Release 3x-0.5B-Chat (standard HF interface, no trust_remote_code needed)

config.json

@@ -37,6 +37,7 @@
     "full_attention"
   ],
   "max_position_embeddings": 32768,
+  "max_window_layers": 21,
   "model_type": "threex2",
   "num_attention_heads": 16,
   "num_hidden_layers": 24,
@@ -44,12 +45,12 @@
   "pad_token_id": 151645,
   "rms_norm_eps": 1e-06,
   "rope_parameters": {
-    "rope_theta": 10000.0,
+    "rope_theta": 1000000.0,
     "rope_type": "default"
   },
-  "rope_theta": 10000.0,
-  "sliding_window": null,
-  "tie_word_embeddings": false,
+  "rope_theta": 1000000.0,
+  "sliding_window": 32768,
+  "tie_word_embeddings": true,
   "transformers_version": "5.2.0",
   "use_cache": true,
   "use_sliding_window": false,

generation_config.json

@@ -1,10 +1,12 @@
 {
-  "_from_model_config": true,
   "bos_token_id": 151643,
-  "eos_token_id": 151645,
-  "output_attentions": false,
-  "output_hidden_states": false,
-  "pad_token_id": 151645,
-  "transformers_version": "5.2.0",
-  "use_cache": true
+  "do_sample": true,
+  "eos_token_id": [
+    151645,
+    151643
+  ],
+  "pad_token_id": 151643,
+  "repetition_penalty": 1.1,
+  "top_p": 0.8,
+  "transformers_version": "5.2.0"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:840b3fe9d6ebf7926947f2c0d4fa841c61b66359ce1c8f4b968c026f43d639e4
-size 2478313760
+oid sha256:7a7cc9681a0a7c1d4640ee1296d10045ce3b4000a98d751f35eb5abc3a2d40d3
+size 1855983640

tokenizer_config.json

@@ -14,7 +14,5 @@
   "pad_token": "<|endoftext|>",
   "split_special_tokens": false,
   "tokenizer_class": "Qwen2Tokenizer",
-  "unk_token": null,
-  "chat_template": "{% for message in messages %}{% if loop.first and messages[0]['role'] != 'system' %}{{ '<|im_start|>system\\nYou are a helpful assistant.<|im_end|>\\n' }}{% endif %}{{'<|im_start|>' + message['role'] + '\\n' + message['content'] + '<|im_end|>' + '\\n'}}{% endfor %}{{ '<|im_start|>assistant\\n' }}",
-  "model_type": "threex"
-}
+  "unk_token": null
+}