File size: 9,650 Bytes
d9adc0d
 
19bce9c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d9adc0d
 
 
 
 
faa378a
 
d9adc0d
faa378a
 
 
d9adc0d
 
faa378a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d9adc0d
 
 
faa378a
 
 
d9adc0d
 
 
 
 
 
faa378a
 
d9adc0d
faa378a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d9adc0d
 
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
{
  "metadata": {
    "total_size": 128849018880,
    "format": "safetensors",
    "model_name": "DeepXR/Helion-2.5-Rnd",
    "version": "2.5.0-rnd",
    "precision": "float16",
    "parameters": "70B",
    "total_shards": 82,
    "created_at": "2025-01-30T00:00:00Z",
    "sha256_checksums_available": true,
    "shard_size_bytes": 1686376448,
    "shard_size_gb": 1.57,
    "note": "Model weights distributed across 82 SafeTensors shards (shard_01 to shard_82), each approximately 1.57GB in FP16 precision."
  },
  "weight_map": {
    "model.embed_tokens.weight": "shard_01.safetensors",
    "model.layers.0.self_attn.q_proj.weight": "shard_02.safetensors",
    "model.layers.0.self_attn.k_proj.weight": "shard_02.safetensors",
    "model.layers.0.self_attn.v_proj.weight": "shard_03.safetensors",
    "model.layers.0.self_attn.o_proj.weight": "shard_03.safetensors",
    "model.layers.0.mlp.gate_proj.weight": "shard_04.safetensors",
    "model.layers.0.mlp.up_proj.weight": "shard_04.safetensors",
    "model.layers.0.mlp.down_proj.weight": "shard_05.safetensors",
    "model.layers.0.input_layernorm.weight": "shard_05.safetensors",
    "model.layers.0.post_attention_layernorm.weight": "shard_05.safetensors",
    "model.layers.1.self_attn.q_proj.weight": "shard_06.safetensors",
    "model.layers.1.self_attn.k_proj.weight": "shard_06.safetensors",
    "model.layers.1.self_attn.v_proj.weight": "shard_07.safetensors",
    "model.layers.1.self_attn.o_proj.weight": "shard_07.safetensors",
    "model.layers.1.mlp.gate_proj.weight": "shard_08.safetensors",
    "model.layers.1.mlp.up_proj.weight": "shard_08.safetensors",
    "model.layers.1.mlp.down_proj.weight": "shard_09.safetensors",
    "model.layers.2.self_attn.q_proj.weight": "shard_10.safetensors",
    "model.layers.2.self_attn.k_proj.weight": "shard_10.safetensors",
    "model.layers.2.self_attn.v_proj.weight": "shard_11.safetensors",
    "model.layers.2.self_attn.o_proj.weight": "shard_11.safetensors",
    "model.layers.2.mlp.gate_proj.weight": "shard_12.safetensors",
    "model.layers.2.mlp.up_proj.weight": "shard_12.safetensors",
    "model.layers.2.mlp.down_proj.weight": "shard_13.safetensors",
    "model.layers.3.self_attn.q_proj.weight": "shard_14.safetensors",
    "model.layers.3.self_attn.k_proj.weight": "shard_14.safetensors",
    "model.layers.3.self_attn.v_proj.weight": "shard_15.safetensors",
    "model.layers.3.self_attn.o_proj.weight": "shard_15.safetensors",
    "model.layers.3.mlp.gate_proj.weight": "shard_16.safetensors",
    "model.layers.3.mlp.up_proj.weight": "shard_16.safetensors",
    "model.layers.3.mlp.down_proj.weight": "shard_17.safetensors",
    "model.layers.29.self_attn.q_proj.weight": "shard_74.safetensors",
    "model.layers.29.self_attn.k_proj.weight": "shard_74.safetensors",
    "model.layers.29.self_attn.v_proj.weight": "shard_75.safetensors",
    "model.layers.29.self_attn.o_proj.weight": "shard_75.safetensors",
    "model.layers.29.mlp.gate_proj.weight": "shard_76.safetensors",
    "model.layers.29.mlp.up_proj.weight": "shard_76.safetensors",
    "model.layers.29.mlp.down_proj.weight": "shard_77.safetensors",
    "model.layers.30.self_attn.q_proj.weight": "shard_78.safetensors",
    "model.layers.30.self_attn.k_proj.weight": "shard_78.safetensors",
    "model.layers.30.self_attn.v_proj.weight": "shard_79.safetensors",
    "model.layers.30.self_attn.o_proj.weight": "shard_79.safetensors",
    "model.layers.30.mlp.gate_proj.weight": "shard_80.safetensors",
    "model.layers.30.mlp.up_proj.weight": "shard_80.safetensors",
    "model.layers.30.mlp.down_proj.weight": "shard_81.safetensors",
    "model.layers.31.self_attn.q_proj.weight": "shard_81.safetensors",
    "model.layers.31.self_attn.k_proj.weight": "shard_81.safetensors",
    "model.layers.31.self_attn.v_proj.weight": "shard_82.safetensors",
    "model.layers.31.self_attn.o_proj.weight": "shard_82.safetensors",
    "model.layers.31.mlp.gate_proj.weight": "shard_82.safetensors",
    "model.layers.31.mlp.up_proj.weight": "shard_82.safetensors",
    "model.layers.31.mlp.down_proj.weight": "shard_82.safetensors",
    "model.layers.31.input_layernorm.weight": "shard_82.safetensors",
    "model.layers.31.post_attention_layernorm.weight": "shard_82.safetensors",
    "model.norm.weight": "shard_82.safetensors",
    "lm_head.weight": "shard_82.safetensors"
  },
  "safetensors_info": {
    "description": "SafeTensors format provides secure, fast, and zero-copy tensor serialization",
    "total_shards": 82,
    "shard_naming": "shard_01.safetensors through shard_82.safetensors",
    "parameters": "70B",
    "precision": "float16",
