9910d7858d9112d29f
--- /dev/null
+++ b/pytorch_model-00029-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62746a327ba66f3f887c848be0644aca6bae43f54921fc4cf743ab9bfd536725
+size 1337620210
diff --git a/pytorch_model-00030-of-00061.bin b/pytorch_model-00030-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..9bf7e625ff1095231ab06b193a7d9f488fbf1024
--- /dev/null
+++ b/pytorch_model-00030-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8842a1d7c11391261f053053b37cdcd848bf84d7db77835e1d192ffcd6e2099b
+size 1337620210
diff --git a/pytorch_model-00031-of-00061.bin b/pytorch_model-00031-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..510c518948614f33faeb88267001aae9eb106edd
--- /dev/null
+++ b/pytorch_model-00031-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4f0594d8cf0bb8821a92e6bc2967545183a68f718d62b3553ca83a1164c98355
+size 1337620210
diff --git a/pytorch_model-00032-of-00061.bin b/pytorch_model-00032-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8e068d382204fc50683be829c91de8c7fad6fcc5
--- /dev/null
+++ b/pytorch_model-00032-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91b09a16a9b5245d518c396ae745552557359f698690b963361078c9a342eaeb
+size 1337620210
diff --git a/pytorch_model-00033-of-00061.bin b/pytorch_model-00033-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..e821e5d6b5bf3a523d55ce9b1133b29f0f3f83e4
--- /dev/null
+++ b/pytorch_model-00033-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3349ea7eb1e54ca2d603e7fa194334d53bad6f579fbbe3565a063a4b81f2f281
+size 1337620210
diff --git a/pytorch_model-00034-of-00061.bin b/pytorch_model-00034-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..f452bbb4399b825bec53a841f3238351b024585e
--- /dev/null
+++ b/pytorch_model-00034-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71695b7c46939277b452eb36a50a95ad91d70395cb97d7db36618dc74074ac88
+size 1337620210
diff --git a/pytorch_model-00035-of-00061.bin b/pytorch_model-00035-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1505686e293760b3ef1fdbafa886f147ae9514c0
--- /dev/null
+++ b/pytorch_model-00035-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aacf8dcf079bb9430178cc2c87ded6af05263063b0e1cb51ceb4e0000557a071
+size 1337620210
diff --git a/pytorch_model-00036-of-00061.bin b/pytorch_model-00036-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..382b07861a3b7ff6f5fc31178fb1c160c193e9fb
--- /dev/null
+++ b/pytorch_model-00036-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:10c85f6a7d207624073faa4e68d47a88750b41cd8170ada2eccfc85b9ed37cd1
+size 1337620210
diff --git a/pytorch_model-00037-of-00061.bin b/pytorch_model-00037-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..fb4eee72e270acabd604abc7edb0ad591ef6de28
--- /dev/null
+++ b/pytorch_model-00037-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6de0d9974b26e983567b0387d7b87f76f67edf1fb2c0ea03fcb1e2f4e5dc658f
+size 1337620210
diff --git a/pytorch_model-00038-of-00061.bin b/pytorch_model-00038-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1054e11e6e64ca18d894ab2db333b324e09cd279
--- /dev/null
+++ b/pytorch_model-00038-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79406f3923a8cdd9e79f7fbffde3549c701cdda665367a4e846fefab7652bb61
+size 1337620210
diff --git a/pytorch_model-00039-of-00061.bin b/pytorch_model-00039-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1ed8322bb5dbfcfef66e965e19bbaf56136ea0bb
--- /dev/null
+++ b/pytorch_model-00039-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b8484077be7a361eff286ec89856984c7818e93eb776154eb939d7c551167599
+size 1337620210
diff --git a/pytorch_model-00040-of-00061.bin b/pytorch_model-00040-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..4d29f125d121e379c12b509563ff94038b90c558
--- /dev/null
+++ b/pytorch_model-00040-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7746057dfea547498f902c49d9da4fcbe310578d038739a1d29cc8097a704913
+size 1337620210
diff --git a/pytorch_model-00041-of-00061.bin b/pytorch_model-00041-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..784402d977eeea83cafd8eeab174583dcd004a25
--- /dev/null
+++ b/pytorch_model-00041-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a7b982f94acfae8fb32812aa211fd28f140528c3a12621b090a8c4edc03d8b81
+size 1337620210
diff --git a/pytorch_model-00042-of-00061.bin b/pytorch_model-00042-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..a53149bf9c939b90995b3f5941df83ff4a957c7e
--- /dev/null
+++ b/pytorch_model-00042-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:394b8614fcb115753e4fbdc768029c3686628d03d8a0fe40d7ca491d486bb48a
+size 1337620210
diff --git a/pytorch_model-00043-of-00061.bin b/pytorch_model-00043-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..95e9c089076fcaac00d472c85e4092ca512b7578
--- /dev/null
+++ b/pytorch_model-00043-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9f75fc52f1f482f30979119eb91a5ed8f15bb2a8b593987545f09a4b26ec2720
+size 1337620210
diff --git a/pytorch_model-00044-of-00061.bin b/pytorch_model-00044-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..aaf97b4bc522740a403fd5ec938f014678434856
--- /dev/null
+++ b/pytorch_model-00044-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a1469cd8392417220d312e4b0c3d6e6c9afd976425a1cf125997ae2ec744c35
+size 1337620210
diff --git a/pytorch_model-00045-of-00061.bin b/pytorch_model-00045-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..214eb99a6f286c28d529768cf519df84ec724009
--- /dev/null
+++ b/pytorch_model-00045-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51c62a84277bdc35d1ab4b5b65a87e1f204ff4103413bfb3525cf2c3ada59ccd
+size 1337620210
diff --git a/pytorch_model-00046-of-00061.bin b/pytorch_model-00046-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..80cd16d567c791ef34792aa376430d5faaf1d0c2
--- /dev/null
+++ b/pytorch_model-00046-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6ba8b192bd6a3322cb71a4c04b7c6a76f161ae94b2dbf27b9788d5c7753751cc
+size 1337620210
diff --git a/pytorch_model-00047-of-00061.bin b/pytorch_model-00047-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..0df29cef65870934f295d61a463ae575bd93cfaa
--- /dev/null
+++ b/pytorch_model-00047-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cc9603f655e80ccb5d9393c7e43936a1bce9e38438876729e1cd3c67b001e6fd
+size 1337620210
diff --git a/pytorch_model-00048-of-00061.bin b/pytorch_model-00048-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..ebe44978ec8afde6f70b598c03cc38260a05d5c9
--- /dev/null
