Upload 23 files

config.json

@@ -0,0 +1 @@
+{"architectures": ["LLaMAForCausalLM"], "bos_token_id": 0, "eos_token_id": 1, "hidden_act": "silu", "hidden_size": 8192, "intermediate_size": 22016, "initializer_range": 0.02, "max_sequence_length": 2048, "model_type": "llama", "num_attention_heads": 64, "num_hidden_layers": 80, "pad_token_id": -1, "rms_norm_eps": 1e-05, "torch_dtype": "float16", "transformers_version": "4.27.0.dev0", "use_cache": true, "vocab_size": 32000}

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"}

pytorch_model-00000-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0dbdcc12a6adc15533d5ecdfc9056b5ce731b83958229b90685855857453399a
+size 1619037618

pytorch_model-00001-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2618fd7849637364187d036a4991f34bfbe8a613878a36c2cc476701878fdbc1
+size 1619037618

pytorch_model-00002-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:58a5d43d24aff817f2ddee10debe7affb8a319ed3209881671f6d2fd25116d81
+size 1619037618

pytorch_model-00003-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3b7da52b3457436a40f2cd035129c1479bab3a9c57bc50c9262b4f62c86ec8c6
+size 1619037618

pytorch_model-00004-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e79d5a4236d2e9e28e2be453e7812daa5a1b0e92f10e8c95dbf4179caa2d719
+size 1619037618

pytorch_model-00005-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39d85463f8f73e5a86c4c8a42b78334202879ccd42da2a271da773ae09a961aa
+size 1619037618

pytorch_model-00006-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ad8991754366a3905fb3cd4039df29cc309d81c06fe16785861ee2d904618a94
+size 1619037618

pytorch_model-00007-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35b0217250ada9dfef6d076f207705e47a38c857a57de3b8725734fb0c4f96d1
+size 1619037618

pytorch_model-00008-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:400a39a54fba400d45c4de61cdaaaf81dfb650365c00214dc40ee337c99931f1
+size 1619037618

pytorch_model-00009-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fdff7f03a1c0d4f3390b92725a2bfb494e012c4c3ec2a39a2f84f78bd4f76c19
+size 1619037618

pytorch_model-00010-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec83c46a6d229791070d501337fb4f748ea6c8016303bf28bd1851a664c29efc
+size 1619037618

pytorch_model-00011-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc5b715e53966110464b874ed770a08657caac91c05d331007a37f8195693882
+size 1619037618

pytorch_model-00012-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81f8e0768d6ad7d65d12d2f9977d742cf021f9a996b6d7bdba2b671e7b0f05e5
+size 1619037618

pytorch_model-00013-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b45e63ac6b46e2804f6f81416d4666c65fe1c755ddccf14ad3e7bd49b13f3420
+size 1619037618

pytorch_model-00014-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bac544dad09325972659ff5d56cd8a09de8a7d16bcb3f5a53647a46d3c98416c
+size 1619037618

pytorch_model-00015-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9bd9df073c5bafd698e8186dffe8e785d06a7e9816583871a86d4c800336d490
+size 1619037618

pytorch_model-00016-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2909671118d834753416b1ba42fd7c61abb442041e0fe7a1bc4b0e812e032019
+size 1619037618

pytorch_model-00017-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7b972d918470cfd2a619c2f0c7bdd89cd10f58f8603ebd52465a3a62466bd97e
+size 1619037618

pytorch_model-00018-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:598289b912f1f756796d2b8536565bbd87f0eccb631c8a80f698e3891ecc8ce7
+size 1619037618

pytorch_model-00019-of-00081.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dacd5f866d32e9fcc7835ab352972e8195b26aa2deae19fc91e4092e45a9410a
+size 1619037618

