Update src/streamlit_app.py

src/streamlit_app.py

@@ -131,82 +131,76 @@ def calculate_quantized_size(base_size_str, quant_format):
 LLM_DATABASE = {
     "ultra_low": {  # ≤2GB
         "general": [
-      { name: "TinyLlama-1.1B-Chat", size: "2.2GB", description: "Ultra-compact conversational model" },
-      { name: "DistilBERT-base", size: "0.3GB", description: "Efficient BERT variant for NLP tasks" },
-      { name: "all-MiniLM-L6-v2", size: "0.1GB", description: "Sentence embeddings specialist" },
-      { name: "OPT-125M", size: "0.5GB", description: "Meta's lightweight language model" },
-      { name: "GPT-Neo-125M", size: "0.5GB", description: "EleutherAI's compact model" },
-      { name: "DistilGPT-2", size: "0.3GB", description: "Distilled version of GPT-2" },
-      { name: "MobileBERT", size: "0.2GB", description: "Google's mobile-optimized BERT" },
-      { name: "ALBERT-base", size: "0.4GB", description: "A Lite BERT for self-supervised learning" },
-      { name: "RoBERTa-base", size: "0.5GB", description: "Robustly optimized BERT pretraining" },
-      { name: "ELECTRA-small", size: "0.2GB", description: "Efficiently learning encoder representations" },
-      { name: "MobileLLaMA-1B", size: "1.0GB", description: "Mobile-optimized Llama variant" },
-      { name: "GPT-2-small", size: "0.5GB", description: "OpenAI's original small model" },
-      { name: "T5-small", size: "0.2GB", description: "Text-to-Text Transfer Transformer" },
-      { name: "FLAN-T5-small", size: "0.3GB", description: "Instruction-tuned T5" },
-      { name: "UL2-small", size: "0.8GB", description: "Unified Language Learner" },
-      { name: "DeBERTa-v3-small", size: "0.4GB", description: "Microsoft's enhanced BERT" },
-      { name: "CANINE-s", size: "0.5GB", description: "Character-level model" },
-      { name: "Longformer-base", size: "0.6GB", description: "Long document understanding" },
-      { name: "BigBird-small", size: "0.7GB", description: "Sparse attention model" },
-      { name: "Reformer-small", size: "0.3GB", description: "Memory-efficient transformer" },
-      { name: "FNet-small", size: "0.4GB", description: "Fourier transform model" },
-      { name: "Synthesizer-small", size: "0.3GB", description: "Synthetic attention patterns" },
-      { name: "GPT-Neo-1.3B", size: "1.3GB", description: "EleutherAI's 1.3B model" },
-      { name: "OPT-350M", size: "0.7GB", description: "Meta's 350M parameter model" },
-      { name: "BLOOM-560M", size: "1.1GB", description: "BigScience's small multilingual" }
+  { "name": "TinyLlama-1.1B-Chat", "size": "2.2GB", "description": "Ultra-compact conversational model" },
+  { "name": "DistilBERT-base", "size": "0.3GB", "description": "Efficient BERT variant for NLP tasks" },
+  { "name": "all-MiniLM-L6-v2", "size": "0.1GB", "description": "Sentence embeddings specialist" },
+  { "name": "OPT-125M", "size": "0.5GB", "description": "Meta's lightweight language model" },
+  { "name": "GPT-Neo-125M", "size": "0.5GB", "description": "EleutherAI's compact model" },
+  { "name": "DistilGPT-2", "size": "0.3GB", "description": "Distilled version of GPT-2" },
+  { "name": "MobileBERT", "size": "0.2GB", "description": "Google's mobile-optimized BERT" },
+  { "name": "ALBERT-base", "size": "0.4GB", "description": "A Lite BERT for self-supervised learning" },
+  { "name": "RoBERTa-base", "size": "0.5GB", "description": "Robustly optimized BERT pretraining" },
+  { "name": "ELECTRA-small", "size": "0.2GB", "description": "Efficiently learning encoder representations" },
+  { "name": "MobileLLaMA-1B", "size": "1.0GB", "description": "Mobile-optimized Llama variant" },
+  { "name": "GPT-2-small", "size": "0.5GB", "description": "OpenAI's original small model" },
+  { "name": "T5-small", "size": "0.2GB", "description": "Text-to-Text Transfer Transformer" },
+  { "name": "FLAN-T5-small", "size": "0.3GB", "description": "Instruction-tuned T5" },
+  { "name": "UL2-small", "size": "0.8GB", "description": "Unified Language Learner" },
+  { "name": "DeBERTa-v3-small", "size": "0.4GB", "description": "Microsoft's enhanced BERT" },
+  { "name": "CANINE-s", "size": "0.5GB", "description": "Character-level model" },