+++ b/pytorch_model-00048-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:353d1a8c0f8d39f55637ea22e20d6ef429bf8c0fba87be79b572a54183348e29
+size 1337620210
diff --git a/pytorch_model-00049-of-00061.bin b/pytorch_model-00049-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..83a1d872cea6cf59590dba6a73b57d2d064fb8d9
--- /dev/null
+++ b/pytorch_model-00049-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e4994a4bc02d5300d58da999dc6cb12d4d1c8e5c159f3f5ba92349940c2f969e
+size 1337620210
diff --git a/pytorch_model-00050-of-00061.bin b/pytorch_model-00050-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b7ceeb6382925ed1e1282fd6a657486d9bc40b1c
--- /dev/null
+++ b/pytorch_model-00050-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b3cb3a3cb1d1e0800803ab3e6f214e2214ff4c8d02e7140fae5a57a4f9be8fc
+size 1337620210
diff --git a/pytorch_model-00051-of-00061.bin b/pytorch_model-00051-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1dce3e2851200c415124a3d0cd0e62a25992cef5
--- /dev/null
+++ b/pytorch_model-00051-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:79a71fcb296bd0dfa123bad6195f3f48e47a3f47d327367aaa6df8b19e82573b
+size 1337620210
diff --git a/pytorch_model-00052-of-00061.bin b/pytorch_model-00052-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..7c4b29cd04e960f955b3707d1d3a5a1f066d6fdc
--- /dev/null
+++ b/pytorch_model-00052-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4ed64fd83c7c389452bfa36c3f23b6da450b22cbdf9cb63b31da7454786ef158
+size 1337620210
diff --git a/pytorch_model-00053-of-00061.bin b/pytorch_model-00053-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..82353375964be7aeae4fec4959196a87223ecd61
--- /dev/null
+++ b/pytorch_model-00053-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2dc4b785351dd33416d7741f1b4b33fc8839578d68f40dbaea86557a150d8f94
+size 1337620210
diff --git a/pytorch_model-00054-of-00061.bin b/pytorch_model-00054-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..b79369b350f789ec818fdce711e075fabb0cf617
--- /dev/null
+++ b/pytorch_model-00054-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b3ff736ce6eb4c328c5d01a05b7fc2996000b68911bf11c43780c2bbce5e376
+size 1337620210
diff --git a/pytorch_model-00055-of-00061.bin b/pytorch_model-00055-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..8244eb1ce49c451eff90a42fb8d3add44dadf500
--- /dev/null
+++ b/pytorch_model-00055-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff70a42c747bbcec0ce3d0a811181058a398d989504087b10ac5c7118538dd4d
+size 1337620210
diff --git a/pytorch_model-00056-of-00061.bin b/pytorch_model-00056-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..06e4502462a50d25cc7a6218e72b20cf5b31497f
--- /dev/null
+++ b/pytorch_model-00056-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8f30751c0a02123c9527f5136ecb9330d3ab791669705f7a4a176a58e6c30a21
+size 1337620210
diff --git a/pytorch_model-00057-of-00061.bin b/pytorch_model-00057-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..5594a7298d7b57b124ac36c930f525d80e8de221
--- /dev/null
+++ b/pytorch_model-00057-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:afd7ba782e05564ed215656f1bfe9d72e27e01379740ee312cc1736670131437
+size 1337620210
diff --git a/pytorch_model-00058-of-00061.bin b/pytorch_model-00058-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..1c5e37d6da8eac67d8b5c66394b10eb9ca1112a1
--- /dev/null
+++ b/pytorch_model-00058-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f353a577561e159f98f60d12348e666ebd0eab27ed10c5804c43ac01966ca90
+size 1337620210
diff --git a/pytorch_model-00059-of-00061.bin b/pytorch_model-00059-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..67a621deb744c2e639184f7743f6d3a5a9c1cd7e
--- /dev/null
+++ b/pytorch_model-00059-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4778f514afa332126841cab26fdfe7b352f5bec0f929cbcdbea962ee1d21faa8
+size 1337620210
diff --git a/pytorch_model-00060-of-00061.bin b/pytorch_model-00060-of-00061.bin
new file mode 100644
index 0000000000000000000000000000000000000000..abbe0decbb0cbbdfe018af75df3f8ade3b0aeb1a
--- /dev/null
+++ b/pytorch_model-00060-of-00061.bin
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7742a45080931322a60806b8064eaef97e8b526442151e7dfff433425c5e8f14
+size 1064974796
diff --git a/pytorch_model.bin.index.json b/pytorch_model.bin.index.json
new file mode 100644
index 0000000000000000000000000000000000000000..1cf70c46d585389bc0d525fb4de81d5cb84f011c
--- /dev/null
+++ b/pytorch_model.bin.index.json
@@ -0,0 +1 @@
+{"weight_map": {"model.layers.0.input_layernorm.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.post_attention_layernorm.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.self_attn.q_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.self_attn.k_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.self_attn.v_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.self_attn.o_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.mlp.gate_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.mlp.down_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.mlp.up_proj.weight": "pytorch_model-00000-of-00061.bin", "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00000-of-00061.bin", "model.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00061.bin", "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00061.bin", "model.layers.2.input_layernorm.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.post_attention_layernorm.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.self_attn.q_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.self_attn.k_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.self_attn.v_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.self_attn.o_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.mlp.gate_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.mlp.down_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.mlp.up_proj.weight": "pytorch_model-00002-of-00061.bin", "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00061.bin", "model.layers.3.input_layernorm.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.post_attention_layernorm.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.self_attn.q_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.self_attn.k_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.self_attn.v_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.self_attn.o_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.mlp.gate_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.mlp.down_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.mlp.up_proj.weight": "pytorch_model-00003-of-00061.bin", "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00061.bin", "model.layers.4.input_layernorm.