pytorch_model.bin.index.json

@@ -0,0 +1 @@
+{"weight_map": {"model.decoder.layers.0.attention_norm.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.ffn_norm.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.self_attn.q_proj.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.self_attn.k_proj.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.self_attn.v_proj.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.self_attn.o_proj.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.feed_forward.w1.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.feed_forward.w2.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.0.feed_forward.w3.weight": "pytorch_model-00000-of-00081.bin", "model.decoder.layers.1.attention_norm.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.ffn_norm.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.feed_forward.w1.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.feed_forward.w2.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.1.feed_forward.w3.weight": "pytorch_model-00001-of-00081.bin", "model.decoder.layers.2.attention_norm.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.ffn_norm.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.self_attn.q_proj.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.self_attn.k_proj.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.self_attn.v_proj.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.self_attn.o_proj.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.feed_forward.w1.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.feed_forward.w2.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.2.feed_forward.w3.weight": "pytorch_model-00002-of-00081.bin", "model.decoder.layers.3.attention_norm.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.ffn_norm.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.self_attn.q_proj.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.self_attn.k_proj.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.self_attn.v_proj.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.self_attn.o_proj.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.feed_forward.w1.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.feed_forward.w2.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.3.feed_forward.w3.weight": "pytorch_model-00003-of-00081.bin", "model.decoder.layers.4.attention_norm.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.ffn_norm.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.self_attn.q_proj.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.self_attn.k_proj.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.self_attn.v_proj.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.self_attn.o_proj.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.feed_forward.w1.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.feed_forward.w2.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.4.feed_forward.w3.weight": "pytorch_model-00004-of-00081.bin", "model.decoder.layers.5.attention_norm.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.ffn_norm.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.self_attn.q_proj.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.self_attn.k_proj.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.self_attn.v_proj.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.self_attn.o_proj.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.feed_forward.w1.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.feed_forward.w2.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.5.feed_forward.w3.weight": "pytorch_model-00005-of-00081.bin", "model.decoder.layers.6.attention_norm.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.ffn_norm.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.self_attn.q_proj.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.self_attn.k_proj.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.self_attn.v_proj.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.self_attn.o_proj.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.feed_forward.w1.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.feed_forward.w2.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.6.feed_forward.w3.weight": "pytorch_model-00006-of-00081.bin", "model.decoder.layers.7.attention_norm.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.ffn_norm.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.self_attn.q_proj.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.self_attn.k_proj.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.self_attn.v_proj.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.self_attn.o_proj.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.feed_forward.w1.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.feed_forward.w2.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.7.feed_forward.w3.weight": "pytorch_model-00007-of-00081.bin", "model.decoder.layers.8.attention_norm.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.ffn_norm.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.self_attn.q_proj.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.self_attn.k_proj.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.self_attn.v_proj.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.self_attn.o_proj.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.feed_forward.w1.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.feed_forward.w2.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.8.feed_forward.w3.weight": "pytorch_model-00008-of-00081.bin", "model.decoder.layers.9.attention_norm.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.ffn_norm.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.self_attn.q_proj.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.self_attn.k_proj.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.self_attn.v_proj.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.self_attn.o_proj.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.feed_forward.w1.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.feed_forward.w2.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.9.feed_forward.w3.weight": "pytorch_model-00009-of-00081.bin", "model.decoder.layers.10.attention_norm.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.ffn_norm.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.self_attn.q_proj.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.self_attn.k_proj.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.self_attn.v_proj.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.self_attn.o_proj.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.feed_forward.w1.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.feed_forward.w2.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.10.feed_forward.w3.weight": "pytorch_model-00010-of-00081.bin", "model.decoder.layers.11.attention_norm.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.ffn_norm.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.self_attn.q_proj.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.self_attn.k_proj.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.self_attn.v_proj.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.self_attn.o_proj.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.feed_forward.w1.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.feed_forward.w2.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.11.feed_forward.w3.weight": "pytorch_model-00011-of-00081.bin", "model.decoder.layers.12.attention_norm.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.ffn_norm.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.self_attn.q_proj.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.self_attn.k_proj.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.self_attn.v_proj.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.self_attn.o_proj.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.feed_forward.w1.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.feed_forward.w2.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.12.feed_forward.w3.weight": "pytorch_model-00012-of-00081.bin", "model.decoder.layers.13.attention_norm.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.ffn_norm.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.self_attn.q_proj.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.self_attn.k_proj.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.self_attn.v_proj.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.self_attn.o_proj.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.feed_forward.w1.