增加convert_lm_logits_dty

Llama-2-13b-chat-hf/bert4torch_config.json

@@ -16,5 +16,6 @@
 	"rope_rank": "updown",
     "tie_word_embeddings": false,
 	"torch_dtype": "float16",
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config":  {"skip_special_tokens": true, "add_special_tokens": false},"end_id": 2}
     }

Llama-2-13b-hf/bert4torch_config.json

@@ -16,5 +16,6 @@
 	"rope_rank": "updown",
     "tie_word_embeddings": false,
 	"torch_dtype": "float16",
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config":  {"skip_special_tokens": true, "add_special_tokens": false},"end_id": 2}
     }

Llama-2-7b-chat-hf/bert4torch_config.json

@@ -10,5 +10,6 @@
     "skip_init": true,
     "layer_norm_eps": 1e-6,
     "rope_rank": "updown",
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config":  {"skip_special_tokens": true, "add_special_tokens": false},"end_id": 2}
 }

Llama-2-7b-hf/bert4torch_config.json

@@ -10,5 +10,6 @@
     "skip_init": true,
     "layer_norm_eps": 1e-5,
     "rope_rank": "updown",
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config":  {"skip_special_tokens": true, "add_special_tokens": false},"end_id": 2}
 }

Meta-Llama-3-8B-Instruct/bert4torch_config.json

@@ -16,5 +16,6 @@
 	"torch_dtype": "bfloat16",
 	"tie_word_embeddings": false,
 	"attention_probs_dropout_prob": 0.0,
+    "convert_lm_logits_dtype": "float32",
 	"generation_config": {"tokenizer_config":  {"skip_special_tokens": true, "add_special_tokens": false}, "end_id": [128001, 128009]}
 }

Meta-Llama-3-8B/bert4torch_config.json

@@ -16,5 +16,6 @@
 	"torch_dtype": "bfloat16",
 	"tie_word_embeddings": false,
 	"attention_probs_dropout_prob": 0.0,
+    "convert_lm_logits_dtype": "float32",
 	"generation_config": {"tokenizer_config":  {"skip_special_tokens": true, "add_special_tokens": false}, "end_id": [128001, 128009]}
 }

Qwen1.5-0.5B-Chat/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 21,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen1.5-0.5B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 21,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

Qwen1.5-1.8B-Chat/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 21,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen1.5-1.8B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 21,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

Qwen1.5-14B-Chat/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 35,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen1.5-14B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 35,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

Qwen1.5-7B-Chat/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 28,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen1.5-7B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 28,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

Qwen2-0.5B-Instruct/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 24,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen2-0.5B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 24,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

Qwen2-1.5B-Instruct/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 28,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen2-1.5B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 32768,
 	"max_window_layers": 28,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

Qwen2-7B-Instruct/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 131072,
 	"max_window_layers": 28,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151644, 151645],
                           "max_length": 32768}
 }

Qwen2-7B/bert4torch_config.json

@@ -23,6 +23,7 @@
 	"max_position_embeddings": 32768,
 	"sliding_window": 131072,
 	"max_window_layers": 28,
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "end_id": [151643],
                           "max_length": 32768}
 }

llama-13b/bert4torch_config.json

@@ -12,5 +12,6 @@
 	"segment_vocab_size": 0,
 	"skip_init": true,
 	"rope_rank": "updown",
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "max_length": 2048, "end_id": 2}
 }

llama-7b/bert4torch_config.json

@@ -12,5 +12,6 @@
 	"segment_vocab_size": 0,
 	"skip_init": true,
 	"rope_rank": "updown",
+    "convert_lm_logits_dtype": "float32",
     "generation_config": {"tokenizer_config": {"skip_special_tokens": true}, "max_length": 2048, "end_id": 2}
 }