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.post_attention_layernorm.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.self_attn.q_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.self_attn.k_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.self_attn.v_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.self_attn.o_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.mlp.gate_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.mlp.down_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.mlp.up_proj.weight": "pytorch_model-00004-of-00061.bin", "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00061.bin", "model.layers.5.input_layernorm.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.post_attention_layernorm.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.self_attn.q_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.self_attn.k_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.self_attn.v_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.self_attn.o_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.mlp.gate_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.mlp.down_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.mlp.up_proj.weight": "pytorch_model-00005-of-00061.bin", "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00061.bin", "model.layers.6.input_layernorm.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.post_attention_layernorm.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.self_attn.q_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.self_attn.k_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.self_attn.v_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.self_attn.o_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.mlp.gate_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.mlp.down_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.mlp.up_proj.weight": "pytorch_model-00006-of-00061.bin", "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00061.bin", "model.layers.7.input_layernorm.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.post_attention_layernorm.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.self_attn.q_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.self_attn.k_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.self_attn.v_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.self_attn.o_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.mlp.gate_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.mlp.down_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.mlp.up_proj.weight": "pytorch_model-00007-of-00061.bin", "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00061.bin", "model.layers.8.input_layernorm.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.post_attention_layernorm.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.self_attn.q_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.self_attn.k_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.self_attn.v_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.self_attn.o_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.mlp.gate_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.mlp.down_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.mlp.up_proj.weight": "pytorch_model-00008-of-00061.bin", "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00061.bin", "model.layers.9.input_layernorm.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.post_attention_layernorm.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.self_attn.q_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.self_attn.k_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.self_attn.v_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.self_attn.o_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.mlp.gate_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.mlp.down_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.mlp.up_proj.weight": "pytorch_model-00009-of-00061.bin", "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00061.bin", "model.layers.10.input_layernorm.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.post_attention_layernorm.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.self_attn.q_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.self_attn.k_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.self_attn.v_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.self_attn.o_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.mlp.gate_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.mlp.down_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.mlp.up_proj.weight": "pytorch_model-00010-of-00061.bin", "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00061.bin", "model.layers.11.input_layernorm.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.post_attention_layernorm.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.self_attn.q_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.self_attn.k_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.self_attn.v_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.self_attn.o_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.mlp.gate_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.mlp.down_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.mlp.up_proj.weight": "pytorch_model-00011-of-00061.bin", "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00061.bin", "model.layers.12.input_layernorm.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.post_attention_layernorm.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.self_attn.q_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.self_attn.k_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.self_attn.v_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.self_attn.o_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.mlp.gate_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.mlp.down_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.mlp.up_proj.weight": "pytorch_model-00012-of-00061.bin", "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00061.bin", "model.layers.13.input_layernorm.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.post_attention_layernorm.