Initial commit

generation_config.json

@@ -0,0 +1,14 @@
+{
+  "max_length": 8192,
+  "do_sample": false,
+  "use_cache": true,
+  "temperature": 0.3,
+  "top_k": 5,
+  "top_p": 0.85,
+  "repetition_penalty": 1.03,
+  "pad_token_id": 3,
+  "bos_token_id": 1,
+  "eos_token_id": 2,
+  "user_token_id": 4,
+  "bot_token_id": 5
+}

model.safetensors.index.json

@@ -0,0 +1 @@
+{"metadata": {"total_size": 453131665408}, "weight_map": {"transformer.word_embeddings.weight": "model-00001-of-00050.safetensors", "transformer.h.0.input_layernorm.weight": "model-00001-of-00050.safetensors", "transformer.h.0.post_attention_layernorm.weight": "model-00001-of-00050.safetensors", "transformer.h.0.self_attention.dense.weight": "model-00001-of-00050.safetensors", "transformer.h.0.self_attention.dense.bias": "model-00001-of-00050.safetensors", "transformer.h.0.self_attention.query.weight": "model-00001-of-00050.safetensors", "transformer.h.0.self_attention.key_value.weight": "model-00001-of-00050.safetensors", "transformer.h.0.mlp.gate_proj.weight": "model-00001-of-00050.safetensors", "transformer.h.0.mlp.down_proj.weight": "model-00001-of-00050.safetensors", "transformer.h.0.mlp.down_proj.bias": "model-00001-of-00050.safetensors", "transformer.h.0.mlp.up_proj.weight": "model-00001-of-00050.safetensors", "transformer.h.1.input_layernorm.weight": "model-00002-of-00050.safetensors", "transformer.h.1.post_attention_layernorm.weight": "model-00002-of-00050.safetensors", "transformer.h.1.self_attention.dense.weight": "model-00002-of-00050.safetensors", "transformer.h.1.self_attention.dense.bias": "model-00002-of-00050.safetensors", "transformer.h.1.self_attention.query.weight": "model-00002-of-00050.safetensors", "transformer.h.1.self_attention.key_value.weight": "model-00002-of-00050.safetensors", "transformer.h.1.mlp.gate_proj.weight": "model-00002-of-00050.safetensors", "transformer.h.1.mlp.down_proj.weight": "model-00002-of-00050.safetensors", "transformer.h.1.mlp.down_proj.bias": "model-00002-of-00050.safetensors", "transformer.h.1.mlp.up_proj.weight": "model-00002-of-00050.safetensors", "transformer.h.2.input_layernorm.weight": "model-00002-of-00050.safetensors", "transformer.h.2.post_attention_layernorm.weight": "model-00002-of-00050.safetensors", "transformer.h.2.self_attention.dense.weight": "model-00002-of-00050.safetensors", "transformer.h.2.self_attention.dense.bias": "model-00002-of-00050.safetensors", "transformer.h.2.self_attention.query.weight": "model-00002-of-00050.safetensors", "transformer.h.2.self_attention.key_value.weight": "model-00002-of-00050.safetensors", "transformer.h.2.mlp.gate_proj.weight": "model-00002-of-00050.safetensors", "transformer.h.2.mlp.down_proj.weight": "model-00002-of-00050.safetensors", "transformer.h.2.mlp.down_proj.bias": "model-00002-of-00050.safetensors", "transformer.h.2.mlp.up_proj.weight": "model-00002-of-00050.safetensors", "transformer.h.3.input_layernorm.weight": "model-00003-of-00050.safetensors", "transformer.h.3.post_attention_layernorm.weight": "model-00003-of-00050.safetensors", "transformer.h.3.self_attention.dense.weight": "model-00003-of-00050.safetensors", "transformer.h.3.self_attention.dense.bias": "model-00003-of-00050.safetensors", "transformer.h.3.self_attention.query.weight": "model-00003-of-00050.safetensors", "transformer.h.3.self_attention.key_value.weight": "model-00003-of-00050.safetensors", "transformer.h.3.mlp.gate_proj.weight": "model-00003-of-00050.safetensors", "transformer.h.3.mlp.down_proj.weight": "model-00003-of-00050.safetensors", "transformer.h.3.mlp.down_proj.bias": "model-00003-of-00050.safetensors", "transformer.h.3.mlp.up_proj.weight": "model-00003-of-00050.safetensors", "transformer.h.4.input_layernorm.weight": "model-00003-of-00050.safetensors", "transformer.h.4.post_attention_layernorm.weight": "model-00003-of-00050.safetensors", "transformer.h.4.self_attention.dense.weight": "model-00003-of-00050.safetensors", "transformer.h.4.self_attention.dense.bias": "model-00003-of-00050.safetensors", "transformer.h.4.self_attention.query.weight": "model-00003-of-00050.safetensors", "transformer.h.4.self_attention.key_value.weight": "model-00003-of-00050.safetensors", "transformer.h.4.mlp.gate_proj.weight": "model-00003-of-00050.safetensors", "transformer.h.4.mlp.down_proj.weight": "model-00003-of-00050.safetensors", "transformer.h.4.mlp.down_proj.bias": "model-00003-of-00050.safetensors", "transformer.h.4.mlp.up_proj.weight": "model-00003-of-00050.safetensors", "transformer.h.5.input_layernorm.weight": "model-00004-of-00050.safetensors", "transformer.h.5.post_attention_layernorm.weight": "model-00004-of-00050.safetensors", "transformer.h.5.self_attention.dense.weight": "model-00004-of-00050.safetensors", "transformer.h.5.self_attention.dense.bias": "model-00004-of-00050.safetensors", "transformer.h.5.self_attention.query.weight": "model-00004-of-00050.safetensors", "transformer.h.5.self_attention.key_value.weight": "model-00004-of-00050.safetensors", "transformer.h.5.mlp.gate_proj.weight": "model-00004-of-00050.safetensors", "transformer.h.5.mlp.down_proj.weight": "model-00004-of-00050.safetensors", "transformer.h.5.mlp.down_proj.bias": "model-00004-of-00050.safetensors", "transformer.h.5.mlp.up_proj.weight": "model-00004-of-00050.safetensors", "transformer.h.6.input_layernorm.weight": "model-00004-of-00050.safetensors", "transformer.h.6.post_attention_layernorm.weight": "model-00004-of-00050.safetensors", "transformer.h.6.self_attention.dense.weight": "model-00004-of-00050.safetensors", "transformer.h.6.self_attention.dense.bias": "model-00004-of-00050.safetensors", "transformer.h.6.self_attention.query.weight": "model-00004-of-00050.safetensors", "transformer.h.6.self_attention.key_value.weight": "model-00004-of-00050.safetensors", "transformer.h.6.mlp.gate_proj.weight": "model-00004-of-00050.safetensors", "transformer.h.6.mlp.down_proj.weight": "model-00004-of-00050.safetensors", "transformer.h.6.mlp.down_proj.bias": "model-00004-of-00050.safetensors", "transformer.h.6.mlp.up_proj.weight": "model-00004-of-00050.safetensors", "transformer.h.7.input_layernorm.weight": "model-00005-of-00050.safetensors", "transformer.h.7.post_attention_layernorm.weight": "model-00005-of-00050.safetensors", "transformer.h.7.self_attention.dense.weight": "model-00005-of-00050.safetensors", "transformer.h.7.self_attention.dense.bias": "model-00005-of-00050.safetensors", "transformer.h.7.self_attention.query.weight": "model-00005-of-00050.safetensors", "transformer.h.7.self_attention.key_value.weight": "model-00005-of-00050.safetensors", "transformer.h.7.mlp.gate_proj.weight": "model-00005-of-00050.safetensors", "transformer.h.7.mlp.down_proj.weight": "model-00005-of-00050.safetensors", "transformer.h.7.mlp.down_proj.bias": "model-00005-of-00050.safetensors", "transformer.h.7.mlp.up_proj.weight": "model-00005-of-00050.safetensors", "transformer.h.8.input_layernorm.weight": "model-00005-of-00050.safetensors", "transformer.h.8.post_attention_layernorm.weight": "model-00005-of-00050.safetensors", "transformer.h.8.self_attention.dense.weight": "model-00005-of-00050.safetensors", "transformer.h.8.self_attention.dense.bias": "model-00005-of-00050.safetensors", "transformer.h.8.self_attention.query.weight": "model-00005-of-00050.safetensors", "transformer.h.8.self_attention.key_value.weight": "model-00005-of-00050.safetensors", "transformer.h.8.mlp.gate_proj.weight": "model-00005-of-00050.safetensors", "transformer.h.8.mlp.down_proj.weight": "model-00005-of-00050.safetensors", "transformer.h.8.mlp.down_proj.bias": "model-00005-of-00050.safetensors", "transformer.h.8.mlp.up_proj.weight": "model-00005-of-00050.safetensors", "transformer.h.9.input_layernorm.weight": "model-00006-of-00050.safetensors", "transformer.h.9.post_attention_layernorm.weight": "model-00006-of-00050.safetensors", "transformer.h.9.self_attention.dense.weight": "model-00006-of-00050.safetensors", "transformer.h.9.self_attention.dense.bias": "model-00006-of-00050.safetensors", "transformer.h.9.self_attention.query.weight": "model-00006-of-00050.safetensors", "transformer.h.9.self_attention.key_value.weight": "model-00006-of-00050.safetensors", "transformer.h.9.mlp.gate_proj.weight": "model-00006-of-00050.safetensors", "transformer.h.9.mlp.down_proj.weight": "model-00006-of-00050.safetensors", "transformer.h.9.mlp.down_proj.bias": "model-00006-of-00050.safetensors", "transformer.h.9.mlp.up_proj.weight": "model-00006-of-00050.safetensors", "transformer.h.10.input_layernorm.weight": "model-00006-of-00050.safetensors", "transformer.h.10.post_attention_layernorm.weight": "model-00006-of-00050.safetensors", "transformer.h.10.self_attention.dense.weight": "model-00006-of-00050.safetensors", "transformer.h.10.self_attention.dense.bias": "model-00006-of-00050.safetensors", "transformer.h.10.self_attention.query.weight": "model-00006-of-00050.safetensors", "transformer.h.10.self_attention.key_value.weight": "model-00006-of-00050.safetensors", "transformer.h.10.mlp.gate_proj.weight": "model-00006-of-00050.safetensors", "transformer.h.10.mlp.down_proj.weight": "model-00006-of-00050.safetensors", "transformer.h.10.mlp.down_proj.bias": "model-00006-of-00050.safetensors", "transformer.h.10.mlp.up_proj.weight": "model-00006-of-00050.safetensors", "transformer.h.11.input_layernorm.weight": "model-00007-of-00050.safetensors", "transformer.h.11.post_attention_layernorm.weight": "model-00007-of-00050.safetensors", "transformer.h.11.self_attention.dense.weight": "model-00007-of-00050.safetensors", "transformer.h.11.self_attention.dense.bias": "model-00007-of-00050.safetensors", "transformer.h.11.self_attention.query.weight": "model-00007-of-00050.safetensors", "transformer.h.11.self_attention.key_value.weight": "model-00007-of-00050.safetensors", "transformer.h.11.mlp.gate_proj.weight": "model-00007-of-00050