Upload 38 files

config.json

@@ -0,0 +1,18 @@
+{
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "hidden_act": "silu",
+  "hidden_size": 4096,
+  "initializer_range": 0.02,
+  "intermediate_size": 11008,
+  "max_position_embeddings": 2048,
+  "model_type": "llama",
+  "num_attention_heads": 32,
+  "num_hidden_layers": 32,
+  "pad_token_id": 0,
+  "rms_norm_eps": 1e-06,
+  "tie_word_embeddings": false,
+  "transformers_version": "4.28.1",
+  "use_cache": false,
+  "vocab_size": 46943
+}

pytorch_model-1-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40f7cd8109b160c3b40aa3d2a593962cbf604fd887524781d73bd7b9bbede46b
+size 404769799

pytorch_model-10-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2562e5f3317c2b2f651c9d38fc8836e4d36176c8087fe1b1d948d31c7d4e448b
+size 404769799

pytorch_model-11-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d3efbb95e9023308dcfc56875fcc4ad81ce1e34e679549042d36a4f26e17337f
+size 404769799

pytorch_model-12-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:057c38e4273af35974edb1ab3db6b97ecbd8a3d0fd6163804a124de9f21c436f
+size 404769799

pytorch_model-13-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d49010711ca14810b028429cce5f68a48eda454e8d013eaaca002cc868bb432
+size 404769799

pytorch_model-14-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f59004cf6a6a2aa953b65b12644383bbbc49b00a575e18dd733dffd272be4ae6
+size 404769799

pytorch_model-15-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:45e1b59ddd0e58f09833d73222de3a3686c6cff8051cf79e1ca9e88f0ac57e68
+size 404769799

pytorch_model-16-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f1732abaa311b425950bda0914804c56a57403216058b9ae79ca09afd570e22
+size 404769799

pytorch_model-17-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a471941b4a37120a9605b94d6c05b6dca68aec48c17578b6107bbb4342d08241
+size 404769799

pytorch_model-18-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8a3edc7e234004c0dc4c5b9ca9ded2111def172ec6e887a4939cbd9951d66e63
+size 404769799

pytorch_model-19-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3574a85973fdc2e544649bdec55af106184bc066f3ea11eb5b32940d886aded
+size 404769799

pytorch_model-2-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0d2db8fa9b44b5bbf3d6f593b387ee025c34fccfc9ab3f4a182ea7646ccd8828
+size 404769799

pytorch_model-20-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92c41443defcf3ede95e71a53d047706bd5fe1242845e74664ad15020846d57d
+size 404769799

pytorch_model-21-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53137e9508f7a14c60594f99d429ba66fb7fbd4a4fe5bf75668af956a20a66d3
+size 404769799

pytorch_model-22-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8b0363260aeec9f4cd070bf652b34a1d55b536497f0376ea9bba24eb7fb1f28d
+size 404769799

pytorch_model-23-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:71f633017ed837373a866ca829cf9c8f51c2e5fcd88286e20d3713496917a639
+size 404769799

pytorch_model-24-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6eacbfe28e5069c64671ffd9efb240877d0ca571879a21fa3c219e2c1ecd5eac
+size 404769799

pytorch_model-25-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cdbe9eebaaa1c0a9151b33d010e3e8d147ed7ba911054a8a3a54e4d174d63ba7
+size 404769799

pytorch_model-26-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c2cd09fe255fb5d9ba21f06b4c86e66ed69d9e972242dbe71cdd2d0f48160476
+size 404769799

pytorch_model-27-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ceb2e3924ced9e7f461dfa519c463be3770b47a76ee91702f56ec11c1ce8035
+size 404769799

pytorch_model-28-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e75fb0746156f72fa4ed15be3e770f776fc90a9822e957905c5abd1953b233a6
+size 404769799

pytorch_model-29-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f46fccd70c13e9a7b2d4fa9cd26982576f0ab79df001436acd77a7e2860837f9
+size 404769799

pytorch_model-3-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b4480fab7d69c72f4de4f62e28081377dcd06fd1bb367a753a78dde87c72e485
+size 404769799

pytorch_model-30-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0bb12f1dd18b36518799966a7ba0aec42b2548f9d890b641658cdc3a6d5b4dce
+size 404769799