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.self_attn.q_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.self_attn.k_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.self_attn.v_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.self_attn.o_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.mlp.gate_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.mlp.down_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.mlp.up_proj.weight": "pytorch_model-00013-of-00061.bin", "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00061.bin", "model.layers.14.input_layernorm.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.post_attention_layernorm.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.self_attn.q_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.self_attn.k_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.self_attn.v_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.self_attn.o_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.mlp.gate_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.mlp.down_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.mlp.up_proj.weight": "pytorch_model-00014-of-00061.bin", "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00014-of-00061.bin", "model.layers.15.input_layernorm.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.post_attention_layernorm.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.self_attn.q_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.self_attn.k_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.self_attn.v_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.self_attn.o_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.mlp.gate_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.mlp.down_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.mlp.up_proj.weight": "pytorch_model-00015-of-00061.bin", "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00015-of-00061.bin", "model.layers.16.input_layernorm.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.post_attention_layernorm.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.self_attn.q_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.self_attn.k_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.self_attn.v_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.self_attn.o_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.mlp.gate_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.mlp.down_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.mlp.up_proj.weight": "pytorch_model-00016-of-00061.bin", "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00016-of-00061.bin", "model.layers.17.input_layernorm.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.post_attention_layernorm.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.self_attn.q_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.self_attn.k_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.self_attn.v_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.self_attn.o_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.mlp.gate_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.mlp.down_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.mlp.up_proj.weight": "pytorch_model-00017-of-00061.bin", "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00017-of-00061.bin", "model.layers.18.input_layernorm.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.post_attention_layernorm.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.self_attn.q_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.self_attn.k_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.self_attn.v_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.self_attn.o_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.mlp.gate_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.mlp.down_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.mlp.up_proj.weight": "pytorch_model-00018-of-00061.bin", "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00018-of-00061.bin", "model.layers.19.input_layernorm.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.post_attention_layernorm.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.self_attn.q_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.self_attn.k_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.self_attn.v_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.self_attn.o_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.mlp.gate_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.mlp.down_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.mlp.up_proj.weight": "pytorch_model-00019-of-00061.bin", "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00019-of-00061.bin", "model.layers.20.input_layernorm.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.post_attention_layernorm.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.self_attn.q_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.self_attn.k_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.self_attn.v_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.self_attn.o_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.mlp.gate_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.mlp.down_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.mlp.up_proj.weight": "pytorch_model-00020-of-00061.bin", "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00020-of-00061.bin", "model.layers.21.input_layernorm.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.post_attention_layernorm.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.self_attn.q_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.self_attn.k_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.self_attn.v_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.self_attn.o_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.mlp.gate_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.mlp.down_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.mlp.up_proj.weight": "pytorch_model-00021-of-00061.bin", "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00021-of-00061.bin", "model.layers.22.input_layernorm.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.post_attention_layernorm.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.self_attn.q_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.self_attn.k_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.self_attn.v_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.self_attn.o_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.mlp.gate_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.mlp.down_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.mlp.up_proj.weight": "pytorch_model-00022-of-00061.bin", "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00022-of-00061.bin", "model.layers.23.input_layernorm.