File size: 25,874 Bytes
a9a7525
aec30a9
 
 
 
 
a9a7525
aec30a9
 
 
 
 
a9a7525
19c0773
 
 
 
 
 
 
 
 
8f0017d
 
 
a9a7525
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
aec30a9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a9a7525
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
aec30a9
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
# TinyLLama-4.6M-v0.0-F16.gguf - GGUF Internal File Dump

- Endian: LITTLE endian

## Key Value Metadata Store

There are 40 key-value pairs in this file

| POS | TYPE      | Count | Key                                    | Value                                                                            |
|----:|:----------|------:|:---------------------------------------|:---------------------------------------------------------------------------------|
|   1 | UINT32    |     1 | GGUF.version                           | 3                                                                                |
|   2 | UINT64    |     1 | GGUF.tensor_count                      | 75                                                                               |
|   3 | UINT64    |     1 | GGUF.kv_count                          | 37                                                                               |
|   4 | STRING    |     1 | general.architecture                   | `llama`                                                                          |
|   5 | STRING    |     1 | general.type                           | `model`                                                                          |
|   6 | STRING    |     1 | general.name                           | `TinyLLama`                                                                      |
|   7 | STRING    |     1 | general.author                         | `Maykeye`                                                                        |
|   8 | STRING    |     1 | general.version                        | `v0.0`                                                                           |
|   9 | STRING    |     1 | general.description                    | `This gguf is ported from a fir`...`M but using Llama architecture`              |
|  10 | STRING    |     1 | general.quantized_by                   | `Mofosyne`                                                                       |
|  11 | STRING    |     1 | general.size_label                     | `4.6M`                                                                           |
|  12 | STRING    |     1 | general.license                        | `apache-2.0`                                                                     |
|  13 | STRING    |     1 | general.license.name                   | `Apache License Version 2.0, January 2004`                                       |
|  14 | STRING    |     1 | general.license.link                   | `https://huggingface.co/dataset`...`ob/main/markdown/apache-2.0.md`              |
|  15 | STRING    |     1 | general.url                            | `https://huggingface.co/mofosyne/TinyLLama-v0-llamafile`                         |
|  16 | STRING    |     1 | general.repo_url                       | `https://huggingface.co/mofosyne/TinyLLama-v0-llamafile`                         |
|  17 | STRING    |     1 | general.source.url                     | `https://huggingface.co/Maykeye/TinyLLama-v0`                                    |
|  18 | STRING    |     1 | general.source.repo_url                | `https://huggingface.co/Maykeye/TinyLLama-v0`                                    |
|  19 | [STRING]  |     5 | general.tags                           | [ `text generation`, `transformer`, `llama`, `tiny`, `tiny model` ]              |
|  20 | [STRING]  |     1 | general.languages                      | [ `en` ]                                                                         |
|  21 | [STRING]  |     2 | general.datasets                       | [ `https://hugging`...`-GPT4-train.txt`, `https://hugging`...`-GPT4-valid.txt` ] |
|  22 | UINT32    |     1 | llama.block_count                      | 8                                                                                |
|  23 | UINT32    |     1 | llama.context_length                   | 2048                                                                             |
|  24 | UINT32    |     1 | llama.embedding_length                 | 64                                                                               |
|  25 | UINT32    |     1 | llama.feed_forward_length              | 256                                                                              |
|  26 | UINT32    |     1 | llama.attention.head_count             | 16                                                                               |
|  27 | FLOAT32   |     1 | llama.attention.layer_norm_rms_epsilon | 1e-06                                                                            |
|  28 | UINT32    |     1 | general.file_type                      | 1                                                                                |
|  29 | UINT32    |     1 | llama.vocab_size                       | 32000                                                                            |
|  30 | UINT32    |     1 | llama.rope.dimension_count             | 4                                                                                |
|  31 | STRING    |     1 | tokenizer.ggml.model                   | `llama`                                                                          |
|  32 | STRING    |     1 | tokenizer.ggml.pre                     | `default`                                                                        |
|  33 | [STRING]  | 32000 | tokenizer.ggml.tokens                  | [ `<unk>`, `<s>`, `</s>`, `<0x00>`, `<0x01>`, ... ]                              |
|  34 | [FLOAT32] | 32000 | tokenizer.ggml.scores                  | [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ]                                       |
|  35 | [INT32]   | 32000 | tokenizer.ggml.token_type              | [ 2, 3, 3, 6, 6, 6, 6, ... ]                                                     |
|  36 | UINT32    |     1 | tokenizer.ggml.bos_token_id            | 1                                                                                |
|  37 | UINT32    |     1 | tokenizer.ggml.eos_token_id            | 2                                                                                |
|  38 | UINT32    |     1 | tokenizer.ggml.unknown_token_id        | 0                                                                                |
|  39 | UINT32    |     1 | tokenizer.ggml.padding_token_id        | 0                                                                                |
|  40 | UINT32    |     1 | general.quantization_version           | 2                                                                                |