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.feed_forward.w2.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.13.feed_forward.w3.weight": "pytorch_model-00013-of-00081.bin", "model.decoder.layers.14.attention_norm.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.ffn_norm.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.self_attn.q_proj.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.self_attn.k_proj.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.self_attn.v_proj.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.self_attn.o_proj.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.feed_forward.w1.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.feed_forward.w2.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.14.feed_forward.w3.weight": "pytorch_model-00014-of-00081.bin", "model.decoder.layers.15.attention_norm.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.ffn_norm.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.self_attn.q_proj.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.self_attn.k_proj.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.self_attn.v_proj.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.self_attn.o_proj.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.feed_forward.w1.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.feed_forward.w2.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.15.feed_forward.w3.weight": "pytorch_model-00015-of-00081.bin", "model.decoder.layers.16.attention_norm.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.ffn_norm.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.self_attn.q_proj.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.self_attn.k_proj.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.self_attn.v_proj.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.self_attn.o_proj.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.feed_forward.w1.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.feed_forward.w2.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.16.feed_forward.w3.weight": "pytorch_model-00016-of-00081.bin", "model.decoder.layers.17.attention_norm.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.ffn_norm.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.self_attn.q_proj.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.self_attn.k_proj.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.self_attn.v_proj.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.self_attn.o_proj.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.feed_forward.w1.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.feed_forward.w2.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.17.feed_forward.w3.weight": "pytorch_model-00017-of-00081.bin", "model.decoder.layers.18.attention_norm.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.ffn_norm.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.self_attn.q_proj.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.self_attn.k_proj.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.self_attn.v_proj.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.self_attn.o_proj.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.feed_forward.w1.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.feed_forward.w2.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.18.feed_forward.w3.weight": "pytorch_model-00018-of-00081.bin", "model.decoder.layers.19.attention_norm.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.ffn_norm.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.self_attn.q_proj.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.self_attn.k_proj.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.self_attn.v_proj.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.self_attn.o_proj.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.feed_forward.w1.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.feed_forward.w2.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.19.feed_forward.w3.weight": "pytorch_model-00019-of-00081.bin", "model.decoder.layers.20.attention_norm.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.ffn_norm.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.self_attn.q_proj.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.self_attn.k_proj.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.self_attn.v_proj.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.self_attn.o_proj.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.feed_forward.w1.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.feed_forward.w2.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.20.feed_forward.w3.weight": "pytorch_model-00020-of-00081.bin", "model.decoder.layers.21.attention_norm.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.ffn_norm.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.self_attn.q_proj.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.self_attn.k_proj.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.self_attn.v_proj.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.self_attn.o_proj.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.feed_forward.w1.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.feed_forward.w2.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.21.feed_forward.w3.weight": "pytorch_model-00021-of-00081.bin", "model.decoder.layers.22.attention_norm.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.ffn_norm.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.self_attn.q_proj.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.self_attn.k_proj.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.self_attn.v_proj.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.self_attn.o_proj.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.feed_forward.w1.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.feed_forward.w2.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.22.feed_forward.w3.weight": "pytorch_model-00022-of-00081.bin", "model.decoder.layers.23.attention_norm.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.ffn_norm.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.self_attn.q_proj.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.self_attn.k_proj.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.self_attn.v_proj.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.self_attn.o_proj.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.feed_forward.w1.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.feed_forward.w2.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.23.feed_forward.w3.weight": "pytorch_model-00023-of-00081.bin", "model.decoder.layers.24.attention_norm.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.ffn_norm.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.self_attn.q_proj.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.self_attn.k_proj.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.self_attn.v_proj.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.self_attn.o_proj.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.feed_forward.w1.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.feed_forward.w2.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.24.feed_forward.w3.weight": "pytorch_model-00024-of-00081.bin", "model.decoder.layers.25.attention_norm.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.ffn_norm.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.self_attn.q_proj.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.self_attn.k_proj.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.self_attn.v_proj.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.self_attn.o_proj.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.feed_forward.w1.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.feed_forward.w2.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.25.feed_forward.w3.weight": "pytorch_model-00025-of-00081.bin", "model.decoder.layers.26.attention_norm.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.ffn_norm.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.self_attn.q_proj.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.self_attn.k_proj.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.self_attn.v_proj.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.self_attn.o_proj.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.feed_forward.w1.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.feed_forward.w2.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.26.feed_forward.w3.weight": "pytorch_model-00026-of-00081.bin", "model.decoder.layers.27.attention_norm.