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.post_attention_layernorm.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.self_attn.q_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.self_attn.k_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.self_attn.v_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.self_attn.o_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.mlp.gate_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.mlp.down_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.mlp.up_proj.weight": "pytorch_model-00023-of-00061.bin", "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00023-of-00061.bin", "model.layers.24.input_layernorm.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.post_attention_layernorm.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.self_attn.q_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.self_attn.k_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.self_attn.v_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.self_attn.o_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.mlp.gate_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.mlp.down_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.mlp.up_proj.weight": "pytorch_model-00024-of-00061.bin", "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00024-of-00061.bin", "model.layers.25.input_layernorm.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.post_attention_layernorm.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.self_attn.q_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.self_attn.k_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.self_attn.v_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.self_attn.o_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.mlp.gate_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.mlp.down_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.mlp.up_proj.weight": "pytorch_model-00025-of-00061.bin", "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00025-of-00061.bin", "model.layers.26.input_layernorm.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.post_attention_layernorm.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.self_attn.q_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.self_attn.k_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.self_attn.v_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.self_attn.o_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.mlp.gate_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.mlp.down_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.mlp.up_proj.weight": "pytorch_model-00026-of-00061.bin", "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00026-of-00061.bin", "model.layers.27.input_layernorm.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.post_attention_layernorm.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.self_attn.q_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.self_attn.k_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.self_attn.v_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.self_attn.o_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.mlp.gate_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.mlp.down_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.mlp.up_proj.weight": "pytorch_model-00027-of-00061.bin", "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00027-of-00061.bin", "model.layers.28.input_layernorm.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.post_attention_layernorm.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.self_attn.q_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.self_attn.k_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.self_attn.v_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.self_attn.o_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.mlp.gate_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.mlp.down_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.mlp.up_proj.weight": "pytorch_model-00028-of-00061.bin", "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00028-of-00061.bin", "model.layers.29.input_layernorm.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.post_attention_layernorm.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.self_attn.q_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.self_attn.k_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.self_attn.v_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.self_attn.o_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.mlp.gate_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.mlp.down_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.mlp.up_proj.weight": "pytorch_model-00029-of-00061.bin", "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00029-of-00061.bin", "model.layers.30.input_layernorm.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.post_attention_layernorm.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.self_attn.q_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.self_attn.k_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.self_attn.v_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.self_attn.o_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.mlp.gate_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.mlp.down_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.mlp.up_proj.weight": "pytorch_model-00030-of-00061.bin", "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00030-of-00061.bin", "model.layers.31.input_layernorm.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.post_attention_layernorm.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.self_attn.q_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.self_attn.k_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.self_attn.v_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.self_attn.o_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.mlp.gate_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.mlp.down_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.mlp.up_proj.weight": "pytorch_model-00031-of-00061.bin", "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00031-of-00061.bin", "model.layers.32.input_layernorm.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.post_attention_layernorm.