+  { "name": "Longformer-base", "size": "0.6GB", "description": "Long document understanding" },
+  { "name": "BigBird-small", "size": "0.7GB", "description": "Sparse attention model" },
+  { "name": "Reformer-small", "size": "0.3GB", "description": "Memory-efficient transformer" },
+  { "name": "FNet-small", "size": "0.4GB", "description": "Fourier transform model" },
+  { "name": "Synthesizer-small", "size": "0.3GB", "description": "Synthetic attention patterns" },
+  { "name": "GPT-Neo-1.3B", "size": "1.3GB", "description": "EleutherAI's 1.3B model" },
+  { "name": "OPT-350M", "size": "0.7GB", "description": "Meta's 350M parameter model" },
+  { "name": "BLOOM-560M", "size": "1.1GB", "description": "BigScience's small multilingual" }
+
+
     ],
         "code": [
-      { name: "CodeT5-small", size: "0.3GB", description: "Compact code generation model" },
-      { name: "Replit-code-v1-3B", size: "1.2GB", description: "Code completion specialist" },
-      { name: "UnixCoder-base", size: "0.5GB", description: "Microsoft's code understanding model" },
-      { name: "CodeBERT-base", size: "0.5GB", description: "Bimodal pre-trained model for programming" },
-      { name: "GraphCodeBERT-base", size: "0.5GB", description: "Pre-trained model with data flow" },
-      { name: "CodeT5-base", size: "0.9GB", description: "Identifier-aware unified pre-trained encoder-decoder" },
-      { name: "PyCodeGPT-110M", size: "0.4GB", description: "Python code generation specialist" },
-      { name: "CodeParrot-110M", size: "0.4GB", description: "GPT-2 model trained on Python code" },
-      { name: "CodeSearchNet-small", size: "0.6GB", description: "Code search and understanding" },
-      { name: "CuBERT-small", size: "0.4GB", description: "Google's code understanding" },
-      { name: "CodeGPT-small", size: "0.5GB", description: "Microsoft's code GPT" },
-      { name: "PLBART-small", size: "0.7GB", description: "Programming language BART" },
-      { name: "TreeBERT-small", size: "0.6GB", description: "Tree-based code representation" },
-      { name: "CoTexT-small", size: "0.5GB", description: "Code and text pre-training" },
-      { name: "SynCoBERT-small", size: "0.6GB", description: "Syntax-guided code BERT" },
+      
+  { "name": "CodeT5-small", "size": "0.3GB", "description": "Compact code generation model" },
+  { "name": "Replit-code-v1-3B", "size": "1.2GB", "description": "Code completion specialist" },
+  { "name": "UnixCoder-base", "size": "0.5GB", "description": "Microsoft's code understanding model" },
+  { "name": "CodeBERT-base", "size": "0.5GB", "description": "Bimodal pre-trained model for programming" },
+  { "name": "GraphCodeBERT-base", "size": "0.5GB", "description": "Pre-trained model with data flow" },
+  { "name": "CodeT5-base", "size": "0.9GB", "description": "Identifier-aware unified pre-trained encoder-decoder" },
+  { "name": "PyCodeGPT-110M", "size": "0.4GB", "description": "Python code generation specialist" },
+  { "name": "CodeParrot-110M", "size": "0.4GB", "description": "GPT-2 model trained on Python code" },
+  { "name": "CodeSearchNet-small", "size": "0.6GB", "description": "Code search and understanding" },
+  { "name": "CuBERT-small", "size": "0.4GB", "description": "Google's code understanding" },
+  { "name": "CodeGPT-small", "size": "0.5GB", "description": "Microsoft's code GPT" },
+  { "name": "PLBART-small", "size": "0.7GB", "description": "Programming language BART" },
+  { "name": "TreeBERT-small", "size": "0.6GB", "description": "Tree-based code representation" },
+  { "name": "CoTexT-small", "size": "0.5GB", "description": "Code and text pre-training" },
+  { "name": "SynCoBERT-small", "size": "0.6GB", "description": "Syntax-guided code BERT" }
+
+
         ]
     },
     "low": {  # 3-4GB
         "general": [
-      { name: "Phi-1.5", size: "2.8GB", description: "Microsoft's efficient reasoning model" },
-      { name: "Gemma-2B", size: "1.4GB", description: "Google's compact foundation model" },
-      { name: "OpenLLaMA-3B", size: "2.1GB", description: "Open source LLaMA reproduction" },
-      { name: "RedPajama-3B", size: "2.0GB", description: "Together AI's open model" },