.safetensors", "transformer.h.11.mlp.down_proj.weight": "model-00007-of-00050.safetensors", "transformer.h.11.mlp.down_proj.bias": "model-00007-of-00050.safetensors", "transformer.h.11.mlp.up_proj.weight": "model-00007-of-00050.safetensors", "transformer.h.12.input_layernorm.weight": "model-00007-of-00050.safetensors", "transformer.h.12.post_attention_layernorm.weight": "model-00007-of-00050.safetensors", "transformer.h.12.self_attention.dense.weight": "model-00007-of-00050.safetensors", "transformer.h.12.self_attention.dense.bias": "model-00007-of-00050.safetensors", "transformer.h.12.self_attention.query.weight": "model-00007-of-00050.safetensors", "transformer.h.12.self_attention.key_value.weight": "model-00007-of-00050.safetensors", "transformer.h.12.mlp.gate_proj.weight": "model-00007-of-00050.safetensors", "transformer.h.12.mlp.down_proj.weight": "model-00007-of-00050.safetensors", "transformer.h.12.mlp.down_proj.bias": "model-00007-of-00050.safetensors", "transformer.h.12.mlp.up_proj.weight": "model-00007-of-00050.safetensors", "transformer.h.13.input_layernorm.weight": "model-00008-of-00050.safetensors", "transformer.h.13.post_attention_layernorm.weight": "model-00008-of-00050.safetensors", "transformer.h.13.self_attention.dense.weight": "model-00008-of-00050.safetensors", "transformer.h.13.self_attention.dense.bias": "model-00008-of-00050.safetensors", "transformer.h.13.self_attention.query.weight": "model-00008-of-00050.safetensors", "transformer.h.13.self_attention.key_value.weight": "model-00008-of-00050.safetensors", "transformer.h.13.mlp.gate_proj.weight": "model-00008-of-00050.safetensors", "transformer.h.13.mlp.down_proj.weight": "model-00008-of-00050.safetensors", "transformer.h.13.mlp.down_proj.bias": "model-00008-of-00050.safetensors", "transformer.h.13.mlp.up_proj.weight": "model-00008-of-00050.safetensors", "transformer.h.14.input_layernorm.weight": "model-00008-of-00050.safetensors", "transformer.h.14.post_attention_layernorm.weight": "model-00008-of-00050.safetensors", "transformer.h.14.self_attention.dense.weight": "model-00008-of-00050.safetensors", "transformer.h.14.self_attention.dense.bias": "model-00008-of-00050.safetensors", "transformer.h.14.self_attention.query.weight": "model-00008-of-00050.safetensors", "transformer.h.14.self_attention.key_value.weight": "model-00008-of-00050.safetensors", "transformer.h.14.mlp.gate_proj.weight": "model-00008-of-00050.safetensors", "transformer.h.14.mlp.down_proj.weight": "model-00008-of-00050.safetensors", "transformer.h.14.mlp.down_proj.bias": "model-00008-of-00050.safetensors", "transformer.h.14.mlp.up_proj.weight": "model-00008-of-00050.safetensors", "transformer.h.15.input_layernorm.weight": "model-00009-of-00050.safetensors", "transformer.h.15.post_attention_layernorm.weight": "model-00009-of-00050.safetensors", "transformer.h.15.self_attention.dense.weight": "model-00009-of-00050.safetensors", "transformer.h.15.self_attention.dense.bias": "model-00009-of-00050.safetensors", "transformer.h.15.self_attention.query.weight": "model-00009-of-00050.safetensors", "transformer.h.15.self_attention.key_value.weight": "model-00009-of-00050.safetensors", "transformer.h.15.mlp.gate_proj.weight": "model-00009-of-00050.safetensors", "transformer.h.15.mlp.down_proj.weight": "model-00009-of-00050.safetensors", "transformer.h.15.mlp.down_proj.bias": "model-00009-of-00050.safetensors", "transformer.h.15.mlp.up_proj.weight": "model-00009-of-00050.safetensors", "transformer.h.16.input_layernorm.weight": "model-00009-of-00050.safetensors", "transformer.h.16.post_attention_layernorm.weight": "model-00009-of-00050.safetensors", "transformer.h.16.self_attention.dense.weight": "model-00009-of-00050.safetensors", "transformer.h.16.self_attention.dense.bias": "model-00009-of-00050.safetensors", "transformer.h.16.self_attention.query.weight": "model-00009-of-00050.safetensors", "transformer.h.16.self_attention.key_value.weight": "model-00009-of-00050.safetensors", "transformer.h.16.mlp.gate_proj.weight": "model-00009-of-00050.safetensors", "transformer.h.16.mlp.down_proj.weight": "model-00009-of-00050.safetensors", "transformer.h.16.mlp.down_proj.bias": "model-00009-of-00050.safetensors", "transformer.h.16.mlp.up_proj.weight": "model-00009-of-00050.safetensors", "transformer.h.17.input_layernorm.weight": "model-00010-of-00050.safetensors", "transformer.h.17.post_attention_layernorm.weight": "model-00010-of-00050.safetensors", "transformer.h.17.self_attention.dense.weight": "model-00010-of-00050.safetensors", "transformer.h.17.self_attention.dense.bias": "model-00010-of-00050.safetensors", "transformer.h.17.self_attention.query.weight": "model-00010-of-00050.safetensors", "transformer.h.17.self_attention.key_value.weight": "model-00010-of-00050.safetensors", "transformer.h.17.mlp.gate_proj.weight": "model-00010-of-00050.safetensors", "transformer.h.17.mlp.down_proj.weight": "model-00010-of-00050.safetensors", "transformer.h.17.mlp.down_proj.bias": "model-00010-of-00050.safetensors", "transformer.h.17.mlp.up_proj.weight": "model-00010-of-00050.safetensors", "transformer.h.18.input_layernorm.weight": "model-00010-of-00050.safetensors", "transformer.h.18.post_attention_layernorm.weight": "model-00010-of-00050.safetensors", "transformer.h.18.self_attention.dense.weight": "model-00010-of-00050.safetensors", "transformer.h.18.self_attention.dense.bias": "model-00010-of-00050.safetensors", "transformer.h.18.self_attention.query.weight": "model-00010-of-00050.safetensors", "transformer.h.18.self_attention.key_value.weight": "model-00010-of-00050.safetensors", "transformer.h.18.mlp.gate_proj.weight": "model-00010-of-00050.safetensors", "transformer.h.18.mlp.down_proj.weight": "model-00010-of-00050.safetensors", "transformer.h.18.mlp.down_proj.bias": "model-00010-of-00050.safetensors", "transformer.h.18.mlp.up_proj.weight": "model-00010-of-00050.safetensors", "transformer.h.19.input_layernorm.weight": "model-00011-of-00050.safetensors", "transformer.h.19.post_attention_layernorm.weight": "model-00011-of-00050.safetensors", "transformer.h.19.self_attention.dense.weight": "model-00011-of-00050.safetensors", "transformer.h.19.self_attention.dense.bias": "model-00011-of-00050.safetensors", "transformer.h.19.self_attention.query.weight": "model-00011-of-00050.safetensors", "transformer.h.19.self_attention.key_value.weight": "model-00011-of-00050.safetensors", "transformer.h.19.mlp.gate_proj.weight": "model-00011-of-00050.safetensors", "transformer.h.19.mlp.down_proj.weight": "model-00011-of-00050.safetensors", "transformer.h.19.mlp.down_proj.bias": "model-00011-of-00050.safetensors", "transformer.h.19.mlp.up_proj.weight": "model-00011-of-00050.safetensors", "transformer.h.20.input_layernorm.weight": "model-00011-of-00050.safetensors", "transformer.h.20.post_attention_layernorm.weight": "model-00011-of-00050.safetensors", "transformer.h.20.self_attention.dense.weight": "model-00011-of-00050.safetensors", "transformer.h.20.self_attention.dense.bias": "model-00011-of-00050.safetensors", "transformer.h.20.self_attention.query.weight": "model-00011-of-00050.safetensors", "transformer.h.20.self_attention.key_value.weight": "model-00011-of-00050.safetensors", "transformer.h.20.mlp.gate_proj.weight": "model-00011-of-00050.safetensors", "transformer.h.20.mlp.down_proj.weight": "model-00011-of-00050.safetensors", "transformer.h.20.mlp.down_proj.bias": "model-00011-of-00050.safetensors", "transformer.h.20.mlp.up_proj.weight": "model-00011-of-00050.safetensors", "transformer.h.21.input_layernorm.weight": "model-00012-of-00050.safetensors", "transformer.h.21.post_attention_layernorm.weight": "model-00012-of-00050.safetensors", "transformer.h.21.self_attention.dense.weight": "model-00012-of-00050.safetensors", "transformer.h.21.self_attention.dense.bias": "model-00012-of-00050.safetensors", "transformer.h.21.self_attention.query.weight": "model-00012-of-00050.safetensors", "transformer.h.21.self_attention.key_value.weight": "model-00012-of-00050.safetensors", "transformer.h.21.mlp.gate_proj.weight": "model-00012-of-00050.safetensors", "transformer.h.21.mlp.down_proj.weight": "model-00012-of-00050.safetensors", "transformer.h.21.mlp.down_proj.bias": "model-00012-of-00050.safetensors", "transformer.h.21.mlp.up_proj.weight": "model-00012-of-00050.safetensors", "transformer.h.22.input_layernorm.weight": "model-00012-of-00050.safetensors", "transformer.h.22.post_attention_layernorm.weight": "model-00012-of-00050.safetensors", "transformer.h.22.self_attention.dense.weight": "model-00012-of-00050.safetensors", "transformer.h.22.self_attention.dense.bias": "model-00012-of-00050.safetensors", "transformer.h.22.self_attention.query.weight": "model-00012-of-00050.safetensors", "transformer.h.22.self_attention.key_value.weight": "model-00012-of-00050.safetensors", "transformer.h.22.mlp.gate_proj.weight": "model-00012-of-00050.safetensors", "transformer.h.22.mlp.down_proj.weight": "model-00012-of-00050.safetensors", "transformer.h.22.mlp.down_proj.bias": "model-00012-of-00050.safetensors", "transformer.h.22.mlp.up_proj.weight": "model-00012-of-00050.safetensors", "transformer.h.23.input_layernorm.weight": "model-00013-of-00050.safetensors", "transformer.h.23.post_attention_layernorm.weight": "model-00013-of-00050.safetensors", "transformer.h.23.self_attention.dense.weight": "model-00013-of-00050.safetensors", "transformer.h.23.self_attention.dense.bias": "model-00013-of-00050.safetensors", "transformer.h.23.self_attention.query.weight": "model-00013-of-00050.safetensors", "transformer.h.23.self_attention.key_value.weight": "model-00013-of-00050.safetensors", "transformer.h.23.mlp.gate_proj.weight": "model-00013-of-00050.safetensors", "transformer.h.23.mlp.down_proj.weight": "model-00013-of-00050.safetensors", "transformer.h.23.mlp.down_proj.bias": "model-00013-of-00050.safetensors", "transformer.h.23.mlp.up_proj.weight": "model-00013-of-00050.safetensors", "transformer.h.24.input_layernorm.weight": "model-00013-of-00050.safetensors", "transformer.h.24.post_attention_layernorm.weight": "model-00013-of-00050.safetensors", "transformer.h.24.self_attention.dense.weight": "model-00013-of-00050.safetensors", "transformer.h.24.self_attention.dense.bias": "model-00013-of-00050.safetensors", "transformer.h.24.self_attention.query.weight": "model-00013-of-00050.safetensors", "transformer.h.24.self_attention.key_value.weight": "model-00013-of-00050.safetensors", "transformer.h.24.mlp.gate_proj.weight": "model-00013-of-00050.safetensors", "transformer.h.24.mlp.down_proj.weight": "model-00013-of-00050.safetensors", "transformer.h.24.mlp.down_proj.bias": "model-00013-of-00050.safetensors", "transformer.h.24.mlp.up_proj.weight": "model-00013-of-00050.safetensors", "transformer.h.25.input_layernorm.weight": "model-00014-of-00050.safetensors", "transformer.h.25.post_attention_layernorm.weight": "model-00014-of-00050.safetensors", "transformer.h.25.self_attention.dense.weight": "model-00014-of-00050.safetensors", "transformer.h.25.self_attention.dense.bias": "model-00014-of-00050.safetensors", "transformer.h.25.self_attention.query.weight": "model-00014-of-00050.safetensors", "transformer.h.25.self_attention.key_value.weight": "model-00014-of-00050.safetensors", "transformer.h.25.mlp.gate_proj.weight": "model-00014-of-00050.safetensors", "transformer.h.25.mlp.down_proj.weight": "model-00014-of-00050.safetensors", "transformer.h.25.mlp.down_proj.bias": "model-00014-of-00050.safetensors", "transformer.h.25.mlp.up_proj.weight": "model-00014-of-00050.safetensors", "transformer.h.26.input_layernorm.weight": "model-00014-of-00050.safetensors", "transformer.h.26.post_attention_layernorm.weight": "model-00014-of-00050.safetensors", "transformer.h.26.self_attention.dense.weight": "model-00014-of-00050.safetensors", "transformer.h.26.self_attention.dense.bias": "model-00014-of-00050.safetensors", "transformer.h.26.self_attention.query.weight": "model-00014-of-00050.safetensors", "transformer.h.26.self_attention.key_value.weight": "model-00014-of-00050.safetensors", "transformer.h.26.mlp.gate_proj.weight": "model-00014-of-00050.safetensors", "transformer.h.26.mlp.down_proj.weight": "model-00014-of-00050.safetensors", "transformer.h.26.mlp.down_proj.bias": "model-00014-of-00050.safetensors", "transformer.h.26.mlp.up_proj.weight": "model-00014-of-00050.safetensors", "transformer.h.27.input_layernorm.weight": "model-00015-of-00050.safetensors", "transformer.h.27.post_attention_layernorm.weight": "model-00015-of-00050.safetensors", "transformer.h.27.self_attention.dense.weight": "model-00015-of-00050.safetensors", "transformer.h.27.self_attention.dense.bias": "model-00015-of-00050.safetensors", "transformer.h.27.self_attention.query.weight": "model-00015-of-00050.safetensors", "transformer.h.27.self_attention.key_value.weight": "model-00015-of-00050.safetensors", "transformer.h.27.mlp.gate_proj.weight": "model-00015-of-00050.safetensors", "transformer.h.27.mlp.down_proj.weight": "model-00015-of-00050.safetensors", "transformer.h.27.mlp.down_proj.bias": "model-00015-of-00050.safetensors", "transformer.h.27.mlp.up_proj.weight": "model-00015-of-00050.safetensors", "transformer.h.28.input_layernorm.weight": "model-00015-of-00050.safetensors", "transformer.h.28.post_attention_layernorm.weight": "model-00015-of-00050.safetensors", "transformer.h.28.self_attention.dense.weight": "model-00015-of-00050.safetensors", "transformer.h.28.self_attention.dense.bias": "model-00015-of-00050.safetensors", "transformer.h.28.self_attention.query.weight": "model-00015-of-00050.safetensors", "transformer.h.28.self_attention.key_value.weight": "model-00015-of-00050.safetensors", "transformer.h.28.mlp.gate_proj.weight": "model-00015-of-00050.safetensors", "transformer.h.28.mlp.down_proj.weight": "model-00015-of-00050.safetensors", "transformer.h.28.mlp.down_proj.bias": "model-00015-of-00050.safetensors", "transformer.h.28.mlp.up_proj.weight": "model-00015-of-00050.safetensors", "transformer.h.29.input_layernorm.weight": "model-00016-of-00050.safetensors", "transformer.h.29.post_attention_layernorm.weight": "model-00016-of-00050.safetensors", "transformer.h.29.self_attention.dense.weight": "model-00016-of-00050.safetensors", "transformer.h.29.self_attention.dense.bias": "model-00016-of-00050.safetensors", "transformer.h.29.self_attention.query.weight": "model-00016-of-00050.safetensors", "transformer.h.29.self_attention.key_value.weight": "model-00016-of-00050.safetensors", "transformer.h.29.mlp.gate_proj.weight": "model-00016-of-00050.safetensors", "transformer.h.29.mlp.down_proj.weight": "model-00016-of-00050.safetensors", "transformer.h.29.mlp.down_proj.bias": "model-00016-of-00050.safetensors", "transformer.h.29.mlp.up_proj.weight": "model-00016-of-00050.safetensors", "transformer.h.30.input_layernorm.weight": "model-00016-of-00050.safetensors", "transformer.h.30.post_attention_layernorm.weight": "model-00016-of-00050.safetensors", "transformer.h.30.self_attention.dense.weight": "model-00016-of-00050.safetensors", "transformer.h.30.self_attention.dense.bias": "model-00016-of-00050.safetensors", "transformer.h.30.self_attention.query.weight": "model-00016-of-00050.safetensors", "transformer.h.30.self_attention.key_value.weight": "model-00016-of-00050.safetensors", "transformer.h.30.mlp.gate_proj.weight": "model-00016-of-00050.safetensors", "transformer.h.30.mlp.down_proj.weight": "model-00016-of-00050.safetensors", "transformer.h.30.mlp.down_proj.bias": "model-00016-of-00050.safetensors", "transformer.h.30.mlp.up_proj.weight": "model-00016-of-00050.safetensors", "transformer.h.31.input_layernorm.weight": "model-00017-of-00050.safetensors", "transformer.h.31.post_attention_layernorm.weight": "model-00017-of-00050.safetensors", "transformer.h.31.self_attention.dense.weight": "model-00017-of-00050.safetensors", "transformer.h.31.self_attention.dense.bias": "model-00017-of-00050.safetensors", "transformer.h.31.self_attention.query.weight": "model-00017-of-00050.safetensors", "transformer.h.31.self_attention.key_value.weight": "model-00017-of-00050.safetensors", "transformer.h.31.mlp.gate_proj.weight": "model-00017-of-00050.safetensors", "transformer.h.31.mlp.down_proj.weight": "model-00017-of-00050.safetensors", "transformer.h.31.mlp.down_proj.bias": "model-00017-of-00050.safetensors", "transformer.h.31.mlp.up_proj.weight": "model-00017-of-00050.safetensors", "transformer.h.32.input_layernorm.weight": "model-00017-of-00050.safetensors", "transformer.h.32.post_attention_layernorm.weight": "model-00017-of-00050.safetensors", "transformer.h.32.self_attention.dense.weight": "model-00017-of-00050.safetensors", "transformer.h.32.self_attention.dense.bias": "model-00017-of-00050.safetensors", "transformer.h.32.self_attention.query.weight": "model-00017-of-00050.safetensors", "transformer.h.32.self_attention.key_value.weight": "model-00017-of-00050.safetensors", "transformer.h.32.mlp.gate_proj.weight": "model-00017-of-00050.safetensors", "transformer.h.32.mlp.down_proj.weight": "model-00017-of-00050.safetensors", "transformer.h.32.mlp.down_proj.bias": "model-00017-of-00050.safetensors", "transformer.h.32.mlp.up_proj.weight": "model-00017-of-00050.safetensors", "transformer.h.33.input_layernorm.weight": "model-00018-of-00050.safetensors", "transformer.h.33.post_attention_layernorm.weight": "model-00018-of-00050.safetensors", "transformer.h.33.self_attention.dense.weight": "model-00018-of-00050.safetensors", "transformer.h.33.self_attention.dense.bias": "model-00018-of-00050.safetensors", "transformer.h.33.self_attention.query.weight": "model-00018-of-00050.safetensors", "transformer.h.33.self_attention.key_value.weight": "model-00018-of-00050.safetensors", "transformer.h.33.mlp.gate_proj.weight": "model-00018-of-00050.safetensors", "transformer.h.33.mlp.down_proj.weight": "model-00018-of-00050.safetensors", "transformer.h.33.mlp.down_proj.bias": "model-00018-of-00050.safetensors", "transformer.h.33.mlp.up_proj.weight": "model-00018-of-00050.safetensors", "transformer.h.34.input_layernorm.weight": "model-00018-of-00050.safetensors", "transformer.h.34.post_attention_layernorm.weight": "model-00018-of-00050.safetensors", "transformer.h.34.self_attention.dense.weight": "model-00018-of-00050.safetensors", "transformer.h.34.self_attention.dense.bias": "model-00018-of-00050.safetensors", "transformer.h.34.self_attention.query.weight": "model-00018-of-00050.safetensors", "transformer.h.34.self_attention.key_value.weight": "model-00018-of-00050.safetensors", "transformer.h.34.mlp.gate_proj.weight": "model-00018-of-00050.safetensors", "transformer.h.34.mlp.down_proj.weight": "model-00018-of-00050.safetensors", "transformer.h.34.mlp.down_proj.bias": "model-00018-of-00050.safetensors", "transformer.h.34.mlp.up_proj.weight": "model-00018-of-00050.safetensors", "transformer.h.35.input_layernorm.weight": "model-00019-of-00050.safetensors", "transformer.h.35.post_attention_layernorm.weight": "model-00019-of-00050.safetensors", "transformer.h.35.self_attention.dense.weight": "model-00019-of-00050.safetensors", "transformer.h.35.self_attention.dense.bias": "model-00019-of-00050.safetensors", "transformer.h.35.self_attention.query.weight": "model-00019-of-00050.safetensors", "transformer.h.35.self_attention.key_value.weight": "model-00019-of-00050.safetensors", "transformer.h.35.mlp.gate_proj.weight": "model-00019-of-00050.safetensors", "transformer.h.35.mlp.down_proj.weight": "model-00019-of-00050.safetensors", "transformer.h.35.mlp.down_proj.bias": "model-00019-of-00050.safetensors", "transformer.h.35.mlp.up_proj.weight": "model-00019-of-00050.safetensors", "transformer.h.36.input_layernorm.weight": "model-00019-of-00050.safetensors", "transformer.h.36.post_attention_layernorm.weight": "model-00019-of-00050.safetensors", "transformer.h.36.self_attention.dense.weight": "model-00019-of-00050.safetensors", "transformer.h.36.self_attention.dense.bias": "model-00019-of-00050.safetensors", "transformer.h.36.self_attention.query.weight": "model-00019-of-00050.safetensors", "transformer.h.36.self_attention.key_value.weight": "model-00019-of-00050.safetensors", "transformer.h.36.mlp.gate_proj.weight": "model-00019-of-00050.safetensors", "transformer.h.36.mlp.down_proj.weight": "model-00019-of-00050.safetensors", "transformer.h.36.mlp.down_proj.bias": "model-00019-of-00050.safetensors", "transformer.h.36.mlp.up_proj.weight": "model-00019-of-00050.safetensors", "transformer.h.37.input_layernorm.weight": "model-00020-of-00050.safetensors", "transformer.h.37.post_attention_layernorm.weight": "model-00020-of-00050.safetensors", "transformer.h.37.self_attention.dense.weight": "model-00020-of-00050.safetensors", "transformer.h.37.self_attention.dense.bias": "model-00020-of-00050.safetensors", "transformer.h.37.self_attention.query.weight": "model-00020-of-00050.safetensors", "transformer.h.37.self_attention.key_value.weight": "model-00020-of-00050.safetensors", "transformer.h.37.mlp.gate_proj.weight": "model-00020-of-00050.safetensors", "transformer.h.37.mlp.down_proj.weight": "model-00020-of-00050.safetensors", "transformer.h.37.mlp.down_proj.bias": "model-00020-of-00050.safetensors", "transformer.h.37.mlp.up_proj.weight": "model-00020-of-00050.safetensors", "transformer.h.38.input_layernorm.weight": "model-00020-of-00050.safetensors", "transformer.h.38.post_attention_layernorm.weight": "model-00020-of-00050.safetensors", "transformer.h.38.self_attention.dense.weight": "model-00020-of-00050.safetensors", "transformer.h.38.self_attention.dense.bias": "model-00020-of-00050.safetensors", "transformer.h.38.self_attention.query.weight": "model-00020-of-00050.safetensors", "transformer.h.38.self_attention.key_value.weight": "model-00020-of-00050.safetensors", "transformer.h.38.mlp.gate_proj.weight": "model-00020-of-00050.safetensors", "transformer.h.38.mlp.down_proj.weight": "model-00020-of-00050.safetensors", "transformer.h.38.mlp.down_proj.bias": "model-00020-of-00050.safetensors", "transformer.h.38.mlp.up_proj.weight": "model-00020-of-00050.safetensors", "transformer.h.39.input_layernorm.weight": "model-00021-of-00050.safetensors", "transformer.h.39.post_attention_layernorm.weight": "model-00021-of-00050.safetensors", "transformer.h.39.self_attention.dense.weight": "model-00021-of-00050.safetensors", "transformer.h.39.self_attention.dense.bias": "model-00021-of-00050.safetensors", "transformer.h.39.self_attention.query.weight": "model-00021-of-00050.safetensors", "transformer.h.39.self_attention.key_value.weight": "model-00021-of-00050.safetensors", "transformer.h.39.mlp.gate_proj.weight": "model-00021-of-00050.safetensors", "transformer.h.39.mlp.down_proj.weight": "model-00021-of-00050.safetensors", "transformer.h.39.mlp.down_proj.bias": "model-00021-of-00050.safetensors", "transformer.h.39.mlp.up_proj.weight": "model-00021-of-00050.safetensors", "transformer.h.40.input_layernorm.weight": "model-00021-of-00050.safetensors", "transformer.h.40.post_attention_layernorm.weight": "model-00021-of-00050.safetensors", "transformer.h.40.self_attention.dense.weight": "model-00021-of-00050.safetensors", "transformer.h.40.self_attention.dense.bias": "model-00021-of-00050.safetensors", "transformer.h.40.self_attention.query.weight": "model-00021-of-00050.safetensors", "transformer.h.40.self_attention.key_value.weight": "model-00021-of-00050.safetensors", "transformer.h.40.mlp.gate_proj.weight": "model-00021-of-00050.safetensors", "transformer.h.40.mlp.down_proj.weight": "model-00021-of-00050.safetensors", "transformer.h.40.mlp.down_proj.bias": "model-00021-of-00050.safetensors", "transformer.h.40.mlp.up_proj.weight": "model-00021-of-00050.safetensors", "transformer.h.41.input_layernorm.weight": "model-00022-of-00050.safetensors", "transformer.h.41.post_attention_layernorm.weight": "model-00022-of-00050.safetensors", "transformer.h.41.self_attention.dense.weight": "model-00022-of-00050.safetensors", "transformer.h.41.self_attention.dense.bias": "model-00022-of-00050.safetensors", "transformer.h.41.self_attention.query.weight": "model-00022-of-00050.safetensors", "transformer.h.41.self_attention.key_value.weight": "model-00022-of-00050.safetensors", "transformer.h.41.mlp.gate_proj.weight": "model-00022-of-00050.safetensors", "transformer.h.41.mlp.down_proj.weight": "model-00022-of-00050.safetensors", "transformer.h.41.mlp.down_proj.bias": "model-00022-of-00050.safetensors", "transformer.h.41.mlp.up_proj.weight": "model-00022-of-00050.safetensors", "transformer.h.42.input_layernorm.weight": "model-00022-of-00050.safetensors", "transformer.h.42.post_attention_layernorm.weight": "model-00022-of-00050.safetensors", "transformer.h.42.self_attention.dense.weight": "model-00022-of-00050.safetensors", "transformer.h.42.self_attention.dense.bias": "model-00022-of-00050.safetensors", "transformer.h.42.self_attention.query.weight": "model-00022-of-00050.safetensors", "transformer.h.42.self_attention.key_value.weight": "model-00022-of-00050.safetensors", "transformer.h.42.mlp.gate_proj.weight": "model-00022-of-00050.safetensors", "transformer.h.42.mlp.down_proj.weight": "model-00022-of-00050.safetensors", "transformer.h.42.mlp.down_proj.bias": "model-00022-of-00050.safetensors", "transformer.h.42.mlp.up_proj.weight": "model-00022-of-00050.safetensors", "transformer.h.43.input_layernorm.weight": "model-00023-of-00050.safetensors", "transformer.h.43.post_attention_layernorm.weight": "model-00023-of-00050.safetensors", "transformer.h.43.self_attention.dense.weight": "model-00023-of-00050.safetensors", "transformer.h.43.self_attention.dense.bias": "model-00023-of-00050.safetensors", "transformer.h.43.self_attention.query.weight": "model-00023-of-00050.safetensors", "transformer.h.43.self_attention.key_value.weight": "model-00023-of-00050.safetensors", "transformer.h.43.mlp.gate_proj.weight": "model-00023-of-00050.safetensors", "transformer.h.43.mlp.down_proj.weight": "model-00023-of-00050.safetensors", "transformer.h.43.mlp.down_proj.bias": "model-00023-of-00050.safetensors", "transformer.h.43.mlp.up_proj.weight": "model-00023-of-00050.safetensors", "transformer.h.44.input_layernorm.weight": "model-00023-of-00050.safetensors", "transformer.h.44.post_attention_layernorm.weight": "model-00023-of-00050.safetensors", "transformer.h.44.self_attention.dense.weight": "model-00023-of-00050.safetensors", "transformer.h.44.self_attention.dense.bias": "model-00023-of-00050.safetensors", "transformer.h.44.self_attention.query.weight": "model-00023-of-00050.safetensors", "transformer.h.44.self_attention.key_value.weight": "model-00023-of-00050.safetensors", "transformer.h.44.mlp.gate_proj.weight": "model-00023-of-00050.safetensors", "transformer.h.44.mlp.down_proj.weight": "model-00023-of-00050.safetensors", "transformer.h.44.mlp.down_proj.bias": "model-00023-of-00050.safetensors", "transformer.h.44.mlp.up_proj.weight": "model-00023-of-00050.safetensors", "transformer.h.45.input_layernorm.weight": "model-00024-of-00050.safetensors", "transformer.h.45.post_attention_layernorm.weight": "model-00024-of-00050.safetensors", "transformer.h.45.self_attention.dense.weight": "model-00024-of-00050.safetensors", "transformer.h.45.self_attention.dense.bias": "model-00024-of-00050.safetensors", "transformer.h.45.self_attention.query.weight": "model-00024-of-00050.safetensors", "transformer.h.45.self_attention.key_value.weight": "model-00024-of-00050.safetensors", "transformer.h.45.mlp.gate_proj.weight": "model-00024-of-00050.safetensors", "transformer.h.45.mlp.down_proj.weight": "model-00024-of-00050.safetensors", "transformer.h.45.mlp.down_proj.bias": "model-00024-of-00050.safetensors", "transformer.h.45.mlp.up_proj.weight": "model-00024-of-00050.safetensors", "transformer.h.46.input_layernorm.weight": "model-00024-of-00050.safetensors", "transformer.h.46.post_attention_layernorm.weight": "model-00024-of-00050.safetensors", "transformer.h.46.self_attention.dense.weight": "model-00024-of-00050.safetensors", "transformer.h.46.self_attention.dense.bias": "model-00024-of-00050.safetensors", "transformer.h.46.self_attention.query.weight": "model-00024-of-00050.safetensors", "transformer.h.46.self_attention.key_value.weight": "model-00024-of-00050.safetensors", "transformer.h.46.mlp.gate_proj.weight": "model-00024-of-00050.safetensors", "transformer.h.46.mlp.down_proj.weight": "model-00024-of-00050.safetensors", "transformer.h.46.mlp.down_proj.bias": "model-00024-of-00050.safetensors", "transformer.h.46.mlp.up_proj.weight": "model-00024-of-00050.safetensors", "transformer.h.47.input_layernorm.weight": "model-00025-of-00050.safetensors", "transformer.h.47.post_attention_layernorm.weight": "model-00025-of-00050.safetensors", "transformer.h.47.self_attention.dense.weight": "model-00025-of-00050.safetensors", "transformer.h.47.self_attention.dense.bias": "model-00025-of-00050.safetensors", "transformer.h.47.self_attention.query.weight": "model-00025-of-00050.safetensors", "transformer.h.47.self_attention.key_value.weight": "model-00025-of-00050.safetensors", "transformer.h.47.mlp.gate_proj.weight": "model-00025-of-00050.safetensors", "transformer.h.47.mlp.down_proj.weight": "model-00025-of-00050.safetensors", "transformer.h.47.mlp.down_proj.bias": "model-00025-of-00050.safetensors", "transformer.h.47.mlp.up_proj.weight": "model-00025-of-00050.safetensors", "transformer.h.48.input_layernorm.weight": "model-00025-of-00050.safetensors", "transformer.h.48.post_attention_layernorm.weight": "model-00025-of-00050.safetensors", "transformer.h.48.self_attention.dense.weight": "model-00025-of-00050.safetensors", "