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.self_attn.q_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.self_attn.k_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.self_attn.v_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.self_attn.o_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.mlp.gate_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.mlp.down_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.mlp.up_proj.weight": "pytorch_model-00032-of-00061.bin", "model.layers.32.self_attn.rotary_emb.inv_freq": "pytorch_model-00032-of-00061.bin", "model.layers.33.input_layernorm.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.post_attention_layernorm.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.self_attn.q_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.self_attn.k_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.self_attn.v_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.self_attn.o_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.mlp.gate_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.mlp.down_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.mlp.up_proj.weight": "pytorch_model-00033-of-00061.bin", "model.layers.33.self_attn.rotary_emb.inv_freq": "pytorch_model-00033-of-00061.bin", "model.layers.34.input_layernorm.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.post_attention_layernorm.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.self_attn.q_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.self_attn.k_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.self_attn.v_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.self_attn.o_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.mlp.gate_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.mlp.down_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.mlp.up_proj.weight": "pytorch_model-00034-of-00061.bin", "model.layers.34.self_attn.rotary_emb.inv_freq": "pytorch_model-00034-of-00061.bin", "model.layers.35.input_layernorm.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.post_attention_layernorm.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.self_attn.q_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.self_attn.k_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.self_attn.v_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.self_attn.o_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.mlp.gate_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.mlp.down_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.mlp.up_proj.weight": "pytorch_model-00035-of-00061.bin", "model.layers.35.self_attn.rotary_emb.inv_freq": "pytorch_model-00035-of-00061.bin", "model.layers.36.input_layernorm.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.post_attention_layernorm.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.self_attn.q_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.self_attn.k_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.self_attn.v_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.self_attn.o_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.mlp.gate_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.mlp.down_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.mlp.up_proj.weight": "pytorch_model-00036-of-00061.bin", "model.layers.36.self_attn.rotary_emb.inv_freq": "pytorch_model-00036-of-00061.bin", "model.layers.37.input_layernorm.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.post_attention_layernorm.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.self_attn.q_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.self_attn.k_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.self_attn.v_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.self_attn.o_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.mlp.gate_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.mlp.down_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.mlp.up_proj.weight": "pytorch_model-00037-of-00061.bin", "model.layers.37.self_attn.rotary_emb.inv_freq": "pytorch_model-00037-of-00061.bin", "model.layers.38.input_layernorm.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.post_attention_layernorm.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.self_attn.q_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.self_attn.k_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.self_attn.v_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.self_attn.o_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.mlp.gate_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.mlp.down_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.mlp.up_proj.weight": "pytorch_model-00038-of-00061.bin", "model.layers.38.self_attn.rotary_emb.inv_freq": "pytorch_model-00038-of-00061.bin", "model.layers.39.input_layernorm.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.post_attention_layernorm.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.self_attn.q_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.self_attn.k_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.self_attn.v_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.self_attn.o_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.mlp.gate_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.mlp.down_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.mlp.up_proj.weight": "pytorch_model-00039-of-00061.bin", "model.layers.39.self_attn.rotary_emb.inv_freq": "pytorch_model-00039-of-00061.bin", "model.layers.40.input_layernorm.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.post_attention_layernorm.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.self_attn.q_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.self_attn.k_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.self_attn.v_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.self_attn.o_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.mlp.gate_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.mlp.down_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.mlp.up_proj.weight": "pytorch_model-00040-of-00061.bin", "model.layers.40.self_attn.rotary_emb.inv_freq": "pytorch_model-00040-of-00061.bin", "model.layers.41.input_layernorm.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.post_attention_layernorm.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.self_attn.q_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.self_attn.k_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.self_attn.v_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.self_attn.o_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.mlp.gate_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.mlp.down_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.mlp.up_proj.weight": "pytorch_model-00041-of-00061.bin", "model.layers.41.self_attn.rotary_emb.inv_freq": "pytorch_model-00041-of-00061.bin", "model.layers.42.input_layernorm.