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.ffn_norm.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.self_attn.q_proj.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.self_attn.k_proj.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.self_attn.v_proj.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.self_attn.o_proj.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.feed_forward.w1.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.feed_forward.w2.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.27.feed_forward.w3.weight": "pytorch_model-00027-of-00081.bin", "model.decoder.layers.28.attention_norm.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.ffn_norm.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.self_attn.q_proj.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.self_attn.k_proj.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.self_attn.v_proj.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.self_attn.o_proj.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.feed_forward.w1.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.feed_forward.w2.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.28.feed_forward.w3.weight": "pytorch_model-00028-of-00081.bin", "model.decoder.layers.29.attention_norm.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.ffn_norm.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.self_attn.q_proj.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.self_attn.k_proj.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.self_attn.v_proj.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.self_attn.o_proj.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.feed_forward.w1.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.feed_forward.w2.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.29.feed_forward.w3.weight": "pytorch_model-00029-of-00081.bin", "model.decoder.layers.30.attention_norm.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.ffn_norm.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.self_attn.q_proj.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.self_attn.k_proj.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.self_attn.v_proj.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.self_attn.o_proj.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.feed_forward.w1.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.feed_forward.w2.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.30.feed_forward.w3.weight": "pytorch_model-00030-of-00081.bin", "model.decoder.layers.31.attention_norm.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.ffn_norm.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.self_attn.q_proj.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.self_attn.k_proj.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.self_attn.v_proj.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.self_attn.o_proj.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.feed_forward.w1.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.feed_forward.w2.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.31.feed_forward.w3.weight": "pytorch_model-00031-of-00081.bin", "model.decoder.layers.32.attention_norm.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.ffn_norm.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.self_attn.q_proj.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.self_attn.k_proj.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.self_attn.v_proj.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.self_attn.o_proj.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.feed_forward.w1.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.feed_forward.w2.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.32.feed_forward.w3.weight": "pytorch_model-00032-of-00081.bin", "model.decoder.layers.33.attention_norm.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.ffn_norm.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.self_attn.q_proj.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.self_attn.k_proj.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.self_attn.v_proj.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.self_attn.o_proj.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.feed_forward.w1.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.feed_forward.w2.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.33.feed_forward.w3.weight": "pytorch_model-00033-of-00081.bin", "model.decoder.layers.34.attention_norm.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.ffn_norm.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.self_attn.q_proj.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.self_attn.k_proj.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.self_attn.v_proj.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.self_attn.o_proj.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.feed_forward.w1.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.feed_forward.w2.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.34.feed_forward.w3.weight": "pytorch_model-00034-of-00081.bin", "model.decoder.layers.35.attention_norm.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.ffn_norm.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.self_attn.q_proj.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.self_attn.k_proj.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.self_attn.v_proj.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.self_attn.o_proj.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.feed_forward.w1.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.feed_forward.w2.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.35.feed_forward.w3.weight": "pytorch_model-00035-of-00081.bin", "model.decoder.layers.36.attention_norm.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.ffn_norm.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.self_attn.q_proj.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.self_attn.k_proj.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.self_attn.v_proj.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.self_attn.o_proj.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.feed_forward.w1.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.feed_forward.w2.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.36.feed_forward.w3.weight": "pytorch_model-00036-of-00081.bin", "model.decoder.layers.37.attention_norm.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.ffn_norm.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.self_attn.q_proj.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.self_attn.k_proj.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.self_attn.v_proj.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.self_attn.o_proj.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.feed_forward.w1.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.feed_forward.w2.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.37.feed_forward.w3.weight": "pytorch_model-00037-of-00081.bin", "model.decoder.layers.38.attention_norm.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.ffn_norm.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.self_attn.q_proj.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.self_attn.k_proj.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.self_attn.v_proj.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.self_attn.o_proj.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.feed_forward.w1.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.feed_forward.w2.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.38.feed_forward.w3.weight": "pytorch_model-00038-of-00081.bin", "model.decoder.layers.39.attention_norm.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.ffn_norm.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.self_attn.q_proj.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.self_attn.k_proj.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.self_attn.v_proj.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.self_attn.o_proj.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.feed_forward.w1.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.feed_forward.w2.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.39.feed_forward.w3.weight": "pytorch_model-00039-of-00081.bin", "model.decoder.layers.40.attention_norm.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.ffn_norm.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.self_attn.q_proj.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.self_attn.k_proj.