Add model-index with benchmark evaluations

Added structured evaluation results from README benchmark table:

**Automated Benchmarks:**
- MMLU: 55.23
- GPQA: 31.47
- IFEval (Instruction following): 74.89
- IFBench: 20.7
- GSM8K (Math reasoning): 58.3
- MGSM (Multilingual math): 55.04
- MMMLU (Multilingual MMLU): 46.73

Total: 7 benchmarks across reasoning, instruction-following, and multilingual capabilities.

This enables the model to appear in leaderboards and makes it easier to compare with other models.

Note: PR #6 (Support tool calls) modifies the tokenizer configuration and should not conflict with this metadata addition.

README.md

@@ -17,6 +17,39 @@ tags:
 - liquid
 - lfm2
 - edge
+model-index:
+- name: LFM2-1.2B
+  results:
+  - task:
+      type: text-generation
+    dataset:
+      name: Automated Benchmarks
+      type: benchmark
+    metrics:
+    - name: MMLU
+      type: mmlu
+      value: 55.23
+    - name: GPQA
+      type: gpqa
+      value: 31.47
+    - name: IFEval
+      type: ifeval
+      value: 74.89
+    - name: IFBench
+      type: ifbench
+      value: 20.7
+    - name: GSM8K
+      type: gsm8k
+      value: 58.3
+    - name: MGSM
+      type: mgsm
+      value: 55.04
+    - name: MMMLU
+      type: mmmlu
+      value: 46.73
+    source:
+      name: Model README - Automated Benchmarks
+      url: https://huggingface.co/LiquidAI/LFM2-1.2B
 ---
 
 <center>