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.post_attention_layernorm.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.self_attn.q_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.self_attn.k_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.self_attn.v_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.self_attn.o_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.mlp.gate_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.mlp.down_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.mlp.up_proj.weight": "pytorch_model-00042-of-00061.bin", "model.layers.42.self_attn.rotary_emb.inv_freq": "pytorch_model-00042-of-00061.bin", "model.layers.43.input_layernorm.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.post_attention_layernorm.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.self_attn.q_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.self_attn.k_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.self_attn.v_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.self_attn.o_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.mlp.gate_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.mlp.down_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.mlp.up_proj.weight": "pytorch_model-00043-of-00061.bin", "model.layers.43.self_attn.rotary_emb.inv_freq": "pytorch_model-00043-of-00061.bin", "model.layers.44.input_layernorm.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.post_attention_layernorm.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.self_attn.q_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.self_attn.k_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.self_attn.v_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.self_attn.o_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.mlp.gate_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.mlp.down_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.mlp.up_proj.weight": "pytorch_model-00044-of-00061.bin", "model.layers.44.self_attn.rotary_emb.inv_freq": "pytorch_model-00044-of-00061.bin", "model.layers.45.input_layernorm.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.post_attention_layernorm.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.self_attn.q_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.self_attn.k_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.self_attn.v_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.self_attn.o_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.mlp.gate_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.mlp.down_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.mlp.up_proj.weight": "pytorch_model-00045-of-00061.bin", "model.layers.45.self_attn.rotary_emb.inv_freq": "pytorch_model-00045-of-00061.bin", "model.layers.46.input_layernorm.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.post_attention_layernorm.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.self_attn.q_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.self_attn.k_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.self_attn.v_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.self_attn.o_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.mlp.gate_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.mlp.down_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.mlp.up_proj.weight": "pytorch_model-00046-of-00061.bin", "model.layers.46.self_attn.rotary_emb.inv_freq": "pytorch_model-00046-of-00061.bin", "model.layers.47.input_layernorm.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.post_attention_layernorm.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.self_attn.q_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.self_attn.k_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.self_attn.v_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.self_attn.o_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.mlp.gate_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.mlp.down_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.mlp.up_proj.weight": "pytorch_model-00047-of-00061.bin", "model.layers.47.self_attn.rotary_emb.inv_freq": "pytorch_model-00047-of-00061.bin", "model.layers.48.input_layernorm.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.post_attention_layernorm.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.self_attn.q_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.self_attn.k_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.self_attn.v_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.self_attn.o_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.mlp.gate_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.mlp.down_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.mlp.up_proj.weight": "pytorch_model-00048-of-00061.bin", "model.layers.48.self_attn.rotary_emb.inv_freq": "pytorch_model-00048-of-00061.bin", "model.layers.49.input_layernorm.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.post_attention_layernorm.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.self_attn.q_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.self_attn.k_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.self_attn.v_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.self_attn.o_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.mlp.gate_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.mlp.down_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.mlp.up_proj.weight": "pytorch_model-00049-of-00061.bin", "model.layers.49.self_attn.rotary_emb.inv_freq": "pytorch_model-00049-of-00061.bin", "model.layers.50.input_layernorm.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.post_attention_layernorm.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.self_attn.q_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.self_attn.k_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.self_attn.v_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.self_attn.o_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.mlp.gate_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.mlp.down_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.mlp.up_proj.weight": "pytorch_model-00050-of-00061.bin", "model.layers.50.self_attn.rotary_emb.inv_freq": "pytorch_model-00050-of-00061.bin", "model.layers.51.input_layernorm.