Upload model

2025-04-21 06:53:03 +00:00
parent d8525e7660
commit e5e37c1a4d
10 changed files with 363 additions and 393 deletions
--- a/config.json
+++ b/config.json
@@ -1,7 +1,7 @@
 {
-  "_name_or_path": "/mnt/data/models/Qwen2___5-7B-Instruct",
+  "_name_or_path": "/mnt/workspace/DistilQwen25-DS3-0324/7B",
  "architectures": [
-    "Qwen2ForCausalLM"
+    "Qwen2Model"
  ],
  "attention_dropout": 0.0,
  "bos_token_id": 151643,
@@ -21,8 +21,8 @@
  "rope_theta": 1000000.0,
  "sliding_window": null,
  "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
-  "transformers_version": "4.46.1",
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.3",
  "use_cache": false,
  "use_sliding_window": false,
  "vocab_size": 152064