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.self_attn.v_proj.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.self_attn.o_proj.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.feed_forward.w1.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.feed_forward.w2.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.40.feed_forward.w3.weight": "pytorch_model-00040-of-00081.bin", "model.decoder.layers.41.attention_norm.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.ffn_norm.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.self_attn.q_proj.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.self_attn.k_proj.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.self_attn.v_proj.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.self_attn.o_proj.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.feed_forward.w1.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.feed_forward.w2.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.41.feed_forward.w3.weight": "pytorch_model-00041-of-00081.bin", "model.decoder.layers.42.attention_norm.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.ffn_norm.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.self_attn.q_proj.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.self_attn.k_proj.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.self_attn.v_proj.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.self_attn.o_proj.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.feed_forward.w1.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.feed_forward.w2.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.42.feed_forward.w3.weight": "pytorch_model-00042-of-00081.bin", "model.decoder.layers.43.attention_norm.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.ffn_norm.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.self_attn.q_proj.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.self_attn.k_proj.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.self_attn.v_proj.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.self_attn.o_proj.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.feed_forward.w1.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.feed_forward.w2.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.43.feed_forward.w3.weight": "pytorch_model-00043-of-00081.bin", "model.decoder.layers.44.attention_norm.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.ffn_norm.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.self_attn.q_proj.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.self_attn.k_proj.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.self_attn.v_proj.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.self_attn.o_proj.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.feed_forward.w1.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.feed_forward.w2.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.44.feed_forward.w3.weight": "pytorch_model-00044-of-00081.bin", "model.decoder.layers.45.attention_norm.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.ffn_norm.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.self_attn.q_proj.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.self_attn.k_proj.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.self_attn.v_proj.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.self_attn.o_proj.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.feed_forward.w1.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.feed_forward.w2.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.45.feed_forward.w3.weight": "pytorch_model-00045-of-00081.bin", "model.decoder.layers.46.attention_norm.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.ffn_norm.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.self_attn.q_proj.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.self_attn.k_proj.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.self_attn.v_proj.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.self_attn.o_proj.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.feed_forward.w1.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.feed_forward.w2.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.46.feed_forward.w3.weight": "pytorch_model-00046-of-00081.bin", "model.decoder.layers.47.attention_norm.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.ffn_norm.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.self_attn.q_proj.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.self_attn.k_proj.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.self_attn.v_proj.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.self_attn.o_proj.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.feed_forward.w1.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.feed_forward.w2.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.47.feed_forward.w3.weight": "pytorch_model-00047-of-00081.bin", "model.decoder.layers.48.attention_norm.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.ffn_norm.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.self_attn.q_proj.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.self_attn.k_proj.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.self_attn.v_proj.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.self_attn.o_proj.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.feed_forward.w1.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.feed_forward.w2.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.48.feed_forward.w3.weight": "pytorch_model-00048-of-00081.bin", "model.decoder.layers.49.attention_norm.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.ffn_norm.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.self_attn.q_proj.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.self_attn.k_proj.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.self_attn.v_proj.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.self_attn.o_proj.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.feed_forward.w1.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.feed_forward.w2.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.49.feed_forward.w3.weight": "pytorch_model-00049-of-00081.bin", "model.decoder.layers.50.attention_norm.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.ffn_norm.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.self_attn.q_proj.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.self_attn.k_proj.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.self_attn.v_proj.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.self_attn.o_proj.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.feed_forward.w1.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.feed_forward.w2.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.50.feed_forward.w3.weight": "pytorch_model-00050-of-00081.bin", "model.decoder.layers.51.attention_norm.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.ffn_norm.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.self_attn.q_proj.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.self_attn.k_proj.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.self_attn.v_proj.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.self_attn.o_proj.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.feed_forward.w1.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.feed_forward.w2.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.51.feed_forward.w3.weight": "pytorch_model-00051-of-00081.bin", "model.decoder.layers.52.attention_norm.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.ffn_norm.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.self_attn.q_proj.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.self_attn.k_proj.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.self_attn.v_proj.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.self_attn.o_proj.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.feed_forward.w1.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.feed_forward.w2.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.52.feed_forward.w3.weight": "pytorch_model-00052-of-00081.bin", "model.decoder.layers.53.attention_norm.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.ffn_norm.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.self_attn.q_proj.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.self_attn.k_proj.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.self_attn.v_proj.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.self_attn.o_proj.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.feed_forward.w1.