File size: 6,731 Bytes
21434ad 9fe7f69 21434ad 9fe7f69 21434ad 65c2c66 706ebc2 21434ad 706ebc2 21434ad |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 |
{
"model_family": "Helion",
"version": "2.5",
"release_type": "research_and_development",
"variants": {
"base": {
"name": "Helion-2.5-Rnd",
"full_name": "DeepXR/Helion-2.5-Rnd",
"description": "Base research model with full precision (FP16)",
"parameters": "70B",
"precision": "float16",
"context_length": 131072,
"safetensors_shards": 83,
"shard_naming": "shard_00 to shard_82",
"shard_size_gb": 1.69,
"shard_size_gib": 1.57,
"total_size_gb": 140.27,
"status": "active",
"recommended_use": [
"Research",
"Development",
"High-accuracy inference"
],
"hardware_requirements": {
"min_vram_gb": 145,
"min_gpus": 2,
"recommended_gpu": "A100 80GB"
}
},
"instruct": {
"name": "Helion-2.5-Rnd-Instruct",
"full_name": "DeepXR/Helion-2.5-Rnd-Instruct",
"description": "Instruction-tuned variant optimized for following instructions",
"parameters": "70B",
"precision": "bfloat16",
"context_length": 131072,
"status": "planned",
"recommended_use": [
"Instruction following",
"Task completion",
"Structured outputs"
],
"fine_tuning": {
"type": "supervised",
"data_focus": "instruction_pairs"
}
},
"chat": {
"name": "Helion-2.5-Rnd-Chat",
"full_name": "DeepXR/Helion-2.5-Rnd-Chat",
"description": "Conversational variant optimized for multi-turn dialogue",
"parameters": "70B",
"precision": "bfloat16",
"context_length": 131072,
"status": "planned",
"recommended_use": [
"Conversational AI",
"Customer service",
"Interactive applications"
],
"fine_tuning": {
"type": "rlhf",
"data_focus": "conversational_data"
}
},
"code": {
"name": "Helion-2.5-Rnd-Code",
"full_name": "DeepXR/Helion-2.5-Rnd-Code",
"description": "Code-specialized variant with enhanced programming capabilities",
"parameters": "70B",
"precision": "bfloat16",
"context_length": 131072,
"status": "planned",
"recommended_use": [
"Code generation",
"Code review",
"Bug fixing",
"Documentation"
],
"fine_tuning": {
"type": "supervised",
"data_focus": "code_repositories"
},
"enhanced_languages": [
"Python",
"JavaScript",
"TypeScript",
"Rust",
"Go",
"Java"
]
},
"math": {
"name": "Helion-2.5-Rnd-Math",
"full_name": "DeepXR/Helion-2.5-Rnd-Math",
"description": "Mathematics-specialized variant for advanced problem solving",
"parameters": "70B",
"precision": "bfloat16",
"context_length": 131072,
"status": "planned",
"recommended_use": [
"Mathematical reasoning",
"Proof generation",
"Problem solving",
"Educational applications"
],
"fine_tuning": {
"type": "supervised",
"data_focus": "mathematical_proofs"
}
}
},
"deployment_configurations": {
"production": {
"description": "Production-ready configuration with optimizations",
"settings": {
"tensor_parallel_size": 4,
"gpu_memory_utilization": 0.95,
"max_batch_size": 32,
"enable_prefix_caching": true,
"enable_chunked_prefill": true
}
},
"development": {
"description": "Development configuration for testing",
"settings": {
"tensor_parallel_size": 2,
"gpu_memory_utilization": 0.85,
"max_batch_size": 8,
"enable_prefix_caching": false,
"enable_chunked_prefill": false
}
},
"research": {
"description": "Research configuration for experimentation",
"settings": {
"tensor_parallel_size": 2,
"gpu_memory_utilization": 0.90,
"max_batch_size": 4,
"enable_prefix_caching": false,
"enable_chunked_prefill": false,
"enable_logging": true
}
}
},
"comparison_matrix": {
"base_vs_instruct": {
"base_advantages": [
"More flexible for fine-tuning",
"Better for creative tasks",
"Less constrained outputs"
],
"instruct_advantages": [
"Better instruction following",
"More structured outputs",
"Improved task completion"
]
},
"base_vs_chat": {
"base_advantages": [
"Better for single-turn tasks",
"More diverse outputs",
"Flexible formatting"
],
"chat_advantages": [
"Better conversation coherence",
"Improved context awareness",
"Natural dialogue flow"
]
}
},
"migration_guide": {
"from_base_to_instruct": {
"steps": [
"Update prompt format to instruction style",
"Adjust temperature (typically lower)",
"Add explicit task descriptions",
"Use structured output formats"
],
"example_prompt_change": {
"base": "Write a function to sort a list",
"instruct": "### Instruction:\nWrite a Python function that sorts a list in ascending order.\n\n### Response:"
}
},
"from_base_to_chat": {
"steps": [
"Convert to chat message format",
"Add system prompts",
"Maintain conversation history",
"Use appropriate message roles"
],
"example_format_change": {
"base": "Hello, how are you?",
"chat": [
{
"role": "system",
"content": "You are a helpful assistant."
},
{
"role": "user",
"content": "Hello, how are you?"
}
]
}
}
},
"version_history": {
"2.5.0-rnd": {
"release_date": "2025-01-30",
"status": "current",
"changes": [
"Initial research release",
"70B parameter model",
"131K context with YARN",
"SafeTensors format (96 shards)",
"Full precision (BF16)"
]
}
},
"roadmap": {
"upcoming_variants": [
{
"name": "Helion-2.5-Rnd-Instruct",
"expected": "Q2 2025",
"status": "in_development"
},
{
"name": "Helion-2.5-Rnd-Chat",
"expected": "Q2 2025",
"status": "planned"
},
{
"name": "Helion-2.5-Rnd-Code",
"expected": "Q3 2025",
"status": "planned"
}
],
"future_features": [
"Multi-modal capabilities",
"Extended context to 256K",
"Improved multilingual support",
"Domain-specific variants"
]
}
} |