    "shard_size_gb": 1.57,
    "total_size_gb": 120.0,
    "shard_distribution": "Each transformer layer distributed across ~2.5 shards for balanced loading",
    "benefits": [
      "No arbitrary code execution during loading",
      "Lazy loading support for memory efficiency",
      "Fast deserialization without pickle",
      "Tensor metadata validation",
      "Cross-platform compatibility",
      "Memory-mapped file support",
      "Parallel loading across multiple GPUs"
    ],
    "verification": "Each file includes SHA256 checksum for integrity verification",
    "status": "released"
  },
  "shard_structure": {
    "embedding_layer": {
      "shards": ["shard_01.safetensors"],
      "size_gb": 1.57
    },
    "transformer_layers": {
      "layers_per_shard": "~0.39",
      "shards_range": "shard_02.safetensors to shard_81.safetensors",
      "size_per_shard_gb": 1.57
    },
    "output_layer": {
      "shards": ["shard_82.safetensors"],
      "includes": ["model.norm.weight", "lm_head.weight"],
      "size_gb": 1.57
    }
  }
} {
    "total_size": 140737488355328,
    "format": "safetensors",
    "model_name": "DeepXR/Helion-2.5-Rnd",
    "version": "2.5.0-rnd",
    "precision": "bfloat16",
    "parameters": "70B",
    "total_shards": 96,
    "created_at": "2025-01-30T00:00:00Z",
    "sha256_checksums_available": true,
    "shard_size_avg": "1.46GB",
    "note": "SafeTensors shards will be available soon. Model weights distributed across 96 files for optimal loading."
  },
  "weight_map": {
    "model.embed_tokens.weight": "model-00001-of-00096.safetensors",
    "model.layers.0.self_attn.q_proj.weight": "model-00002-of-00096.safetensors",
    "model.layers.0.self_attn.k_proj.weight": "model-00002-of-00096.safetensors",
    "model.layers.0.self_attn.v_proj.weight": "model-00003-of-00096.safetensors",
    "model.layers.0.self_attn.o_proj.weight": "model-00003-of-00096.safetensors",
    "model.layers.0.mlp.gate_proj.weight": "model-00004-of-00096.safetensors",
    "model.layers.0.mlp.up_proj.weight": "model-00004-of-00096.safetensors",
    "model.layers.0.mlp.down_proj.weight": "model-00005-of-00096.safetensors",
    "model.layers.0.input_layernorm.weight": "model-00005-of-00096.safetensors",
    "model.layers.0.post_attention_layernorm.weight": "model-00005-of-00096.safetensors",
    "model.layers.1.self_attn.q_proj.weight": "model-00006-of-00096.safetensors",
    "model.layers.1.self_attn.k_proj.weight": "model-00006-of-00096.safetensors",
    "model.layers.1.self_attn.v_proj.weight": "model-00007-of-00096.safetensors",
    "model.layers.1.self_attn.o_proj.weight": "model-00007-of-00096.safetensors",
    "model.layers.1.mlp.gate_proj.weight": "model-00008-of-00096.safetensors",
    "model.layers.1.mlp.up_proj.weight": "model-00008-of-00096.safetensors",
    "model.layers.1.mlp.down_proj.weight": "model-00009-of-00096.safetensors",
    "model.layers.31.self_attn.q_proj.weight": "model-00092-of-00096.safetensors",
    "model.layers.31.self_attn.k_proj.weight": "model-00093-of-00096.safetensors",
    "model.layers.31.self_attn.v_proj.weight": "model-00093-of-00096.safetensors",
    "model.layers.31.self_attn.o_proj.weight": "model-00094-of-00096.safetensors",
    "model.layers.31.mlp.gate_proj.weight": "model-00094-of-00096.safetensors",
    "model.layers.31.mlp.up_proj.weight": "model-00095-of-00096.safetensors",
    "model.layers.31.mlp.down_proj.weight": "model-00095-of-00096.safetensors",
    "model.layers.31.input_layernorm.weight": "model-00096-of-00096.safetensors",
    "model.layers.31.post_attention_layernorm.weight": "model-00096-of-00096.safetensors",
    "model.norm.weight": "model-00096-of-00096.safetensors",
    "lm_head.weight": "model-00096-of-00096.safetensors"
  },
  "safetensors_info": {
    "description": "SafeTensors format provides secure, fast, and zero-copy tensor serialization",
    "total_shards": 96,
    "parameters": "70B",
    "shard_distribution": "Each transformer layer distributed across ~3 shards for balanced loading",
    "benefits": [
      "No arbitrary code execution during loading",
      "Lazy loading support for memory efficiency",
      "Fast deserialization without pickle",
      "Tensor metadata validation",
      "Cross-platform compatibility",
      "Memory-mapped file support",
      "Parallel loading across multiple GPUs"
    ],
    "verification": "Each file will include SHA256 checksum for integrity verification",
    "status": "In preparation - 96 shards to be released soon"
  },
  "shard_structure": {
    "embedding_layer": {
      "shards": ["model-00001-of-00096.safetensors"],
      "size_estimate": "~1.0GB"
    },
    "transformer_layers": {
      "layers_per_shard": "~0.33",
      "shards_range": "model-00002-of-00096 to model-00095-of-00096",
      "size_estimate": "~1.5GB per shard"
    },
    "output_layer": {
      "shards": ["model-00096-of-00096.safetensors"],
      "includes": ["model.norm.weight", "lm_head.weight"],
      "size_estimate": "~1.0GB"
    }
  }
}