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.feed_forward.w2.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.53.feed_forward.w3.weight": "pytorch_model-00053-of-00081.bin", "model.decoder.layers.54.attention_norm.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.ffn_norm.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.self_attn.q_proj.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.self_attn.k_proj.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.self_attn.v_proj.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.self_attn.o_proj.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.feed_forward.w1.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.feed_forward.w2.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.54.feed_forward.w3.weight": "pytorch_model-00054-of-00081.bin", "model.decoder.layers.55.attention_norm.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.ffn_norm.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.self_attn.q_proj.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.self_attn.k_proj.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.self_attn.v_proj.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.self_attn.o_proj.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.feed_forward.w1.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.feed_forward.w2.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.55.feed_forward.w3.weight": "pytorch_model-00055-of-00081.bin", "model.decoder.layers.56.attention_norm.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.ffn_norm.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.self_attn.q_proj.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.self_attn.k_proj.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.self_attn.v_proj.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.self_attn.o_proj.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.feed_forward.w1.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.feed_forward.w2.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.56.feed_forward.w3.weight": "pytorch_model-00056-of-00081.bin", "model.decoder.layers.57.attention_norm.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.ffn_norm.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.self_attn.q_proj.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.self_attn.k_proj.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.self_attn.v_proj.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.self_attn.o_proj.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.feed_forward.w1.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.feed_forward.w2.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.57.feed_forward.w3.weight": "pytorch_model-00057-of-00081.bin", "model.decoder.layers.58.attention_norm.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.ffn_norm.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.self_attn.q_proj.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.self_attn.k_proj.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.self_attn.v_proj.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.self_attn.o_proj.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.feed_forward.w1.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.feed_forward.w2.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.58.feed_forward.w3.weight": "pytorch_model-00058-of-00081.bin", "model.decoder.layers.59.attention_norm.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.ffn_norm.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.self_attn.q_proj.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.self_attn.k_proj.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.self_attn.v_proj.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.self_attn.o_proj.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.feed_forward.w1.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.feed_forward.w2.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.59.feed_forward.w3.weight": "pytorch_model-00059-of-00081.bin", "model.decoder.layers.60.attention_norm.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.ffn_norm.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.self_attn.q_proj.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.self_attn.k_proj.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.self_attn.v_proj.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.self_attn.o_proj.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.feed_forward.w1.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.feed_forward.w2.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.60.feed_forward.w3.weight": "pytorch_model-00060-of-00081.bin", "model.decoder.layers.61.attention_norm.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.ffn_norm.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.self_attn.q_proj.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.self_attn.k_proj.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.self_attn.v_proj.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.self_attn.o_proj.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.feed_forward.w1.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.feed_forward.w2.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.61.feed_forward.w3.weight": "pytorch_model-00061-of-00081.bin", "model.decoder.layers.62.attention_norm.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.ffn_norm.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.self_attn.q_proj.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.self_attn.k_proj.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.self_attn.v_proj.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.self_attn.o_proj.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.feed_forward.w1.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.feed_forward.w2.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.62.feed_forward.w3.weight": "pytorch_model-00062-of-00081.bin", "model.decoder.layers.63.attention_norm.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.ffn_norm.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.self_attn.q_proj.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.self_attn.k_proj.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.self_attn.v_proj.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.self_attn.o_proj.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.feed_forward.w1.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.feed_forward.w2.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.63.feed_forward.w3.weight": "pytorch_model-00063-of-00081.bin", "model.decoder.layers.64.attention_norm.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.ffn_norm.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.self_attn.q_proj.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.self_attn.k_proj.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.self_attn.v_proj.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.self_attn.o_proj.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.feed_forward.w1.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.feed_forward.w2.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.64.feed_forward.w3.weight": "pytorch_model-00064-of-00081.bin", "model.decoder.layers.65.attention_norm.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.ffn_norm.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.self_attn.q_proj.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.self_attn.k_proj.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.self_attn.v_proj.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.self_attn.o_proj.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.feed_forward.w1.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.feed_forward.w2.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.65.feed_forward.w3.weight": "pytorch_model-00065-of-00081.bin", "model.decoder.layers.66.attention_norm.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.ffn_norm.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.self_attn.q_proj.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.self_attn.k_proj.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.self_attn.v_proj.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.self_attn.o_proj.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.feed_forward.w1.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.feed_forward.w2.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.66.feed_forward.w3.weight": "pytorch_model-00066-of-00081.bin", "model.decoder.layers.67.attention_norm.