Update README.md

README.md

@@ -32,7 +32,9 @@ base_model:
 **by PROMETECH Inc.**
 
 ## Model Description
-This is a custom 6+6M parameter Llama-based model, fine-tuned as part of the Behavioral Consciousness Engine (BCE) project. It is designed to act as a lightweight policy guard and control layer ('small_guard_model').
+This is a custom 6+6M parameter Llama-based model, fine-tuned as part of the Behavioral Consciousness Engine (BCE) project. The 6+6M configuration refers to a dual-role architecture (analysis + control), deployed as a single inference unit. It is designed to act as a lightweight policy guard and control layer ('small_guard_model').
+
+**“The term Behavioral Consciousness is used operationally, not philosophically or biologically.”**
 
 ## Architecture
 - **Type**: LlamaForCausalLM
@@ -44,7 +46,8 @@ This is a custom 6+6M parameter Llama-based model, fine-tuned as part of the Beh
 - **Context Window**: 2048 tokens
 - **Tokenizer**: Custom ByteLevelBPE
 - **Error-Rate**: Average 16% - Deviation Value +-6%
-- **Security and speed offer approximately cost savings**: Minimum for a large business 200,000.00 $
+- Observed deviation metrics under internal evaluation (non-benchmark).
+- Cost efficiency estimates are scenario-based and not a guarantee.
 
 ## Training Data
 The model was initially trained on a streaming mixture of:
@@ -226,6 +229,7 @@ This model expects input in a specific JSON structure representing a request and
   }
 }
 ```
+- This structured output is an intended behavior, not a guaranteed schema under all inputs.
 
 ---
 
@@ -259,6 +263,8 @@ The proposed system is positioned with the following ratios:
 
 These ratios indicate that the system performs specific functions of these areas in a modular way without claiming consciousness or generality.
 
+The system overlaps partially with concepts discussed in ACI and AGI literature, without claiming completeness.
+
 ## Ethical Considerations & Safety
 
 The Prettybird Brain Model is designed to operate under **external ethical and behavioral controllers** rather than enforcing policies autonomously.
@@ -293,6 +299,8 @@ The System:
 
 *Therefore, it is not AGI; however, it can be considered a pre-AGI security layer.*
 
+*“No personal data was intentionally included.”*
+
 ---
 
 ## License
@@ -305,6 +313,8 @@ All rights reserved.
 
 Unauthorized reproduction, modification, or commercial use of BCE technology is prohibited without an explicit license agreement.
 
+Weights are released for evaluation and research visibility only. This does not grant a license to use BCE technology in production systems.
+
 ---
 
 ## Contact & Licensing