undo PR 20 (#22)

- undo PR 20 (ebaf4f3e76c91ca9e67751c5eb088b71842ccd25)
- Update sentence_bert_config.json (64c8869ce5bf1876b0cd88e4039cbc32ae52c3dc)
- Update tokenizer_config.json (dde01c7ef2499209ac7da7f5c572b15197f74187)


Co-authored-by: Michael <[email protected]>

config.json

@@ -14,7 +14,7 @@
   "hidden_size": 1536,
   "initializer_range": 0.02,
   "intermediate_size": 8960,
-  "max_position_embeddings": 2048,
+  "max_position_embeddings": 131072,
   "max_window_layers": 21,
   "model_type": "qwen2",
   "num_attention_heads": 12,
@@ -22,7 +22,7 @@
   "num_key_value_heads": 2,
   "rms_norm_eps": 1e-06,
   "rope_theta": 1000000.0,
-  "sliding_window": 2048,
+  "sliding_window": 131072,
   "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.41.2",

sentence_bert_config.json

@@ -1,4 +1,4 @@
 {
-  "max_seq_length": 2048,
+  "max_seq_length": 32768,
   "do_lower_case": false
 }

tokenizer_config.json

@@ -39,7 +39,7 @@
   "clean_up_tokenization_spaces": false,
   "eos_token": "<|endoftext|>",
   "errors": "replace",
-  "model_max_length": 2048,
+  "model_max_length": 32768,
   "pad_token": "<|endoftext|>",
   "split_special_tokens": false,
   "tokenizer_class": "Qwen2Tokenizer",