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.post_attention_layernorm.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.self_attn.q_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.self_attn.k_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.self_attn.v_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.self_attn.o_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.mlp.gate_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.mlp.down_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.mlp.up_proj.weight": "pytorch_model-00051-of-00061.bin", "model.layers.51.self_attn.rotary_emb.inv_freq": "pytorch_model-00051-of-00061.bin", "model.layers.52.input_layernorm.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.post_attention_layernorm.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.self_attn.q_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.self_attn.k_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.self_attn.v_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.self_attn.o_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.mlp.gate_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.mlp.down_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.mlp.up_proj.weight": "pytorch_model-00052-of-00061.bin", "model.layers.52.self_attn.rotary_emb.inv_freq": "pytorch_model-00052-of-00061.bin", "model.layers.53.input_layernorm.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.post_attention_layernorm.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.self_attn.q_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.self_attn.k_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.self_attn.v_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.self_attn.o_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.mlp.gate_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.mlp.down_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.mlp.up_proj.weight": "pytorch_model-00053-of-00061.bin", "model.layers.53.self_attn.rotary_emb.inv_freq": "pytorch_model-00053-of-00061.bin", "model.layers.54.input_layernorm.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.post_attention_layernorm.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.self_attn.q_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.self_attn.k_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.self_attn.v_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.self_attn.o_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.mlp.gate_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.mlp.down_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.mlp.up_proj.weight": "pytorch_model-00054-of-00061.bin", "model.layers.54.self_attn.rotary_emb.inv_freq": "pytorch_model-00054-of-00061.bin", "model.layers.55.input_layernorm.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.post_attention_layernorm.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.self_attn.q_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.self_attn.k_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.self_attn.v_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.self_attn.o_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.mlp.gate_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.mlp.down_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.mlp.up_proj.weight": "pytorch_model-00055-of-00061.bin", "model.layers.55.self_attn.rotary_emb.inv_freq": "pytorch_model-00055-of-00061.bin", "model.layers.56.input_layernorm.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.post_attention_layernorm.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.self_attn.q_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.self_attn.k_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.self_attn.v_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.self_attn.o_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.mlp.gate_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.mlp.down_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.mlp.up_proj.weight": "pytorch_model-00056-of-00061.bin", "model.layers.56.self_attn.rotary_emb.inv_freq": "pytorch_model-00056-of-00061.bin", "model.layers.57.input_layernorm.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.post_attention_layernorm.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.self_attn.q_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.self_attn.k_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.self_attn.v_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.self_attn.o_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.mlp.gate_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.mlp.down_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.mlp.up_proj.weight": "pytorch_model-00057-of-00061.bin", "model.layers.57.self_attn.rotary_emb.inv_freq": "pytorch_model-00057-of-00061.bin", "model.layers.58.input_layernorm.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.post_attention_layernorm.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.self_attn.q_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.self_attn.k_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.self_attn.v_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.self_attn.o_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.mlp.gate_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.mlp.down_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.mlp.up_proj.weight": "pytorch_model-00058-of-00061.bin", "model.layers.58.self_attn.rotary_emb.inv_freq": "pytorch_model-00058-of-00061.bin", "model.layers.59.input_layernorm.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.post_attention_layernorm.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.self_attn.q_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.self_attn.k_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.self_attn.v_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.self_attn.o_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.mlp.gate_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.mlp.down_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.mlp.up_proj.weight": "pytorch_model-00059-of-00061.bin", "model.layers.59.self_attn.rotary_emb.inv_freq": "pytorch_model-00059-of-00061.bin", "model.norm.weight": "pytorch_model-00060-of-00061.bin", "model.embed_tokens.weight": "pytorch_model-00060-of-00061.bin", "lm_head.weight": "pytorch_model-00060-of-00061.bin"}, "metadata": {"total_size": 65057894912}}
\ No newline at end of file