pytorch_model-31-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:09c297ed271d77cf6981df7b87361067f93787f3fb2cbcc2abf87e02bb169400
+size 404769799

pytorch_model-32-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:37b3d83fcda5e5416b074fdb87d2932198c8d5400505da012fd29ed245e138de
+size 404769799

pytorch_model-33-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a52449667509c95acc699e2845a0aa8388ba331fbf73ba6ab5d8124c0edfe249
+size 1153680547

pytorch_model-4-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bf35b03760212fed7c619bbcc7c11fe7a7d07b7980478ad7e172b77c48686f8b
+size 404769799

pytorch_model-5-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1e955572da54bad0d2cb0e39811c493441e37672163e6743ed6f93591c34ad4
+size 404769799

pytorch_model-6-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f6b788b98c79f61e12498b576cb259b0a5a055db8aab27b80453f5d04cc07d24
+size 404769799

pytorch_model-7-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5b4e574ec82c3616e0a0e6d4c5a8d7de9ac58180b664bdd8388a7ac28cf3f641
+size 404769799

pytorch_model-8-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e48db2ef850ef25a9b7eab7b5989d166fad2921d6ebb6ca1de6892e3574ff72
+size 404769799

pytorch_model-9-of-33.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:17bb7f29264dc87a7703808ffdf843156562371a3bdf43a52816181b80558ed5
+size 404769799

