Helion-V2 / metadata.json
Trouter-Library's picture
Create metadata.json
72b94b9 verified
{
"model_name": "Helion-V2",
"model_version": "2.0",
"model_type": "causal-lm",
"architecture": "helion",
"organization": "DeepXR",
"release_date": "2024-11-15",
"license": "Apache-2.0",
"homepage": "https://huggingface.co/DeepXR/Helion-V2",
"repository": "https://github.com/DeepXR/Helion-V2",
"documentation": "https://docs.deepxr.ai/helion-v2",
"parameters": {
"total": 7200000000,
"trainable": 7200000000,
"non_trainable": 0
},
"model_size": {
"fp32": "28.8 GB",
"fp16": "14.4 GB",
"bfloat16": "14.4 GB",
"int8": "7.2 GB",
"int4": "3.6 GB"
},
"context_length": 8192,
"vocabulary_size": 32768,
"languages": [
"en",
"es",
"fr",
"de",
"it",
"pt",
"nl",
"ru",
"zh",
"ja",
"ko",
"ar",
"hi"
],
"primary_language": "en",
"tasks": [
"text-generation",
"conversational",
"code-generation",
"question-answering",
"summarization",
"translation",
"creative-writing"
],
"framework": {
"name": "transformers",
"version": "4.40.0+",
"backend": "pytorch",
"backend_version": "2.1.0+"
},
"training": {
"dataset_tokens": 2500000000000,
"training_steps": 600000,
"epochs": 3,
"batch_size": 4194304,
"learning_rate": 0.0003,
"optimizer": "AdamW",
"hardware": "128x NVIDIA H100 80GB",
"training_time_days": 21,
"knowledge_cutoff": "2024-10-31"
},
"benchmarks": {
"mmlu": {
"score": 64.2,
"metric": "accuracy",
"shots": 5
},
"humaneval": {
"score": 48.2,
"metric": "pass@1",
"shots": 0
},
"hellaswag": {
"score": 80.5,
"metric": "accuracy",
"shots": 10
},
"truthfulqa": {
"score": 52.1,
"metric": "mc2_accuracy",
"shots": 0
},
"gsm8k": {
"score": 68.7,
"metric": "accuracy",
"shots": 8
},
"arc_challenge": {
"score": 58.3,
"metric": "accuracy",
"shots": 25
},
"mt_bench": {
"score": 7.85,
"metric": "rating",
"shots": 0
}
},
"safety": {
"toxigen_score": 0.08,
"crows_pairs_bias": 54.2,
"content_filtering": true,
"pii_detection": true,
"crisis_detection": true,
"safety_classifiers": [
"hate_speech",
"violence",
"sexual_content",
"self_harm",
"illegal_activity",
"child_safety"
]
},
"deployment": {
"recommended_gpu": [
"NVIDIA A100",
"NVIDIA H100",
"NVIDIA RTX 4090",
"NVIDIA A6000"
],
"minimum_vram_gb": 16,
"recommended_vram_gb": 24,
"supports_quantization": true,
"quantization_formats": [
"int8",
"int4",
"GPTQ",
"AWQ",
"GGUF"
],
"inference_frameworks": [
"transformers",
"vllm",
"llama.cpp",
"text-generation-inference"
]
},
"contact": {
"email": "contact@deepxr.ai",
"github": "https://github.com/DeepXR/Helion-V2",
"twitter": "@DeepXR_AI",
"discord": "https://discord.gg/deepxr"
},
"tags": [
"text-generation",
"conversational",
"code",
"instruction-following",
"pytorch",
"causal-lm",
"llm",
"reasoning",
"multilingual",
"apache-2.0",
"production-ready",
"safe-ai"
],
"checksum": {
"algorithm": "SHA256",
"model_files": "computed_on_upload"
}
}