-      { name: "StableLM-3B", size: "2.3GB", description: "Stability AI's language model" },
-      { name: "Pythia-2.8B", size: "2.8GB", description: "EleutherAI's training suite model" },
-      { name: "GPT-Neo-2.7B", size: "2.7GB", description: "EleutherAI's open GPT model" },
-      { name: "OPT-2.7B", size: "2.7GB", description: "Meta's open pre-trained transformer" },
-      { name: "BLOOM-3B", size: "3.0GB", description: "BigScience's multilingual model" },
-      { name: "GPT-J-6B", size: "3.5GB", description: "EleutherAI's 6B parameter model" },
-      { name: "Cerebras-GPT-2.7B", size: "2.7GB", description: "Cerebras Systems' open model" },
-      { name: "PaLM-2B", size: "2.0GB", description: "Google's Pathways Language Model" },
-      { name: "LaMDA-2B", size: "2.2GB", description: "Google's Language Model for Dialogue" },
-      { name: "FairSeq-2.7B", size: "2.7GB", description: "Facebook's sequence-to-sequence toolkit" },
-      { name: "Megatron-2.5B", size: "2.5GB", description: "NVIDIA's transformer model" },
-      { name: "GLM-2B", size: "2.0GB", description: "General Language Model pretraining" },
-      { name: "CPM-2", size: "2.6GB", description: "Chinese Pre-trained Language Model" },
-      { name: "mT5-small", size: "1.2GB", description: "Multilingual Text-to-Text Transfer" },
-      { name: "ByT5-small", size: "1.5GB", description: "Byte-level Text-to-Text Transfer" },
-      { name: "Switch-2B", size: "2.0GB", description: "Switch Transformer sparse model" },
-      { name: "GPT-NeoX-2B", size: "2.0GB", description: "EleutherAI's NeoX architecture" },
-      { name: "OPT-1.3B", size: "1.3GB", description: "Meta's 1.3B parameter model" },
-      { name: "BLOOM-1B7", size: "1.7GB", description: "BigScience's 1.7B model" },
-      { name: "Pythia-1.4B", size: "1.4GB", description: "EleutherAI's 1.4B model" },
-      { name: "StableLM-Alpha-3B", size: "2.2GB", description: "Stability AI's alpha model" },
-      { name: "OpenLLM-3B", size: "2.1GB", description: "Open-sourced language model" },
-      { name: "Dolly-v1-6B", size: "3.0GB", description: "Databricks' instruction model" },
-      { name: "GPT4All-J-6B", size: "3.2GB", description: "Nomic AI's assistant model" },
-      { name: "Vicuna-3B", size: "2.1GB", description: "UC Berkeley's 3B chat model" },
-      { name: "Alpaca-3B", size: "2.0GB", description: "Stanford's 3B instruction model" }
+     
+  { "name": "Phi-1.5", "size": "2.8GB", "description": "Microsoft's efficient reasoning model" },
+  { "name": "Gemma-2B", "size": "1.4GB", "description": "Google's compact foundation model" },
+  { "name": "OpenLLaMA-3B", "size": "2.1GB", "description": "Open source LLaMA reproduction" },
+  { "name": "RedPajama-3B", "size": "2.0GB", "description": "Together AI's open model" },
+  { "name": "StableLM-3B", "size": "2.3GB", "description": "Stability AI's language model" },
+  { "name": "Pythia-2.8B", "size": "2.8GB", "description": "EleutherAI's training suite model" },
+  { "name": "GPT-Neo-2.7B", "size": "2.7GB", "description": "EleutherAI's open GPT model" },
+  { "name": "OPT-2.7B", "size": "2.7GB", "description": "Meta's open pre-trained transformer" },
+  { "name": "BLOOM-3B", "size": "3.0GB", "description": "BigScience's multilingual model" },
+  { "name": "GPT-J-6B", "size": "3.5GB", "description": "EleutherAI's 6B parameter model" },
+  { "name": "Cerebras-GPT-2.7B", "size": "2.7GB", "description": "Cerebras Systems' open model" },
+  { "name": "PaLM-2B", "size": "2.0GB", "description": "Google's Pathways Language Model" },
+  { "name": "LaMDA-2B", "size": "2.2GB", "description": "Google's Language Model for Dialogue" },
+  { "name": "FairSeq-2.7B", "size": "2.7GB", "description": "Facebook's sequence-to-sequence toolkit" },
+  { "name": "Megatron-2.5B", "size": "2.5GB", "description": "NVIDIA's transformer model" },
+  { "name": "GLM-2B", "size": "2.0GB", "description": "General Language Model pretraining" },
+  { "name": "CPM-2", "size": "2.6GB", "description": "Chinese"},
+
         ],
         "code": [
             {"name": "CodeGen-2B", "size": "1.8GB", "description": "Salesforce code model", "parameters": "2B", "context": "2K"},