## Tensors Overview ~5M Elements

Total number of elements in all tensors: 4621376 Elements

- [Base Tensor Group - ~4M Elements](#base)
- [Block 0 Tensor Group - ~66K Elements](#blk_0)
- [Block 1 Tensor Group - ~66K Elements](#blk_1)
- [Block 2 Tensor Group - ~66K Elements](#blk_2)
- [Block 3 Tensor Group - ~66K Elements](#blk_3)
- [Block 4 Tensor Group - ~66K Elements](#blk_4)
- [Block 5 Tensor Group - ~66K Elements](#blk_5)
- [Block 6 Tensor Group - ~66K Elements](#blk_6)
- [Block 7 Tensor Group - ~66K Elements](#blk_7)

### Tensor Data Offset

This table contains the offset and data segment relative to start of file

| T_ID | Tensor Layer Name        |  Data Offset (B) |    Data Size (B) |
|-----:|:-------------------------|-----------------:|-----------------:|
|    0 | output.weight            |          0xba8e0 |         0x3e8000 |
|    1 | token_embd.weight        |         0x4a28e0 |         0x3e8000 |
|    2 | blk.0.attn_norm.weight   |         0x88a8e0 |            0x100 |
|    3 | blk.0.ffn_down.weight    |         0x88a9e0 |           0x8000 |
|    4 | blk.0.ffn_gate.weight    |         0x8929e0 |           0x8000 |
|    5 | blk.0.ffn_up.weight      |         0x89a9e0 |           0x8000 |
|    6 | blk.0.ffn_norm.weight    |         0x8a29e0 |            0x100 |
|    7 | blk.0.attn_k.weight      |         0x8a2ae0 |           0x2000 |
|    8 | blk.0.attn_output.weight |         0x8a4ae0 |           0x2000 |
|    9 | blk.0.attn_q.weight      |         0x8a6ae0 |           0x2000 |
|   10 | blk.0.attn_v.weight      |         0x8a8ae0 |           0x2000 |
|   11 | blk.1.attn_norm.weight   |         0x8aaae0 |            0x100 |
|   12 | blk.1.ffn_down.weight    |         0x8aabe0 |           0x8000 |
|   13 | blk.1.ffn_gate.weight    |         0x8b2be0 |           0x8000 |
|   14 | blk.1.ffn_up.weight      |         0x8babe0 |           0x8000 |
|   15 | blk.1.ffn_norm.weight    |         0x8c2be0 |            0x100 |
|   16 | blk.1.attn_k.weight      |         0x8c2ce0 |           0x2000 |
|   17 | blk.1.attn_output.weight |         0x8c4ce0 |           0x2000 |
|   18 | blk.1.attn_q.weight      |         0x8c6ce0 |           0x2000 |
|   19 | blk.1.attn_v.weight      |         0x8c8ce0 |           0x2000 |
|   20 | blk.2.attn_norm.weight   |         0x8cace0 |            0x100 |
|   21 | blk.2.ffn_down.weight    |         0x8cade0 |           0x8000 |
|   22 | blk.2.ffn_gate.weight    |         0x8d2de0 |           0x8000 |
|   23 | blk.2.ffn_up.weight      |         0x8dade0 |           0x8000 |
|   24 | blk.2.ffn_norm.weight    |         0x8e2de0 |            0x100 |
|   25 | blk.2.attn_k.weight      |         0x8e2ee0 |           0x2000 |
|   26 | blk.2.attn_output.weight |         0x8e4ee0 |           0x2000 |
|   27 | blk.2.attn_q.weight      |         0x8e6ee0 |           0x2000 |
|   28 | blk.2.attn_v.weight      |         0x8e8ee0 |           0x2000 |
|   29 | blk.3.attn_norm.weight   |         0x8eaee0 |            0x100 |
|   30 | blk.3.ffn_down.weight    |         0x8eafe0 |           0x8000 |
|   31 | blk.3.ffn_gate.weight    |         0x8f2fe0 |           0x8000 |
