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"
    ]
  }
}