pytorch_model.bin.index.json

@@ -0,0 +1 @@
+{"weight_map": {"model.layers.0.self_attn.q_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.self_attn.k_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.self_attn.v_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.self_attn.o_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.mlp.gate_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.mlp.down_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.mlp.up_proj.weight": "pytorch_model-1-of-33.bin", "model.layers.0.input_layernorm.weight": "pytorch_model-1-of-33.bin", "model.layers.0.post_attention_layernorm.weight": "pytorch_model-1-of-33.bin", "model.layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-1-of-33.bin", "model.layers.1.self_attn.q_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.self_attn.k_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.self_attn.v_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.self_attn.o_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.mlp.gate_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.mlp.down_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.mlp.up_proj.weight": "pytorch_model-2-of-33.bin", "model.layers.1.input_layernorm.weight": "pytorch_model-2-of-33.bin", "model.layers.1.post_attention_layernorm.weight": "pytorch_model-2-of-33.bin", "model.layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-2-of-33.bin", "model.layers.2.self_attn.q_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.self_attn.k_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.self_attn.v_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.self_attn.o_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.mlp.gate_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.mlp.down_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.mlp.up_proj.weight": "pytorch_model-3-of-33.bin", "model.layers.2.input_layernorm.weight": "pytorch_model-3-of-33.bin", "model.layers.2.post_attention_layernorm.weight": "pytorch_model-3-of-33.bin", "model.layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-3-of-33.bin", "model.layers.3.self_attn.q_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.self_attn.k_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.self_attn.v_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.self_attn.o_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.mlp.gate_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.mlp.down_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.mlp.up_proj.weight": "pytorch_model-4-of-33.bin", "model.layers.3.input_layernorm.weight": "pytorch_model-4-of-33.bin", "model.layers.3.post_attention_layernorm.weight": "pytorch_model-4-of-33.bin", "model.layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-4-of-33.bin", "model.layers.4.self_attn.q_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.self_attn.k_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.self_attn.v_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.self_attn.o_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.mlp.gate_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.mlp.down_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.mlp.up_proj.weight": "pytorch_model-5-of-33.bin", "model.layers.4.input_layernorm.weight": "pytorch_model-5-of-33.bin", "model.layers.4.post_attention_layernorm.weight": "pytorch_model-5-of-33.bin", "model.layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-5-of-33.bin", "model.layers.5.self_attn.q_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.self_attn.k_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.self_attn.v_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.self_attn.o_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.mlp.gate_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.mlp.down_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.mlp.up_proj.weight": "pytorch_model-6-of-33.bin", "model.layers.5.input_layernorm.weight": "pytorch_model-6-of-33.bin", "model.layers.5.post_attention_layernorm.weight": "pytorch_model-6-of-33.bin", "model.layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-6-of-33.bin", "model.layers.6.self_attn.q_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.self_attn.k_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.self_attn.v_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.self_attn.o_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.mlp.gate_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.mlp.down_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.mlp.up_proj.weight": "pytorch_model-7-of-33.bin", "model.layers.6.input_layernorm.weight": "pytorch_model-7-of-33.bin", "model.layers.6.post_attention_layernorm.weight": "pytorch_model-7-of-33.bin", "model.layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-7-of-33.bin", "model.layers.7.self_attn.q_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.self_attn.k_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.self_attn.v_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.self_attn.o_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.mlp.gate_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.mlp.down_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.mlp.up_proj.weight": "pytorch_model-8-of-33.bin", "model.layers.7.input_layernorm.weight": "pytorch_model-8-of-33.bin", "model.layers.7.post_attention_layernorm.weight": "pytorch_model-8-of-33.bin", "model.layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-8-of-33.bin", "model.layers.8.self_attn.q_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.self_attn.k_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.self_attn.v_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.self_attn.o_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.mlp.gate_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.mlp.down_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.mlp.up_proj.weight": "pytorch_model-9-of-33.bin", "model.layers.8.input_layernorm.weight": "pytorch_model-9-of-33.bin", "model.layers.8.post_attention_layernorm.weight": "pytorch_model-9-of-33.bin", "model.layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-9-of-33.bin", "model.layers.9.self_attn.q_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.self_attn.k_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.self_attn.v_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.self_attn.o_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.mlp.gate_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.mlp.down_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.mlp.up_proj.weight": "pytorch_model-10-of-33.bin", "model.layers.9.input_layernorm.weight": "pytorch_model-10-of-33.bin", "model.layers.9.post_attention_layernorm.weight": "pytorch_model-10-of-33.bin", "model.layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-10-of-33.bin", "model.layers.10.self_attn.q_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.self_attn.k_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.self_attn.v_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.self_attn.o_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.mlp.gate_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.mlp.down_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.mlp.up_proj.weight": "pytorch_model-11-of-33.bin", "model.layers.10.input_layernorm.weight": "pytorch_model-11-of-33.bin", "model.layers.10.post_attention_layernorm.weight": "pytorch_model-11-of-33.bin", "model.layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-11-of-33.bin", "model.layers.11.self_attn