Add Devstral Small 1.1

keras_hub/src/models/mistral/mistral_presets.py

@@ -42,4 +42,12 @@
         },
         "kaggle_handle": "kaggle://keras/mistral/keras/mistral_0.3_instruct_7b_en/1",
     },
+    "devstral_small_1_1": {
+        "metadata": {
+            "description": "Devstral Small 1.1 finetuned from Mistral-Small-3.1 24B base model",
+            "params": 23572403200,
+            "path": "devstral_small_1_1",
+        },
+        # "kaggle_handle": "kaggle://keras/mistral/keras/devstral_small_1_1/1",
+    },
 }

keras_hub/src/utils/transformers/convert_mistral.py

@@ -113,4 +113,6 @@ def convert_weights(backbone, loader, transformers_config):
 
 
 def convert_tokenizer(cls, preset, **kwargs):
+    if "devstral" in preset.lower():
+        preset = "mistralai/Mistral-Small-24B-Base-2501"
     return cls(get_file(preset, "tokenizer.model"), **kwargs)

tools/checkpoint_conversion/convert_mistral_checkpoints.py

@@ -22,6 +22,7 @@
     "mistral_instruct_7b_en": "mistralai/Mistral-7B-Instruct-v0.1",
     "mistral_0.2_instruct_7b_en": "mistralai/Mistral-7B-Instruct-v0.2",
     "mistral_0.3_instruct_7b_en": "mistralai/Mistral-7B-Instruct-v0.3",
+    "devstral_small_1_1": "mistralai/Devstral-Small-2507",
 }
 
 FLAGS = flags.FLAGS
@@ -220,7 +221,10 @@ def main(_):
     try:
         # === Load the Huggingface model ===
         hf_model = MistralForCausalLM.from_pretrained(hf_preset)
-        hf_tokenizer = AutoTokenizer.from_pretrained(hf_preset)
+        if "devstral" in hf_preset.lower():
+            hf_tokenizer = AutoTokenizer.from_pretrained("mistralai/Mistral-Small-24B-Base-2501")
+        else:   
+            hf_tokenizer = AutoTokenizer.from_pretrained(hf_preset)
         hf_model.eval()
         print("\n-> Huggingface model and tokenizer loaded")
 
@@ -239,7 +243,10 @@ def main(_):
         )
         keras_hub_backbone = MistralBackbone(**backbone_kwargs)
 
-        keras_hub_tokenizer = MistralTokenizer.from_preset(f"hf://{hf_preset}")
+        if "devstral" in hf_preset.lower():
+            keras_hub_tokenizer = MistralTokenizer.from_preset("hf://mistralai/Mistral-Small-24B-Base-2501")
+        else:
+            keras_hub_tokenizer = MistralTokenizer.from_preset(f"hf://{hf_preset}")
         print("\n-> Keras 3 model and tokenizer loaded.")
 
         # === Port the weights ===