transformer.h.48.self_attention.dense.bias": "model-00025-of-00050.safetensors", "transformer.h.48.self_attention.query.weight": "model-00025-of-00050.safetensors", "transformer.h.48.self_attention.key_value.weight": "model-00025-of-00050.safetensors", "transformer.h.48.mlp.gate_proj.weight": "model-00025-of-00050.safetensors", "transformer.h.48.mlp.down_proj.weight": "model-00025-of-00050.safetensors", "transformer.h.48.mlp.down_proj.bias": "model-00025-of-00050.safetensors", "transformer.h.48.mlp.up_proj.weight": "model-00025-of-00050.safetensors", "transformer.h.49.input_layernorm.weight": "model-00026-of-00050.safetensors", "transformer.h.49.post_attention_layernorm.weight": "model-00026-of-00050.safetensors", "transformer.h.49.self_attention.dense.weight": "model-00026-of-00050.safetensors", "transformer.h.49.self_attention.dense.bias": "model-00026-of-00050.safetensors", "transformer.h.49.self_attention.query.weight": "model-00026-of-00050.safetensors", "transformer.h.49.self_attention.key_value.weight": "model-00026-of-00050.safetensors", "transformer.h.49.mlp.gate_proj.weight": "model-00026-of-00050.safetensors", "transformer.h.49.mlp.down_proj.weight": "model-00026-of-00050.safetensors", "transformer.h.49.mlp.down_proj.bias": "model-00026-of-00050.safetensors", "transformer.h.49.mlp.up_proj.weight": "model-00026-of-00050.safetensors", "transformer.h.50.input_layernorm.weight": "model-00026-of-00050.safetensors", "transformer.h.50.post_attention_layernorm.weight": "model-00026-of-00050.safetensors", "transformer.h.50.self_attention.dense.weight": "model-00026-of-00050.safetensors", "transformer.h.50.self_attention.dense.bias": "model-00026-of-00050.safetensors", "transformer.h.50.self_attention.query.weight": "model-00026-of-00050.safetensors", "transformer.h.50.self_attention.key_value.weight": "model-00026-of-00050.safetensors", "transformer.h.50.mlp.gate_proj.weight": "model-00026-of-00050.safetensors", "transformer.h.50.mlp.down_proj.weight": "model-00026-of-00050.safetensors", "transformer.h.50.mlp.down_proj.bias": "model-00026-of-00050.safetensors", "transformer.h.50.mlp.up_proj.weight": "model-00026-of-00050.safetensors", "transformer.h.51.input_layernorm.weight": "model-00027-of-00050.safetensors", "transformer.h.51.post_attention_layernorm.weight": "model-00027-of-00050.safetensors", "transformer.h.51.self_attention.dense.weight": "model-00027-of-00050.safetensors", "transformer.h.51.self_attention.dense.bias": "model-00027-of-00050.safetensors", "transformer.h.51.self_attention.query.weight": "model-00027-of-00050.safetensors", "transformer.h.51.self_attention.key_value.weight": "model-00027-of-00050.safetensors", "transformer.h.51.mlp.gate_proj.weight": "model-00027-of-00050.safetensors", "transformer.h.51.mlp.down_proj.weight": "model-00027-of-00050.safetensors", "transformer.h.51.mlp.down_proj.bias": "model-00027-of-00050.safetensors", "transformer.h.51.mlp.up_proj.weight": "model-00027-of-00050.safetensors", "transformer.h.52.input_layernorm.weight": "model-00027-of-00050.safetensors", "transformer.h.52.post_attention_layernorm.weight": "model-00027-of-00050.safetensors", "transformer.h.52.self_attention.dense.weight": "model-00027-of-00050.safetensors", "transformer.h.52.self_attention.dense.bias": "model-00027-of-00050.safetensors", "transformer.h.52.self_attention.query.weight": "model-00027-of-00050.safetensors", "transformer.h.52.self_attention.key_value.weight": "model-00027-of-00050.safetensors", "transformer.h.52.mlp.gate_proj.weight": "model-00027-of-00050.safetensors", "transformer.h.52.mlp.down_proj.weight": "model-00027-of-00050.safetensors", "transformer.h.52.mlp.down_proj.bias": "model-00027-of-00050.safetensors", "transformer.h.52.mlp.up_proj.weight": "model-00027-of-00050.safetensors", "transformer.h.53.input_layernorm.weight": "model-00028-of-00050.safetensors", "transformer.h.53.post_attention_layernorm.weight": "model-00028-of-00050.safetensors", "transformer.h.53.self_attention.dense.weight": "model-00028-of-00050.safetensors", "transformer.h.53.self_attention.dense.bias": "model-00028-of-00050.safetensors", "transformer.h.53.self_attention.query.weight": "model-00028-of-00050.safetensors", "transformer.h.53.self_attention.key_value.weight": "model-00028-of-00050.safetensors", "transformer.h.53.mlp.gate_proj.weight": "model-00028-of-00050.safetensors", "transformer.h.53.mlp.down_proj.weight": "model-00028-of-00050.safetensors", "transformer.h.53.mlp.down_proj.bias": "model-00028-of-00050.safetensors", "transformer.h.53.mlp.up_proj.weight": "model-00028-of-00050.safetensors", "transformer.h.54.input_layernorm.weight": "model-00028-of-00050.safetensors", "transformer.h.54.post_attention_layernorm.weight": "model-00028-of-00050.safetensors", "transformer.h.54.self_attention.dense.weight": "model-00028-of-00050.safetensors", "transformer.h.54.self_attention.dense.bias": "model-00028-of-00050.safetensors", "transformer.h.54.self_attention.query.weight": "model-00028-of-00050.safetensors", "transformer.h.54.self_attention.key_value.weight": "model-00028-of-00050.safetensors", "transformer.h.54.mlp.gate_proj.weight": "model-00028-of-00050.safetensors", "transformer.h.54.mlp.down_proj.weight": "model-00028-of-00050.safetensors", "transformer.h.54.mlp.down_proj.bias": "model-00028-of-00050.safetensors", "transformer.h.54.mlp.up_proj.weight": "model-00028-of-00050.safetensors", "transformer.h.55.input_layernorm.weight": "model-00029-of-00050.safetensors", "transformer.h.55.post_attention_layernorm.weight": "model-00029-of-00050.safetensors", "transformer.h.55.self_attention.dense.weight": "model-00029-of-00050.safetensors", "transformer.h.55.self_attention.dense.bias": "model-00029-of-00050.safetensors", "transformer.h.55.self_attention.query.weight": "model-00029-of-00050.safetensors", "transformer.h.55.self_attention.key_value.weight": "model-00029-of-00050.safetensors", "transformer.h.55.mlp.gate_proj.weight": "model-00029-of-00050.safetensors", "transformer.h.55.mlp.down_proj.weight": "model-00029-of-00050.safetensors", "transformer.h.55.mlp.down_proj.bias": "model-00029-of-00050.safetensors", "transformer.h.55.mlp.up_proj.weight": "model-00029-of-00050.safetensors", "transformer.h.56.input_layernorm.weight": "model-00029-of-00050.safetensors", "transformer.h.56.post_attention_layernorm.weight": "model-00029-of-00050.safetensors", "transformer.h.56.self_attention.dense.weight": "model-00029-of-00050.safetensors", "transformer.h.56.self_attention.dense.bias": "model-00029-of-00050.safetensors", "transformer.h.56.self_attention.query.weight": "model-00029-of-00050.safetensors", "transformer.h.56.self_attention.key_value.weight": "model-00029-of-00050.safetensors", "transformer.h.56.mlp.gate_proj.weight": "model-00029-of-00050.safetensors", "transformer.h.56.mlp.down_proj.weight": "model-00029-of-00050.safetensors", "transformer.h.56.mlp.down_proj.bias": "model-00029-of-00050.safetensors", "transformer.h.56.mlp.up_proj.weight": "model-00029-of-00050.safetensors", "transformer.h.57.input_layernorm.weight": "model-00030-of-00050.safetensors", "transformer.h.57.post_attention_layernorm.weight": "model-00030-of-00050.safetensors", "transformer.h.57.self_attention.dense.weight": "model-00030-of-00050.safetensors", "transformer.h.57.self_attention.dense.bias": "model-00030-of-00050.safetensors", "transformer.h.57.self_attention.query.weight": "model-00030-of-00050.safetensors", "transformer.h.57.self_attention.key_value.weight": "model-00030-of-00050.safetensors", "transformer.h.57.mlp.gate_proj.weight": "model-00030-of-00050.safetensors", "transformer.h.57.mlp.down_proj.weight": "model-00030-of-00050.safetensors", "transformer.h.57.mlp.down_proj.bias": "model-00030-of-00050.safetensors", "transformer.h.57.mlp.up_proj.weight": "model-00030-of-00050.safetensors", "transformer.h.58.input_layernorm.weight": "model-00030-of-00050.safetensors", "transformer.h.58.post_attention_layernorm.weight": "model-00030-of-00050.safetensors", "transformer.h.58.self_attention.dense.weight": "model-00030-of-00050.safetensors", "transformer.h.58.self_attention.dense.bias": "model-00030-of-00050.safetensors", "transformer.h.58.self_attention.query.weight": "model-00030-of-00050.safetensors", "transformer.h.58.self_attention.key_value.weight": "model-00030-of-00050.safetensors", "transformer.h.58.mlp.gate_proj.weight": "model-00030-of-00050.safetensors", "transformer.h.58.mlp.down_proj.weight": "model-00030-of-00050.safetensors", "transformer.h.58.mlp.down_proj.bias": "model-00030-of-00050.safetensors", "transformer.h.58.mlp.up_proj.weight": "model-00030-of-00050.safetensors", "transformer.h.59.input_layernorm.weight": "model-00031-of-00050.safetensors", "transformer.h.59.post_attention_layernorm.weight": "model-00031-of-00050.safetensors", "transformer.h.59.self_attention.dense.weight": "model-00031-of-00050.safetensors", "transformer.h.59.self_attention.dense.bias": "model-00031-of-00050.safetensors", "transformer.h.59.self_attention.query.weight": "model-00031-of-00050.safetensors", "transformer.h.59.self_attention.key_value.weight": "model-00031-of-00050.safetensors", "transformer.h.59.mlp.gate_proj.weight": "model-00031-of-00050.safetensors", "transformer.h.59.mlp.down_proj.weight": "model-00031-of-00050.safetensors", "transformer.h.59.mlp.down_proj.bias": "model-00031-of-00050.safetensors", "transformer.h.59.mlp.up_proj.weight": "model-00031-of-00050.safetensors", "transformer.h.60.input_layernorm.weight": "model-00031-of-00050.safetensors", "transformer.h.60.post_attention_layernorm.weight": "model-00031-of-00050.safetensors", "transformer.h.60.self_attention.dense.weight": "model-00031-of-00050.safetensors", "transformer.h.60.self_attention.dense.bias": "model-00031-of-00050.safetensors", "transformer.h.60.self_attention.query.weight": "model-00031-of-00050.safetensors", "transformer.h.60.self_attention.key_value.weight": "model-00031-of-00050.safetensors", "transformer.h.60.mlp.gate_proj.weight": "model-00031-of-00050.safetensors", "transformer.h.60.mlp.down_proj.weight": "model-00031-of-00050.safetensors", "transformer.h.60.mlp.down_proj.bias": "model-00031-of-00050.safetensors", "transformer.h.60.mlp.up_proj.weight": "model-00031-of-00050.safetensors", "transformer.h.61.input_layernorm.weight": "model-00032-of-00050.safetensors", "transformer.h.61.post_attention_layernorm.weight": "model-00032-of-00050.safetensors", "transformer.h.61.self_attention.dense.weight": "model-00032-of-00050.safetensors", "transformer.h.61.self_attention.dense.bias": "model-00032-of-00050.safetensors", "transformer.h.61.self_attention.query.weight": "model-00032-of-00050.safetensors", "transformer.h.61.self_attention.key_value.weight": "model-00032-of-00050.safetensors", "transformer.h.61.mlp.gate_proj.weight": "model-00032-of-00050.safetensors", "transformer.h.61.mlp.down_proj.weight": "model-00032-of-00050.safetensors", "transformer.h.61.mlp.down_proj.bias": "model-00032-of-00050.safetensors", "transformer.h.61.mlp.up_proj.weight": "model-00032-of-00050.safetensors", "transformer.h.62.input_layernorm.weight": "model-00032-of-00050.safetensors", "transformer.h.62.post_attention_layernorm.weight": "model-00032-of-00050.safetensors", "transformer.h.62.self_attention.dense.weight": "model-00032-of-00050.safetensors", "transformer.h.62.self_attention.dense.bias": "model-00032-of-00050.safetensors", "transformer.h.62.self_attention.query.weight": "model-00032-of-00050.safetensors", "transformer.h.62.self_attention.key_value.weight": "model-00032-of-00050.safetensors", "transformer.h.62.mlp.gate_proj.weight": "model-00032-of-00050.safetensors", "transformer.h.62.mlp.down_proj.weight": "model-00032-of-00050.safetensors", "transformer.h.62.mlp.down_proj.bias": "model-00032-of-00050.safetensors", "transformer.h.62.mlp.up_proj.weight": "model-00032-of-00050.safetensors", "transformer.h.63.input_layernorm.weight": "model-00033-of-00050.safetensors", "transformer.h.63.post_attention_layernorm.weight": "model-00033-of-00050.safetensors", "transformer.h.63.self_attention.dense.weight": "model-00033-of-00050.safetensors", "transformer.h.63.self_attention.dense.bias": "model-00033-of-00050.safetensors", "transformer.h.63.self_attention.query.weight": "model-00033-of-00050.safetensors", "transformer.h.63.self_attention.key_value.weight": "model-00033-of-00050.safetensors", "transformer.h.63.mlp.gate_proj.weight": "model-00033-of-00050.safetensors", "transformer.h.63.mlp.down_proj.weight": "model-00033-of-00050.safetensors", "transformer.h.63.mlp.down_proj.bias": "model-00033-of-00050.safetensors", "transformer.h.63.mlp.up_proj.weight": "model-00033-of-00050.safetensors", "transformer.h.64.input_layernorm.weight": "model-00033-of-00050.safetensors", "transformer.h.64.post_attention_layernorm.weight": "model-00033-of-00050.safetensors", "transformer.h.64.self_attention.dense.weight": "model-00033-of-00050.safetensors", "transformer.h.64.self_attention.dense.bias": "model-00033-of-00050.safetensors", "transformer.h.64.self_attention.query.weight": "model-00033-of-00050.safetensors", "transformer.h.64.self_attention.key_value.weight": "model-00033-of-00050.safetensors", "transformer.h.64.mlp.gate_proj.weight": "model-00033-of-00050.safetensors", "transformer.h.64.mlp.down_proj.weight": "model-00033-of-00050.safetensors", "transformer.h.64.mlp.down_proj.bias": "model-00033-of-00050.safetensors", "transformer.h.64.mlp.up_proj.weight": "model-00033-of-00050.safetensors", "transformer.h.65.input_layernorm.weight": "model-00034-of-00050.safetensors", "transformer.h.65.post_attention_layernorm.weight": "model-00034-of-00050.safetensors", "transformer.h.65.self_attention.dense.weight": "model-00034-of-00050.safetensors", "transformer.h.65.self_attention.dense.bias": "model-00034-of-00050.safetensors", "transformer.h.65.self_attention.query.weight": "model-00034-of-00050.safetensors", "transformer.h.65.self_attention.key_value.weight": "model-00034-of-00050.safetensors", "transformer.h.65.mlp.gate_proj.weight": "model-00034-of-00050.safetensors", "transformer.h.65.mlp.down_proj.weight": "model-00034-of-00050.safetensors", "transformer.h.65.mlp.down_proj.bias": "model-00034-of-00050.safetensors", "transformer.h.65.mlp.up_proj.weight": "model-00034-of-00050.safetensors", "transformer.h.66.input_layernorm.weight": "model-00034-of-00050.safetensors", "transformer.h.66.post_attention_layernorm.weight": "model-00034-of-00050.safetensors", "transformer.h.66.self_attention.dense.weight": "model-00034-of-00050.safetensors", "transformer.h.66.self_attention.dense.bias": "model-00034-of-00050.safetensors", "transformer.h.66.self_attention.query.weight": "model-00034-of-00050.safetensors", "transformer.h.66.self_attention.key_value.weight": "model-00034-of-00050.safetensors", "transformer.h.66.mlp.gate_proj.weight": "model-00034-of-00050.safetensors", "transformer.h.66.mlp.down_proj.weight": "model-00034-of-00050.safetensors", "transformer.h.66.mlp.down_proj.bias": "model-00034-of-00050.safetensors", "transformer.h.66.mlp.up_proj.weight": "model-00034-of-00050.safetensors", "transformer.h.67.input_layernorm.weight": "model-00035-of-00050.safetensors", "transformer.h.67.post_attention_layernorm.weight": "model-00035-of-00050.safetensors", "transformer.h.67.self_attention.dense.weight": "model-00035-of-00050.safetensors", "transformer.h.67.self_attention.dense.bias": "model-00035-of-00050.safetensors", "transformer.h.67.self_attention.query.weight": "model-00035-of-00050.safetensors", "transformer.h.67.self_attention.key_value.weight": "model-00035-of-00050.safetensors", "transformer.h.67.mlp.gate_proj.weight": "model-00035-of-00050.safetensors", "transformer.h.67.mlp.down_proj.weight": "model-00035-of-00050.safetensors", "transformer.h.67.mlp.down_proj.bias": "model-00035-of-00050.safetensors", "transformer.h.67.mlp.up_proj.weight": "model-00035-of-00050.safetensors", "transformer.h.68.input_layernorm.weight": "model-00035-of-00050.safetensors", "transformer.h.68.post_attention_layernorm.weight": "model-00035-of-00050.safetensors", "transformer.h.68.self_attention.dense.weight": "model-00035-of-00050.safetensors", "transformer.h.68.self_attention.dense.bias": "model-00035-of-00050.safetensors", "transformer.h.68.self_attention.query.weight": "model-00035-of-00050.safetensors", "transformer.h.68.self_attention.key_value.weight": "model-00035-of-00050.safetensors", "transformer.h.68.mlp.gate_proj.weight": "model-00035-of-00050.safetensors", "transformer.h.68.mlp.down_proj.weight": "model-00035-of-00050.safetensors", "transformer.h.68.mlp.down_proj.bias": "model-00035-of-00050.safetensors", "transformer.h.68.mlp.up_proj.weight": "model-00035-of-00050.safetensors", "transformer.h.69.input_layernorm.weight": "model-00036-of-00050.safetensors", "transformer.h.69.post_attention_layernorm.weight": "model-00036-of-00050.safetensors", "transformer.h.69.self_attention.dense.weight": "model-00036-of-00050.safetensors", "transformer.h.69.self_attention.dense.bias": "model-00036-of-00050.safetensors", "transformer.h.69.self_attention.query.weight": "model-00036-of-00050.safetensors", "transformer.h.69.self_attention.key_value.weight": "model-00036-of-00050.safetensors", "transformer.h.69.mlp.gate_proj.weight": "model-00036-of-00050.safetensors", "transformer.h.69.mlp.down_proj.weight": "model-00036-of-00050.safetensors", "transformer.h.69.mlp.down_proj.bias": "model-00036-of-00050.safetensors", "transformer.h.69.mlp.up_proj.weight": "model-00036-of-00050.safetensors", "transformer.h.70.input_layernorm.weight": "model-00036-of-00050.safetensors", "transformer.h.70.post_attention_layernorm.weight": "model-00036-of-00050.safetensors", "transformer.h.70.self_attention.dense.weight": "model-00036-of-00050.safetensors", "transformer.h.70.self_attention.dense.bias": "model-00036-of-00050.safetensors", "transformer.h.70.self_attention.query.weight": "model-00036-of-00050.safetensors", "transformer.h.70.self_attention.key_value.weight": "model-00036-of-00050.safetensors", "transformer.h.70.mlp.gate_proj.weight": "model-00036-of-00050.safetensors", "transformer.h.70.mlp.down_proj.weight": "model-00036-of-00050.safetensors", "transformer.h.70.mlp.down_proj.bias": "model-00036-of-00050.safetensors", "transformer.h.70.mlp.up_proj.weight": "model-00036-of-00050.safetensors", "transformer.h.71.input_layernorm.weight": "model-00037-of-00050.safetensors", "transformer.h.71.post_attention_layernorm.weight": "model-00037-of-00050.safetensors", "transformer.h.71.self_attention.dense.weight": "model-00037-of-00050.safetensors", "transformer.h.71.self_attention.dense.bias": "model-00037-of-00050.safetensors", "transformer.h.71.self_attention.query.weight": "model-00037-of-00050.safetensors", "transformer.h.71.self_attention.key_value.weight": "model-00037-of-00050.safetensors", "transformer.h.71.mlp.gate_proj.weight": "model-00037-of-00050.safetensors", "transformer.h.71.mlp.down_proj.weight": "model-00037-of-00050.safetensors", "transformer.h.71.mlp.down_proj.bias": "model-00037-of-00050.safetensors", "transformer.h.71.mlp.up_proj.weight": "model-00037-of-00050.safetensors", "transformer.h.72.input_layernorm.weight": "model-00037-of-00050.safetensors", "transformer.h.72.post_attention_layernorm.weight": "model-00037-of-00050.safetensors", "transformer.h.72.self_attention.dense.weight": "model-00037-of-00050.safetensors", "transformer.h.72.self_attention.dense.bias": "model-00037-of-00050.safetensors", "transformer.h.72.self_attention.query.weight": "model-00037-of-00050.safetensors", "transformer.h.72.self_attention.key_value.weight": "model-00037-of-00050.safetensors", "transformer.h.72.mlp.gate_proj.weight": "model-00037-of-00050.safetensors", "transformer.h.72.mlp.down_proj.weight": "model-00037-of-00050.safetensors", "transformer.h.72.mlp.down_proj.bias": "model-00037-of-00050.safetensors", "transformer.h.72.mlp.up_proj.weight": "model-00037-of-00050.safetensors", "transformer.h.73.input_layernorm.weight": "model-00038-of-00050.safetensors", "transformer.h.73.post_attention_layernorm.weight": "model-00038-of-00050.safetensors", "transformer.h.73.self_attention.dense.weight": "model-00038-of-00050.safetensors", "transformer.h.73.self_attention.dense.bias": "model-00038-of-00050.safetensors", "transformer.h.73.self_attention.query.weight": "model-00038-of-00050.safetensors", "transformer.h.73.self_attention.key_value.weight": "model-00038-of-00050.safetensors", "transformer.h.73.mlp.gate_proj.weight": "model-00038-of-00050.safetensors", "transformer.h.73.mlp.down_proj.weight": "model-00038-of-00050.safetensors", "transformer.h.73.mlp.down_proj.bias": "model-00038-of-00050.safetensors", "transformer.h.73.mlp.up_proj.weight": "model-00038-of-00050.safetensors", "transformer.h.74.input_layernorm.weight": "model-00038-of-00050.safetensors", "transformer.h.74.post_attention_layernorm.weight": "model-00038-of-00050.safetensors", "transformer.h.74.self_attention.dense.weight": "model-00038-of-00050.safetensors", "transformer.h.74.self_attention.dense.bias": "model-00038-of-00050.safetensors", "transformer.h.74.self_attention.query.weight": "model-00038-of-00050.safetensors", "transformer.h.74.self_attention.key_value.weight": "model-00038-of-00050.safetensors", "transformer.h.74.mlp.gate_proj.weight": "model-00038-of-00050.safetensors", "transformer.h.74.mlp.down_proj.weight": "model-00038-of-00050.safetensors", "transformer.h.74.mlp.down_proj.bias": "model-00038-of-00050.safetensors", "transformer.h.74.mlp.up_proj.weight": "model-00038-of-00050.safetensors", "transformer.h.75.input_layernorm.weight": "model-00039-of-00050.safetensors", "transformer.h.75.post_attention_layernorm.weight": "model-00039-of-00050.safetensors", "transformer.h.75.self_attention.dense.weight": "model-00039-of-00050.safetensors", "transformer.h.75.self_attention.dense.bias": "model-00039-of-00050.safetensors", "transformer.h.75.self_attention.query.weight": "model-00039-of-00050.safetensors", "transformer.h.75.self_attention.key_value.weight": "model-00039-of-00050.safetensors", "transformer.h.75.mlp.gate_proj.weight": "model-00039-of-00050.safetensors", "transformer.h.75.mlp.down_proj.weight": "model-00039-of-00050.safetensors", "transformer.h.75.mlp.down_proj.bias": "model-00039-of-00050.safetensors", "transformer.h.75.mlp.up_proj.weight": "model-00039-of-00050.safetensors", "transformer.h.76.input_layernorm.weight": "model-00039-of-00050.safetensors", "transformer.h.76.post_attention_layernorm.weight": "model-00039-of-00050.safetensors", "transformer.h.76.self_attention.dense.weight": "model-00039-of-00050.safetensors", "transformer.h.76.self_attention.dense.bias": "model-00039-of-00050.safetensors", "transformer.h.76.self_attention.query.weight": "model-00039-of-00050.safetensors", "transformer.h.76.self_attention.key_value.weight": "model-00039-of-00050.safetensors", "transformer.h.76.mlp.gate_proj.weight": "model-00039-of-00050.safetensors", "transformer.h.76.mlp.down_proj.weight": "model-00039-of-00050.safetensors", "transformer.h.76.mlp.down_proj.bias": "model-00039-of-00050.safetensors", "transformer.h.76.mlp.up_proj.weight": "model-00039-of-00050.safetensors", "transformer.h.77.input_layernorm.weight": "model-00040-of-00050.safetensors", "transformer.h.77.post_attention_layernorm.weight": "model-00040-of-00050.safetensors", "transformer.h.77.self_attention.dense.weight": "model-00040-of-00050.safetensors", "transformer.h.77.self_attention.dense.bias": "model-00040-of-00050.safetensors", "transformer.h.77.self_attention.query.weight": "model-00040-of-00050.safetensors", "transformer.h.77.self_attention.key_value.weight": "model-00040-of-00050.safetensors", "transformer.h.77.mlp.gate_proj.weight": "model-00040-of-00050.safetensors", "transformer.h.77.mlp.down_proj.weight": "model-00040-of-00050.safetensors", "transformer.h.77.mlp.down_proj.bias": "model-00040-of-00050.safetensors", "transformer.h.77.mlp.up_proj.weight": "model-00040-of-00050.safetensors", "transformer.h.78.input_layernorm.weight": "model-00040-of-00050.safetensors", "transformer.h.78.post_attention_layernorm.weight": "model-00040-of-00050.safetensors", "transformer.h.78.self_attention.dense.weight": "model-00040-of-00050.safetensors", "transformer.h.78.self_attention.dense.bias": "model-00040-of-00050.safetensors", "transformer.h.78.self_attention.query.weight": "model-00040-of-00050.safetensors", "transformer.h.78.self_attention.key_value.weight": "model-00040-of-00050.safetensors", "transformer.h.78.mlp.gate_proj.weight": "model-00040-of-00050.safetensors", "transformer.h.78.mlp.down_proj.weight": "model-00040-of-00050.safetensors", "transformer.h.78.mlp.down_proj.bias": "model-00040-of-00050.safetensors", "transformer.h.78.mlp.up_proj.weight": "model-00040-of-00050.safetensors", "transformer.h.79.input_layernorm.weight": "model-00041-of-00050.safetensors", "transformer.h.79.post_attention_layernorm.weight": "model-00041-of-00050.safetensors", "transformer.h.79.self_attention.dense.weight": "model-00041-of-00050.safetensors", "transformer.h.79.self_attention.dense.bias": "model-00041-of-00050.safetensors", "transformer.h.79.self_attention.query.weight": "model-00041-of-00050.safetensors", "transformer.h.79.self_attention.key_value.weight": "model-00041-of-00050.safetensors", "transformer.h.79.mlp.gate_proj.weight": "model-00041-of-00050.safetensors", "transformer.h.79.mlp.down_proj.weight": "model-00041-of-00050.safetensors", "transformer.h.79.mlp.down_proj.bias": "model-00041-of-00050.safetensors", "transformer.h.79.mlp.up_proj.weight": "model-00041-of-00050.safetensors", "transformer.h.80.input_layernorm.weight": "model-00041-of-00050.safetensors", "transformer.h.80.post_attention_layernorm.weight": "model-00041-of-00050.safetensors", "transformer.h.80.self_attention.dense.weight": "model-00041-of-00050.safetensors", "transformer.h.80.self_attention.dense.bias": "model-00041-of-00050.safetensors", "transformer.h.80.self_attention.query.weight": "model-00041-of-00050.safetensors", "transformer.h.80.self_attention.key_value.weight": "model-00041-of-00050.safetensors", "transformer.h.80.mlp.gate_proj.weight": "model-00041-of-00050.safetensors", "transformer.h.80.mlp.down_proj.weight": "model-00041-of-00050.safetensors", "transformer.h.80.mlp.down_proj.bias": "model-00041-of-00050.safetensors", "transformer.h.80.mlp.up_proj.weight": "model-00041-of-00050.safetensors", "transformer.h.81.input_layernorm.weight": "model-00042-of-00050.safetensors", "transformer.h.81.post_attention_layernorm.weight": "model-00042-of-00050.safetensors", "transformer.h.81.self_attention.dense.weight": "model-00042-of-00050.safetensors", "transformer.h.81.self_attention.dense.bias": "model-00042-of-00050.safetensors", "transformer.h.81.self_attention.query.weight": "model-00042-of-00050.safetensors", "transformer.h.81.self_attention.key_value.weight": "model-00042-of-00050.safetensors", "transformer.h.81.mlp.gate_proj.weight": "model-00042-of-00050.safetensors", "transformer.h.81.mlp.down_proj.weight": "model-00042-of-00050.safetensors", "transformer.h.81.mlp.down_proj.bias": "model-00042-of-00050.safetensors", "transformer.h.81.mlp.up_proj.weight": "model-00042-of-00050.safetensors", "transformer.h.82.input_layernorm.weight": "model-00042-of-00050.safetensors", "transformer.h.82.post_attention_layernorm.weight": "model-00042-of-00050.safetensors", "transformer.h.82.self_attention.dense.weight": "model-00042-of-00050.safetensors", "transformer.h.82.self_attention.dense.bias": "model-00042-of-00050.safetensors", "transformer.h.82.self_attention.query.weight": "model-00042-of-00050.safetensors", "transformer.h.82.self_attention.key_value.weight": "model-00042-of-00050.safetensors", "transformer.h.82.mlp.gate_proj.weight": "model-00042-of-00050.safetensors", "transformer.h.82.mlp.down_proj.weight": "model-00042-of-00050.safetensors", "transformer.h.82.mlp.down_proj.bias": "model-00042-of-00050.safetensors", "transformer.h.82.mlp.up_proj.weight": "model-00042-of-00050.safetensors", "transformer.h.83.input_layernorm.weight": "model-00043-of-00050.safetensors", "transformer.h.83.post_attention_layernorm.weight": "model-00043-of-00050.safetensors", "transformer.h.83.self_attention.dense.weight": "model-00043-of-00050.safetensors", "transformer.h.83.self_attention.dense.bias": "model-00043-of-00050.safetensors", "transformer.h.83.self_attention.query.weight": "model-00043-of-00050.safetensors", "transformer.h.83.self_attention.key_value.weight": "model-00043-of-00050.safetensors", "transformer.h.83.mlp.gate_proj.weight": "model-00043-of-00050.safetensors", "transformer.h.83.mlp.down_proj.weight": "model-00043-of-00050.safetensors", "transformer.h.83.mlp.down_proj.bias": "model-00043-of-00050.safetensors", "transformer.h.83.mlp.up_proj.weight": "model-00043-of-00050.safetensors", "transformer.h.84.input_layernorm.weight": "model-00043-of-00050.safetensors", "transformer.h.84.post_attention_layernorm.weight": "model-00043-of-00050.safetensors", "transformer.h.84.self_attention.dense.weight": "model-00043-of-00050.safetensors", "transformer.h.84.self_attention.dense.bias": "model-00043-of-00050.safetensors", "transformer.h.84.self_attention.query.weight": "model-00043-of-00050.safetensors", "transformer.h.84.self_attention.key_value.weight": "model-00043-of-00050.safetensors", "transformer.h.84.mlp.gate_proj.weight": "model-00043-of-00050.safetensors", "transformer.h.84.mlp.down_proj.weight": "model-00043-of-00050.safetensors", "transformer.h.84.mlp.down_proj.bias": "model-00043-of-00050.safetensors", "transformer