.q_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.self_attn.k_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.self_attn.v_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.self_attn.o_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.mlp.gate_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.mlp.down_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.mlp.up_proj.weight": "pytorch_model-12-of-33.bin", "model.layers.11.input_layernorm.weight": "pytorch_model-12-of-33.bin", "model.layers.11.post_attention_layernorm.weight": "pytorch_model-12-of-33.bin", "model.layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-12-of-33.bin", "model.layers.12.self_attn.q_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.self_attn.k_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.self_attn.v_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.self_attn.o_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.mlp.gate_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.mlp.down_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.mlp.up_proj.weight": "pytorch_model-13-of-33.bin", "model.layers.12.input_layernorm.weight": "pytorch_model-13-of-33.bin", "model.layers.12.post_attention_layernorm.weight": "pytorch_model-13-of-33.bin", "model.layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-13-of-33.bin", "model.layers.13.self_attn.q_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.self_attn.k_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.self_attn.v_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.self_attn.o_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.mlp.gate_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.mlp.down_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.mlp.up_proj.weight": "pytorch_model-14-of-33.bin", "model.layers.13.input_layernorm.weight": "pytorch_model-14-of-33.bin", "model.layers.13.post_attention_layernorm.weight": "pytorch_model-14-of-33.bin", "model.layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-14-of-33.bin", "model.layers.14.self_attn.q_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.self_attn.k_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.self_attn.v_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.self_attn.o_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.mlp.gate_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.mlp.down_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.mlp.up_proj.weight": "pytorch_model-15-of-33.bin", "model.layers.14.input_layernorm.weight": "pytorch_model-15-of-33.bin", "model.layers.14.post_attention_layernorm.weight": "pytorch_model-15-of-33.bin", "model.layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-15-of-33.bin", "model.layers.15.self_attn.q_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.self_attn.k_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.self_attn.v_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.self_attn.o_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.mlp.gate_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.mlp.down_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.mlp.up_proj.weight": "pytorch_model-16-of-33.bin", "model.layers.15.input_layernorm.weight": "pytorch_model-16-of-33.bin", "model.layers.15.post_attention_layernorm.weight": "pytorch_model-16-of-33.bin", "model.layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-16-of-33.bin", "model.layers.16.self_attn.q_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.self_attn.k_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.self_attn.v_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.self_attn.o_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.mlp.gate_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.mlp.down_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.mlp.up_proj.weight": "pytorch_model-17-of-33.bin", "model.layers.16.input_layernorm.weight": "pytorch_model-17-of-33.bin", "model.layers.16.post_attention_layernorm.weight": "pytorch_model-17-of-33.bin", "model.layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-17-of-33.bin", "model.layers.17.self_attn.q_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.self_attn.k_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.self_attn.v_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.self_attn.o_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.mlp.gate_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.mlp.down_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.mlp.up_proj.weight": "pytorch_model-18-of-33.bin", "model.layers.17.input_layernorm.weight": "pytorch_model-18-of-33.bin", "model.layers.17.post_attention_layernorm.weight": "pytorch_model-18-of-33.bin", "model.layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-18-of-33.bin", "model.layers.18.self_attn.q_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.self_attn.k_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.self_attn.v_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.self_attn.o_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.mlp.gate_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.mlp.down_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.mlp.up_proj.weight": "pytorch_model-19-of-33.bin", "model.layers.18.input_layernorm.weight": "pytorch_model-19-of-33.bin", "model.layers.18.post_attention_layernorm.weight": "pytorch_model-19-of-33.bin", "model.layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-19-of-33.bin", "model.layers.19.self_attn.q_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.self_attn.k_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.self_attn.v_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.self_attn.o_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.mlp.gate_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.mlp.down_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.mlp.up_proj.weight": "pytorch_model-20-of-33.bin", "model.layers.19.input_layernorm.weight": "pytorch_model-20-of-33.bin", "model.layers.19.post_attention_layernorm.weight": "pytorch_model-20-of-33.bin", "model.layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-20-of-33.bin", "model.layers.20.self_attn.q_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.self_attn.k_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.self_attn.v_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.self_attn.o_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.mlp.gate_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.mlp.down_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.mlp.up_proj.weight": "pytorch_model-21-of-33.bin", "model.layers.20.input_layernorm.weight": "pytorch_model-21-of-33.bin", "model.layers.20.post_attention_layernorm.weight": "pytorch_model-21-of-33.bin", "model.layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-21-of-33.bin", "model.layers.21.self_attn.q_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.self_attn.k_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.self_attn.v_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.self_attn.o_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.mlp.gate_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.mlp.down_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.mlp.up_proj.weight": "pytorch_model-22-of-33.bin", "model.layers.21.input_layernorm.weight": "pytorch_model-22-of-33.bin", "model.layers.21.post_attention_layernorm.weight": "pytorch_model-22-of-33.bin", "model.layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-22-of-33.bin", "model.layers.22.self_attn.q_