Upload working DevOps-SLM - compatible with transformers

README.md

@@ -18,7 +18,7 @@ pipeline_tag: text-generation
 DevOps-SLM is a specialized instruction-tuned language model designed exclusively for DevOps tasks, Kubernetes operations, and infrastructure management. This model provides accurate guidance and step-by-step instructions for complex DevOps workflows.
 
 ## Model Details
-- **Base Architecture**: Custom transformer-based causal language model
+- **Base Architecture**: Transformer-based causal language model
 - **Parameters**: 494M (0.5B)
 - **Model Type**: Instruction-tuned for DevOps domain
 - **Max Sequence Length**: 2048 tokens
@@ -55,24 +55,6 @@ response = tokenizer.decode(outputs[0], skip_special_tokens=True)
 print(response)
 ```
 
-### Docker Integration
-```python
-# Generate Dockerfile
-messages = [
-    {"role": "system", "content": "You are a specialized DevOps assistant."},
-    {"role": "user", "content": "Create a Dockerfile for a Node.js application"}
-]
-```
-
-### CI/CD Pipeline Design
-```python
-# Design CI/CD pipeline
-messages = [
-    {"role": "system", "content": "You are a specialized DevOps assistant."},
-    {"role": "user", "content": "Design a CI/CD pipeline for a microservices application"}
-]
-```
-
 ## Examples
 
 ### Kubernetes Deployment
@@ -83,10 +65,6 @@ messages = [
 **Input**: "Create a Dockerfile for a Python Flask application"
 **Output**: Optimized Dockerfile with proper layering and security practices
 
-### Infrastructure Automation
-**Input**: "Create a Terraform configuration for AWS EKS cluster"
-**Output**: Complete Terraform configuration with proper networking and security
-
 ## Performance
 - **Instruction Following**: >90% accuracy on DevOps tasks
 - **YAML Generation**: >95% syntactically correct output
@@ -94,7 +72,7 @@ messages = [
 - **Response Coherence**: High-quality, contextually appropriate responses
 
 ## Model Architecture
-- **Base**: Custom transformer architecture
+- **Base**: Transformer architecture
 - **Attention**: Multi-head self-attention with group query attention
 - **Activation**: SwiGLU activation functions
 - **Normalization**: RMS normalization

config.json

@@ -1,6 +1,6 @@
 {
   "architectures": [
-    "DevOpsInstructSLMForCausalLM"
+    "Qwen2ForCausalLM"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 151643,
@@ -38,7 +38,7 @@
   ],
   "max_position_embeddings": 32768,
   "max_window_layers": 24,
-  "model_type": "devops_instruct_slm",
+  "model_type": "qwen2",
   "num_attention_heads": 14,
   "num_hidden_layers": 24,
   "num_key_value_heads": 2,
@@ -51,11 +51,8 @@
   "use_cache": true,
   "use_sliding_window": false,
   "vocab_size": 151936,
-  "_name_or_path": "devops-slm-base",
+  "_name_or_path": "lakhera2023/devops-slm",
   "custom_model_name": "DevOps-SLM",
   "training_data": "DevOps documentation, Kubernetes examples, and infrastructure guides",
-  "base_architecture": "Custom transformer architecture for DevOps instruction following",
-  "model_family": "DevOps-AI",
-  "domain_specialization": "DevOps, Kubernetes, Docker, CI/CD, Infrastructure",
-  "instruction_tuning": "Specialized for DevOps task completion and guidance"
+  "domain_specialization": "DevOps, Kubernetes, Docker, CI/CD, Infrastructure"
 }

tokenizer_config.json

@@ -38,8 +38,6 @@
   "model_max_length": 32768,
   "pad_token": "<|endoftext|>",
   "split_special_tokens": false,
-  "tokenizer_class": "DevOpsInstructTokenizer",
-  "unk_token": null,
-  "custom_tokenizer": "DevOps Specialized Tokenizer",
-  "domain_optimized": true
-}
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}