|   32 | blk.3.ffn_up.weight      |         0x8fafe0 |           0x8000 |
|   33 | blk.3.ffn_norm.weight    |         0x902fe0 |            0x100 |
|   34 | blk.3.attn_k.weight      |         0x9030e0 |           0x2000 |
|   35 | blk.3.attn_output.weight |         0x9050e0 |           0x2000 |
|   36 | blk.3.attn_q.weight      |         0x9070e0 |           0x2000 |
|   37 | blk.3.attn_v.weight      |         0x9090e0 |           0x2000 |
|   38 | blk.4.attn_norm.weight   |         0x90b0e0 |            0x100 |
|   39 | blk.4.ffn_down.weight    |         0x90b1e0 |           0x8000 |
|   40 | blk.4.ffn_gate.weight    |         0x9131e0 |           0x8000 |
|   41 | blk.4.ffn_up.weight      |         0x91b1e0 |           0x8000 |
|   42 | blk.4.ffn_norm.weight    |         0x9231e0 |            0x100 |
|   43 | blk.4.attn_k.weight      |         0x9232e0 |           0x2000 |
|   44 | blk.4.attn_output.weight |         0x9252e0 |           0x2000 |
|   45 | blk.4.attn_q.weight      |         0x9272e0 |           0x2000 |
|   46 | blk.4.attn_v.weight      |         0x9292e0 |           0x2000 |
|   47 | blk.5.attn_norm.weight   |         0x92b2e0 |            0x100 |
|   48 | blk.5.ffn_down.weight    |         0x92b3e0 |           0x8000 |
|   49 | blk.5.ffn_gate.weight    |         0x9333e0 |           0x8000 |
|   50 | blk.5.ffn_up.weight      |         0x93b3e0 |           0x8000 |
|   51 | blk.5.ffn_norm.weight    |         0x9433e0 |            0x100 |
|   52 | blk.5.attn_k.weight      |         0x9434e0 |           0x2000 |
|   53 | blk.5.attn_output.weight |         0x9454e0 |           0x2000 |
|   54 | blk.5.attn_q.weight      |         0x9474e0 |           0x2000 |
|   55 | blk.5.attn_v.weight      |         0x9494e0 |           0x2000 |
|   56 | blk.6.attn_norm.weight   |         0x94b4e0 |            0x100 |
|   57 | blk.6.ffn_down.weight    |         0x94b5e0 |           0x8000 |
|   58 | blk.6.ffn_gate.weight    |         0x9535e0 |           0x8000 |
|   59 | blk.6.ffn_up.weight      |         0x95b5e0 |           0x8000 |
|   60 | blk.6.ffn_norm.weight    |         0x9635e0 |            0x100 |
|   61 | blk.6.attn_k.weight      |         0x9636e0 |           0x2000 |
|   62 | blk.6.attn_output.weight |         0x9656e0 |           0x2000 |
|   63 | blk.6.attn_q.weight      |         0x9676e0 |           0x2000 |
|   64 | blk.6.attn_v.weight      |         0x9696e0 |           0x2000 |
|   65 | blk.7.attn_norm.weight   |         0x96b6e0 |            0x100 |
|   66 | blk.7.ffn_down.weight    |         0x96b7e0 |           0x8000 |
|   67 | blk.7.ffn_gate.weight    |         0x9737e0 |           0x8000 |
|   68 | blk.7.ffn_up.weight      |         0x97b7e0 |           0x8000 |
|   69 | blk.7.ffn_norm.weight    |         0x9837e0 |            0x100 |
|   70 | blk.7.attn_k.weight      |         0x9838e0 |           0x2000 |
|   71 | blk.7.attn_output.weight |         0x9858e0 |           0x2000 |
|   72 | blk.7.attn_q.weight      |         0x9878e0 |           0x2000 |
|   73 | blk.7.attn_v.weight      |         0x9898e0 |           0x2000 |
|   74 | output_norm.weight       |         0x98b8e0 |            0x100 |

