Update README.md

README.md

@@ -28,35 +28,28 @@ That second number matters a lot. Low KL divergence means the model's "brain" di
 
 ---
 
-## Base Model Capabilities (inherited from GRM-2.6-Plus)
-
-GRM-2.6-Plus already punches well above its weight class for a 27B model — competitive with much larger systems on reasoning and coding benchmarks. Primal inherits all of that:
-
-- Elite step-by-step reasoning across complex domains
-- Strong code generation and agentic/tool-use workflows
-- High performance on math, science, and general knowledge
-- Practical for local deployment on capable consumer hardware
-
----
-
 ## Usage
 
 ```python
 from vllm import LLM, SamplingParams
 
+# Sampling configuration
 sampling_params = SamplingParams(
     temperature=1.0,
     top_p=0.95,
-    max_tokens=81920
+    max_tokens=81920,
 )
 
+# Load the model
 llm = LLM(model="morikomorizz/GRM-2.6-Plus-Primal")
 
-messages = [{"role": "user", "content": "Your prompt here"}]
+# Run inference
+messages = [
+    {"role": "user", "content": "Your prompt here"},
+]
 
 outputs = llm.chat(messages, sampling_params=sampling_params)
 
 for output in outputs:
     print(output.outputs[0].text)
-```
-*Based on [OrionLLM/GRM-2.6-Plus](https://huggingface.co/OrionLLM/GRM-2.6-Plus) · Architecture: Qwen3.6-27B · Tensor type: BF16*
+```