🔥 Modèle entièrement recréé - Stable et fonctionnel

config.json

@@ -17,14 +17,13 @@
   "num_attention_heads": 16,
   "num_hidden_layers": 16,
   "num_key_value_heads": 4,
-  "pad_token_id": 0,
   "pretraining_tp": 1,
   "rms_norm_eps": 1e-06,
   "rope_scaling": null,
   "rope_theta": 10000.0,
-  "tie_word_embeddings": true,
+  "tie_word_embeddings": false,
   "torch_dtype": "float32",
   "transformers_version": "4.53.3",
-  "use_cache": false,
-  "vocab_size": 16000
+  "use_cache": true,
+  "vocab_size": 32000
 }

generation_config.json

@@ -2,6 +2,5 @@
   "_from_model_config": true,
   "bos_token_id": 1,
   "eos_token_id": 2,
-  "pad_token_id": 0,
   "transformers_version": "4.53.3"
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:975a1be47bae9f26e876054b60b456d20ded1fecb27bb3987bd1e24a42cd8642
-size 2966708480
+oid sha256:fd051e3e5ef8ea0341c729845115363ce44351bdc7e3b6be92251b419209b947
+size 3359924616

tokenizer_config.json

@@ -37,6 +37,7 @@
   "pad_token": "</s>",
   "padding_side": "right",
   "sp_model_kwargs": {},
+  "spaces_between_special_tokens": false,
   "tokenizer_class": "LlamaTokenizer",
   "unk_token": "<unk>",
   "use_default_system_prompt": false