.h.84.mlp.up_proj.weight": "model-00043-of-00050.safetensors", "transformer.h.85.input_layernorm.weight": "model-00044-of-00050.safetensors", "transformer.h.85.post_attention_layernorm.weight": "model-00044-of-00050.safetensors", "transformer.h.85.self_attention.dense.weight": "model-00044-of-00050.safetensors", "transformer.h.85.self_attention.dense.bias": "model-00044-of-00050.safetensors", "transformer.h.85.self_attention.query.weight": "model-00044-of-00050.safetensors", "transformer.h.85.self_attention.key_value.weight": "model-00044-of-00050.safetensors", "transformer.h.85.mlp.gate_proj.weight": "model-00044-of-00050.safetensors", "transformer.h.85.mlp.down_proj.weight": "model-00044-of-00050.safetensors", "transformer.h.85.mlp.down_proj.bias": "model-00044-of-00050.safetensors", "transformer.h.85.mlp.up_proj.weight": "model-00044-of-00050.safetensors", "transformer.h.86.input_layernorm.weight": "model-00044-of-00050.safetensors", "transformer.h.86.post_attention_layernorm.weight": "model-00044-of-00050.safetensors", "transformer.h.86.self_attention.dense.weight": "model-00044-of-00050.safetensors", "transformer.h.86.self_attention.dense.bias": "model-00044-of-00050.safetensors", "transformer.h.86.self_attention.query.weight": "model-00044-of-00050.safetensors", "transformer.h.86.self_attention.key_value.weight": "model-00044-of-00050.safetensors", "transformer.h.86.mlp.gate_proj.weight": "model-00044-of-00050.safetensors", "transformer.h.86.mlp.down_proj.weight": "model-00044-of-00050.safetensors", "transformer.h.86.mlp.down_proj.bias": "model-00044-of-00050.safetensors", "transformer.h.86.mlp.up_proj.weight": "model-00044-of-00050.safetensors", "transformer.h.87.input_layernorm.weight": "model-00045-of-00050.safetensors", "transformer.h.87.post_attention_layernorm.weight": "model-00045-of-00050.safetensors", "transformer.h.87.self_attention.dense.weight": "model-00045-of-00050.safetensors", "transformer.h.87.self_attention.dense.bias": "model-00045-of-00050.safetensors", "transformer.h.87.self_attention.query.weight": "model-00045-of-00050.safetensors", "transformer.h.87.self_attention.key_value.weight": "model-00045-of-00050.safetensors", "transformer.h.87.mlp.gate_proj.weight": "model-00045-of-00050.safetensors", "transformer.h.87.mlp.down_proj.weight": "model-00045-of-00050.safetensors", "transformer.h.87.mlp.down_proj.bias": "model-00045-of-00050.safetensors", "transformer.h.87.mlp.up_proj.weight": "model-00045-of-00050.safetensors", "transformer.h.88.input_layernorm.weight": "model-00045-of-00050.safetensors", "transformer.h.88.post_attention_layernorm.weight": "model-00045-of-00050.safetensors", "transformer.h.88.self_attention.dense.weight": "model-00045-of-00050.safetensors", "transformer.h.88.self_attention.dense.bias": "model-00045-of-00050.safetensors", "transformer.h.88.self_attention.query.weight": "model-00045-of-00050.safetensors", "transformer.h.88.self_attention.key_value.weight": "model-00045-of-00050.safetensors", "transformer.h.88.mlp.gate_proj.weight": "model-00045-of-00050.safetensors", "transformer.h.88.mlp.down_proj.weight": "model-00045-of-00050.safetensors", "transformer.h.88.mlp.down_proj.bias": "model-00045-of-00050.safetensors", "transformer.h.88.mlp.up_proj.weight": "model-00045-of-00050.safetensors", "transformer.h.89.input_layernorm.weight": "model-00046-of-00050.safetensors", "transformer.h.89.post_attention_layernorm.weight": "model-00046-of-00050.safetensors", "transformer.h.89.self_attention.dense.weight": "model-00046-of-00050.safetensors", "transformer.h.89.self_attention.dense.bias": "model-00046-of-00050.safetensors", "transformer.h.89.self_attention.query.weight": "model-00046-of-00050.safetensors", "transformer.h.89.self_attention.key_value.weight": "model-00046-of-00050.safetensors", "transformer.h.89.mlp.gate_proj.weight": "model-00046-of-00050.safetensors", "transformer.h.89.mlp.down_proj.weight": "model-00046-of-00050.safetensors", "transformer.h.89.mlp.down_proj.bias": "model-00046-of-00050.safetensors", "transformer.h.89.mlp.up_proj.weight": "model-00046-of-00050.safetensors", "transformer.h.90.input_layernorm.weight": "model-00046-of-00050.safetensors", "transformer.h.90.post_attention_layernorm.weight": "model-00046-of-00050.safetensors", "transformer.h.90.self_attention.dense.weight": "model-00046-of-00050.safetensors", "transformer.h.90.self_attention.dense.bias": "model-00046-of-00050.safetensors", "transformer.h.90.self_attention.query.weight": "model-00046-of-00050.safetensors", "transformer.h.90.self_attention.key_value.weight": "model-00046-of-00050.safetensors", "transformer.h.90.mlp.gate_proj.weight": "model-00046-of-00050.safetensors", "transformer.h.90.mlp.down_proj.weight": "model-00046-of-00050.safetensors", "transformer.h.90.mlp.down_proj.bias": "model-00046-of-00050.safetensors", "transformer.h.90.mlp.up_proj.weight": "model-00046-of-00050.safetensors", "transformer.h.91.input_layernorm.weight": "model-00047-of-00050.safetensors", "transformer.h.91.post_attention_layernorm.weight": "model-00047-of-00050.safetensors", "transformer.h.91.self_attention.dense.weight": "model-00047-of-00050.safetensors", "transformer.h.91.self_attention.dense.bias": "model-00047-of-00050.safetensors", "transformer.h.91.self_attention.query.weight": "model-00047-of-00050.safetensors", "transformer.h.91.self_attention.key_value.weight": "model-00047-of-00050.safetensors", "transformer.h.91.mlp.gate_proj.weight": "model-00047-of-00050.safetensors", "transformer.h.91.mlp.down_proj.weight": "model-00047-of-00050.safetensors", "transformer.h.91.mlp.down_proj.bias": "model-00047-of-00050.safetensors", "transformer.h.91.mlp.up_proj.weight": "model-00047-of-00050.safetensors", "transformer.h.92.input_layernorm.weight": "model-00047-of-00050.safetensors", "transformer.h.92.post_attention_layernorm.weight": "model-00047-of-00050.safetensors", "transformer.h.92.self_attention.dense.weight": "model-00047-of-00050.safetensors", "transformer.h.92.self_attention.dense.bias": "model-00047-of-00050.safetensors", "transformer.h.92.self_attention.query.weight": "model-00047-of-00050.safetensors", "transformer.h.92.self_attention.key_value.weight": "model-00047-of-00050.safetensors", "transformer.h.92.mlp.gate_proj.weight": "model-00047-of-00050.safetensors", "transformer.h.92.mlp.down_proj.weight": "model-00047-of-00050.safetensors", "transformer.h.92.mlp.down_proj.bias": "model-00047-of-00050.safetensors", "transformer.h.92.mlp.up_proj.weight": "model-00047-of-00050.safetensors", "transformer.h.93.input_layernorm.weight": "model-00048-of-00050.safetensors", "transformer.h.93.post_attention_layernorm.weight": "model-00048-of-00050.safetensors", "transformer.h.93.self_attention.dense.weight": "model-00048-of-00050.safetensors", "transformer.h.93.self_attention.dense.bias": "model-00048-of-00050.safetensors", "transformer.h.93.self_attention.query.weight": "model-00048-of-00050.safetensors", "transformer.h.93.self_attention.key_value.weight": "model-00048-of-00050.safetensors", "transformer.h.93.mlp.gate_proj.weight": "model-00048-of-00050.safetensors", "transformer.h.93.mlp.down_proj.weight": "model-00048-of-00050.safetensors", "transformer.h.93.mlp.down_proj.bias": "model-00048-of-00050.safetensors", "transformer.h.93.mlp.up_proj.weight": "model-00048-of-00050.safetensors", "transformer.h.94.input_layernorm.weight": "model-00048-of-00050.safetensors", "transformer.h.94.post_attention_layernorm.weight": "model-00048-of-00050.safetensors", "transformer.h.94.self_attention.dense.weight": "model-00048-of-00050.safetensors", "transformer.h.94.self_attention.dense.bias": "model-00048-of-00050.safetensors", "transformer.h.94.self_attention.query.weight": "model-00048-of-00050.safetensors", "transformer.h.94.self_attention.key_value.weight": "model-00048-of-00050.safetensors", "transformer.h.94.mlp.gate_proj.weight": "model-00048-of-00050.safetensors", "transformer.h.94.mlp.down_proj.weight": "model-00048-of-00050.safetensors", "transformer.h.94.mlp.down_proj.bias": "model-00048-of-00050.safetensors", "transformer.h.94.mlp.up_proj.weight": "model-00048-of-00050.safetensors", "transformer.h.95.input_layernorm.weight": "model-00049-of-00050.safetensors", "transformer.h.95.post_attention_layernorm.weight": "model-00049-of-00050.safetensors", "transformer.h.95.self_attention.dense.weight": "model-00049-of-00050.safetensors", "transformer.h.95.self_attention.dense.bias": "model-00049-of-00050.safetensors", "transformer.h.95.self_attention.query.weight": "model-00049-of-00050.safetensors", "transformer.h.95.self_attention.key_value.weight": "model-00049-of-00050.safetensors", "transformer.h.95.mlp.gate_proj.weight": "model-00049-of-00050.safetensors", "transformer.h.95.mlp.down_proj.weight": "model-00049-of-00050.safetensors", "transformer.h.95.mlp.down_proj.bias": "model-00049-of-00050.safetensors", "transformer.h.95.mlp.up_proj.weight": "model-00049-of-00050.safetensors", "transformer.ln_f.weight": "model-00050-of-00050.safetensors", "lm_head.weight": "model-00050-of-00050.safetensors"}}

tokenizer_config.json

@@ -0,0 +1,39 @@
+{
+  "tokenizer_class": "TelechatTokenizer",
+  "auto_map": {
+    "AutoTokenizer": [
+      "tokenization_telechat.TelechatTokenizer",
+      null
+    ]
+  },
+  "add_bos_token": false,
+  "add_eos_token": false,
+  "use_fast": false,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": {
+    "__type": "AddedToken",
+    "content": "<_end>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": true
+  },
+  "model_max_length": 100000000,
+  "sp_model_kwargs": {},
+  "pad_token": {
+    "__type": "AddedToken",
+    "content": "<_pad>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": true
+  },
+  "unk_token": {
+    "__type": "AddedToken",
+    "content": "<_end>",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": true
+  }
+}