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.ffn_norm.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.self_attn.q_proj.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.self_attn.k_proj.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.self_attn.v_proj.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.self_attn.o_proj.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.feed_forward.w1.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.feed_forward.w2.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.67.feed_forward.w3.weight": "pytorch_model-00067-of-00081.bin", "model.decoder.layers.68.attention_norm.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.ffn_norm.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.self_attn.q_proj.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.self_attn.k_proj.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.self_attn.v_proj.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.self_attn.o_proj.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.feed_forward.w1.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.feed_forward.w2.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.68.feed_forward.w3.weight": "pytorch_model-00068-of-00081.bin", "model.decoder.layers.69.attention_norm.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.ffn_norm.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.self_attn.q_proj.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.self_attn.k_proj.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.self_attn.v_proj.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.self_attn.o_proj.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.feed_forward.w1.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.feed_forward.w2.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.69.feed_forward.w3.weight": "pytorch_model-00069-of-00081.bin", "model.decoder.layers.70.attention_norm.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.ffn_norm.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.self_attn.q_proj.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.self_attn.k_proj.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.self_attn.v_proj.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.self_attn.o_proj.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.feed_forward.w1.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.feed_forward.w2.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.70.feed_forward.w3.weight": "pytorch_model-00070-of-00081.bin", "model.decoder.layers.71.attention_norm.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.ffn_norm.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.self_attn.q_proj.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.self_attn.k_proj.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.self_attn.v_proj.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.self_attn.o_proj.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.feed_forward.w1.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.feed_forward.w2.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.71.feed_forward.w3.weight": "pytorch_model-00071-of-00081.bin", "model.decoder.layers.72.attention_norm.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.ffn_norm.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.self_attn.q_proj.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.self_attn.k_proj.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.self_attn.v_proj.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.self_attn.o_proj.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.feed_forward.w1.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.feed_forward.w2.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.72.feed_forward.w3.weight": "pytorch_model-00072-of-00081.bin", "model.decoder.layers.73.attention_norm.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.ffn_norm.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.self_attn.q_proj.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.self_attn.k_proj.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.self_attn.v_proj.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.self_attn.o_proj.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.feed_forward.w1.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.feed_forward.w2.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.73.feed_forward.w3.weight": "pytorch_model-00073-of-00081.bin", "model.decoder.layers.74.attention_norm.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.ffn_norm.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.self_attn.q_proj.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.self_attn.k_proj.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.self_attn.v_proj.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.self_attn.o_proj.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.feed_forward.w1.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.feed_forward.w2.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.74.feed_forward.w3.weight": "pytorch_model-00074-of-00081.bin", "model.decoder.layers.75.attention_norm.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.ffn_norm.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.self_attn.q_proj.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.self_attn.k_proj.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.self_attn.v_proj.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.self_attn.o_proj.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.feed_forward.w1.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.feed_forward.w2.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.75.feed_forward.w3.weight": "pytorch_model-00075-of-00081.bin", "model.decoder.layers.76.attention_norm.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.ffn_norm.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.self_attn.q_proj.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.self_attn.k_proj.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.self_attn.v_proj.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.self_attn.o_proj.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.feed_forward.w1.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.feed_forward.w2.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.76.feed_forward.w3.weight": "pytorch_model-00076-of-00081.bin", "model.decoder.layers.77.attention_norm.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.ffn_norm.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.self_attn.q_proj.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.self_attn.k_proj.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.self_attn.v_proj.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.self_attn.o_proj.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.feed_forward.w1.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.feed_forward.w2.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.77.feed_forward.w3.weight": "pytorch_model-00077-of-00081.bin", "model.decoder.layers.78.attention_norm.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.ffn_norm.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.self_attn.q_proj.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.self_attn.k_proj.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.self_attn.v_proj.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.self_attn.o_proj.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.feed_forward.w1.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.feed_forward.w2.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.78.feed_forward.w3.weight": "pytorch_model-00078-of-00081.bin", "model.decoder.layers.79.attention_norm.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.ffn_norm.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.self_attn.q_proj.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.self_attn.k_proj.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.self_attn.v_proj.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.self_attn.o_proj.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.feed_forward.w1.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.feed_forward.w2.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.layers.79.feed_forward.w3.weight": "pytorch_model-00079-of-00081.bin", "model.decoder.norm.weight": "pytorch_model-00080-of-00081.bin", "model.decoder.embed_tokens.weight": "pytorch_model-00080-of-00081.bin", "lm_head.weight": "pytorch_model-00080-of-00081.bin"}, "metadata": {"total_size": 130571321344}}