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.self_attn.k_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.self_attn.v_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.self_attn.o_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.mlp.gate_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.mlp.down_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.mlp.up_proj.weight": "pytorch_model-23-of-33.bin", "model.layers.22.input_layernorm.weight": "pytorch_model-23-of-33.bin", "model.layers.22.post_attention_layernorm.weight": "pytorch_model-23-of-33.bin", "model.layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-23-of-33.bin", "model.layers.23.self_attn.q_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.self_attn.k_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.self_attn.v_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.self_attn.o_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.mlp.gate_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.mlp.down_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.mlp.up_proj.weight": "pytorch_model-24-of-33.bin", "model.layers.23.input_layernorm.weight": "pytorch_model-24-of-33.bin", "model.layers.23.post_attention_layernorm.weight": "pytorch_model-24-of-33.bin", "model.layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-24-of-33.bin", "model.layers.24.self_attn.q_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.self_attn.k_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.self_attn.v_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.self_attn.o_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.mlp.gate_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.mlp.down_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.mlp.up_proj.weight": "pytorch_model-25-of-33.bin", "model.layers.24.input_layernorm.weight": "pytorch_model-25-of-33.bin", "model.layers.24.post_attention_layernorm.weight": "pytorch_model-25-of-33.bin", "model.layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-25-of-33.bin", "model.layers.25.self_attn.q_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.self_attn.k_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.self_attn.v_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.self_attn.o_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.mlp.gate_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.mlp.down_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.mlp.up_proj.weight": "pytorch_model-26-of-33.bin", "model.layers.25.input_layernorm.weight": "pytorch_model-26-of-33.bin", "model.layers.25.post_attention_layernorm.weight": "pytorch_model-26-of-33.bin", "model.layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-26-of-33.bin", "model.layers.26.self_attn.q_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.self_attn.k_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.self_attn.v_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.self_attn.o_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.mlp.gate_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.mlp.down_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.mlp.up_proj.weight": "pytorch_model-27-of-33.bin", "model.layers.26.input_layernorm.weight": "pytorch_model-27-of-33.bin", "model.layers.26.post_attention_layernorm.weight": "pytorch_model-27-of-33.bin", "model.layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-27-of-33.bin", "model.layers.27.self_attn.q_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.self_attn.k_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.self_attn.v_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.self_attn.o_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.mlp.gate_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.mlp.down_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.mlp.up_proj.weight": "pytorch_model-28-of-33.bin", "model.layers.27.input_layernorm.weight": "pytorch_model-28-of-33.bin", "model.layers.27.post_attention_layernorm.weight": "pytorch_model-28-of-33.bin", "model.layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-28-of-33.bin", "model.layers.28.self_attn.q_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.self_attn.k_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.self_attn.v_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.self_attn.o_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.mlp.gate_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.mlp.down_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.mlp.up_proj.weight": "pytorch_model-29-of-33.bin", "model.layers.28.input_layernorm.weight": "pytorch_model-29-of-33.bin", "model.layers.28.post_attention_layernorm.weight": "pytorch_model-29-of-33.bin", "model.layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-29-of-33.bin", "model.layers.29.self_attn.q_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.self_attn.k_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.self_attn.v_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.self_attn.o_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.mlp.gate_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.mlp.down_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.mlp.up_proj.weight": "pytorch_model-30-of-33.bin", "model.layers.29.input_layernorm.weight": "pytorch_model-30-of-33.bin", "model.layers.29.post_attention_layernorm.weight": "pytorch_model-30-of-33.bin", "model.layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-30-of-33.bin", "model.layers.30.self_attn.q_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.self_attn.k_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.self_attn.v_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.self_attn.o_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.mlp.gate_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.mlp.down_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.mlp.up_proj.weight": "pytorch_model-31-of-33.bin", "model.layers.30.input_layernorm.weight": "pytorch_model-31-of-33.bin", "model.layers.30.post_attention_layernorm.weight": "pytorch_model-31-of-33.bin", "model.layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-31-of-33.bin", "model.layers.31.self_attn.q_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.self_attn.k_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.self_attn.v_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.self_attn.o_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.mlp.gate_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.mlp.down_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.mlp.up_proj.weight": "pytorch_model-32-of-33.bin", "model.layers.31.input_layernorm.weight": "pytorch_model-32-of-33.bin", "model.layers.31.post_attention_layernorm.weight": "pytorch_model-32-of-33.bin", "model.layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-32-of-33.bin", "model.embed_tokens.weight": "pytorch_model-33-of-33.bin", "model.norm.weight": "pytorch_model-33-of-33.bin", "lm_head.weight": "pytorch_model-33-of-33.bin"}, "metadata": {"total_size": 13721661440}}

special_tokens_map.json

@@ -0,0 +1,23 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39ea56d83c2c49dd507c25b99a6974b31149790b5f49015ee1437a9e84f43b2f
+size 700093

tokenizer_config.json

@@ -0,0 +1,33 @@
+{
+  "add_bos_token": true,
+  "add_eos_token": false,
+  "bos_token": {
+    "__type": "AddedToken",
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  },
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": null,
+  "sp_model_kwargs": {},
+  "tokenizer_class": "LlamaTokenizer",
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
+}