### <a name="base">Base Tensor Group : ~4M Elements</a>

| T_ID | Tensor Layer Name  | Human Friendly Tensor Layer Name | Elements      | Shape              | Type |
|-----:|:-------------------|:---------------------------------|:--------------|:-------------------|:-----|
|    0 | output.weight      | Output (W)                       | (~2M) 2048000 | 64 x 32000 x 1 x 1 | F16  |
|    1 | token_embd.weight  | Token Embedding (W)              | (~2M) 2048000 | 64 x 32000 x 1 x 1 | F16  |
|   74 | output_norm.weight | Output Normalization (W)         | ( 64)      64 | 64 x     1 x 1 x 1 | F32  |

- Total elements in base: ( ~4M) 4096064
- Percentage of total elements: 88.63%


### <a name="blk_0">Block 0 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|    2 | blk.0.attn_norm.weight   | Block 0 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|    3 | blk.0.ffn_down.weight    | Block 0 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|    4 | blk.0.ffn_gate.weight    | Block 0 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|    5 | blk.0.ffn_up.weight      | Block 0 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|    6 | blk.0.ffn_norm.weight    | Block 0 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|    7 | blk.0.attn_k.weight      | Block 0 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|    8 | blk.0.attn_output.weight | Block 0 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|    9 | blk.0.attn_q.weight      | Block 0 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   10 | blk.0.attn_v.weight      | Block 0 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.0: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_1">Block 1 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   11 | blk.1.attn_norm.weight   | Block 1 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   12 | blk.1.ffn_down.weight    | Block 1 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   13 | blk.1.ffn_gate.weight    | Block 1 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   14 | blk.1.ffn_up.weight      | Block 1 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   15 | blk.1.ffn_norm.weight    | Block 1 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   16 | blk.1.attn_k.weight      | Block 1 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   17 | blk.1.attn_output.weight | Block 1 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   18 | blk.1.attn_q.weight      | Block 1 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   19 | blk.1.attn_v.weight      | Block 1 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.1: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_2">Block 2 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   20 | blk.2.attn_norm.weight   | Block 2 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   21 | blk.2.ffn_down.weight    | Block 2 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   22 | blk.2.ffn_gate.weight    | Block 2 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   23 | blk.2.ffn_up.weight      | Block 2 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   24 | blk.2.ffn_norm.weight    | Block 2 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   25 | blk.2.attn_k.weight      | Block 2 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   26 | blk.2.attn_output.weight | Block 2 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   27 | blk.2.attn_q.weight      | Block 2 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   28 | blk.2.attn_v.weight      | Block 2 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.2: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_3">Block 3 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   29 | blk.3.attn_norm.weight   | Block 3 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   30 | blk.3.ffn_down.weight    | Block 3 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   31 | blk.3.ffn_gate.weight    | Block 3 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   32 | blk.3.ffn_up.weight      | Block 3 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   33 | blk.3.ffn_norm.weight    | Block 3 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   34 | blk.3.attn_k.weight      | Block 3 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   35 | blk.3.attn_output.weight | Block 3 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   36 | blk.3.attn_q.weight      | Block 3 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   37 | blk.3.attn_v.weight      | Block 3 