Adding Neuron-optimized model files

🤖 Neuron Export Bot: On behalf of [badaoui](https://huggingface.co/badaoui), adding AWS Neuron-optimized model files.
Neuron-optimized models can achieve high-performance inference on AWS Inferentia and Trainium chips. Learn more:
- [AWS Neuron Documentation](https://awsdocs-neuron.readthedocs-hosted.com)
- [🤗 Optimum Neuron Guide](https://huggingface.co/docs/optimum-neuron/index)

.gitattributes

@@ -8,3 +8,4 @@
 *.onnx filter=lfs diff=lfs merge=lfs -text
 *.msgpack filter=lfs diff=lfs merge=lfs -text
 model.safetensors filter=lfs diff=lfs merge=lfs -text
+model.neuron filter=lfs diff=lfs merge=lfs -text

README.md

@@ -8,11 +8,12 @@ metrics:
 tags:
 - bert
 - language-model
-- flaubert 
-- flue 
+- flaubert
+- flue
 - french
 - flaubert-small
 - cased
+- neuron
 ---
 
 # FlauBERT: Unsupervised Language Model Pre-training for French

config.json

@@ -1,4 +1,5 @@
 {
+  "_attn_implementation_autoset": true,
   "amp": 1,
   "architectures": [
     "FlaubertWithLMHeadModel"
@@ -50,6 +51,32 @@
   "n_heads": 8,
   "n_langs": 1,
   "n_layers": 6,
+  "neuron": {
+    "auto_cast": null,
+    "auto_cast_type": null,
+    "compiler_type": "neuronx-cc",
+    "compiler_version": "2.19.8089.0+8ab9f450",
+    "disable_fallback": false,
+    "disable_fast_relayout": false,
+    "dynamic_batch_size": false,
+    "inline_weights_to_neff": true,
+    "input_names": [
+      "input_ids",
+      "attention_mask",
+      "token_type_ids"
+    ],
+    "model_type": "flaubert",
+    "optlevel": "2",
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_names": [
+      "logits"
+    ],
+    "static_batch_size": 1,
+    "static_sequence_length": 128,
+    "task": "fill-mask",
+    "tensor_parallel_size": 1
+  },
   "pad_index": 2,
   "pad_token_id": 2,
   "pre_norm": true,
@@ -63,6 +90,9 @@
   "summary_type": "first",
   "summary_use_proj": true,
   "tokens_per_batch": -1,
+  "torch_dtype": "float32",
+  "torchscript": true,
+  "transformers_version": "4.51.3",
   "unk_index": 3,
   "use_lang_emb": true,
   "vocab_size": 68729,

model.neuron

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15a48c135e21d866e4b13451d42b8c694e015ce6840fda268287f4b0a5587de8
+size 333028477