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.3: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_4">Block 4 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   38 | blk.4.attn_norm.weight   | Block 4 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   39 | blk.4.ffn_down.weight    | Block 4 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   40 | blk.4.ffn_gate.weight    | Block 4 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   41 | blk.4.ffn_up.weight      | Block 4 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   42 | blk.4.ffn_norm.weight    | Block 4 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   43 | blk.4.attn_k.weight      | Block 4 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   44 | blk.4.attn_output.weight | Block 4 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   45 | blk.4.attn_q.weight      | Block 4 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   46 | blk.4.attn_v.weight      | Block 4 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.4: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_5">Block 5 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   47 | blk.5.attn_norm.weight   | Block 5 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   48 | blk.5.ffn_down.weight    | Block 5 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   49 | blk.5.ffn_gate.weight    | Block 5 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   50 | blk.5.ffn_up.weight      | Block 5 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   51 | blk.5.ffn_norm.weight    | Block 5 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   52 | blk.5.attn_k.weight      | Block 5 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   53 | blk.5.attn_output.weight | Block 5 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   54 | blk.5.attn_q.weight      | Block 5 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   55 | blk.5.attn_v.weight      | Block 5 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.5: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_6">Block 6 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   56 | blk.6.attn_norm.weight   | Block 6 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   57 | blk.6.ffn_down.weight    | Block 6 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   58 | blk.6.ffn_gate.weight    | Block 6 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   59 | blk.6.ffn_up.weight      | Block 6 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   60 | blk.6.ffn_norm.weight    | Block 6 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   61 | blk.6.attn_k.weight      | Block 6 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   62 | blk.6.attn_output.weight | Block 6 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   63 | blk.6.attn_q.weight      | Block 6 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   64 | blk.6.attn_v.weight      | Block 6 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.6: (~66K) 65664
- Percentage of total elements: 1.42%


### <a name="blk_7">Block 7 Tensor Group : ~66K Elements</a>

| T_ID | Tensor Layer Name        | Human Friendly Tensor Layer Name               | Elements     | Shape             | Type |
|-----:|:-------------------------|:-----------------------------------------------|:-------------|:------------------|:-----|
|   65 | blk.7.attn_norm.weight   | Block 7 Attention Normalization (W)            | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   66 | blk.7.ffn_down.weight    | Block 7 Feed-Forward Network "Down" (W)        | (~16K) 16384 | 256 x  64 x 1 x 1 | F16  |
|   67 | blk.7.ffn_gate.weight    | Block 7 Feed-Forward Network "Gate" (W)        | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   68 | blk.7.ffn_up.weight      | Block 7 Feed-Forward Network "Up" (W)          | (~16K) 16384 |  64 x 256 x 1 x 1 | F16  |
|   69 | blk.7.ffn_norm.weight    | Block 7 Feed-Forward Network Normalization (W) | (  64)    64 |  64 x   1 x 1 x 1 | F32  |
|   70 | blk.7.attn_k.weight      | Block 7 Attention Key (W)                      | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   71 | blk.7.attn_output.weight | Block 7 Attention Output (W)                   | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   72 | blk.7.attn_q.weight      | Block 7 Attention Query (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |
|   73 | blk.7.attn_v.weight      | Block 7 Attention Value (W)                    | ( ~4K)  4096 |  64 x  64 x 1 x 1 | F16  |

- Total elements in blk.7: (~66K) 65664
- Percentage of total elements: 1.42%