Hugging Face's logo

mofosyne
/
TinyLLama-v0-5M-F16-llamafile

Text Generation
GGUF
llamafile
English
model-conversion
Inference Endpoints
Model card Files Community
mofosyne commited on
Commit
19c0773
•
1 Parent(s): 809d127

update dumpfile with correct kv array rendering

Browse files
Files changed (3) hide show
  1. Tinyllama-4.6M-v0.0-F16.md → Tinyllama-4.6M-v0.0-F16.dump.md +19 -19
  2. llama.cpp +1 -1
  3. llamafile-creation.sh +4 -5
Tinyllama-4.6M-v0.0-F16.md → Tinyllama-4.6M-v0.0-F16.dump.md RENAMED
@@ -11,20 +11,20 @@ There are 36 key-value pairs in this file
11
  | 1 | UINT32 | 1 | GGUF.version | 3 |
12
  | 2 | UINT64 | 1 | GGUF.tensor_count | 75 |
13
  | 3 | UINT64 | 1 | GGUF.kv_count | 33 |
14
- | 4 | STRING | 1 | general.architecture | 'llama' |
15
- | 5 | STRING | 1 | general.type | 'model' |
16
- | 6 | STRING | 1 | general.name | 'TinyLLama' |
17
- | 7 | STRING | 1 | general.author | 'Maykeye' |
18
- | 8 | STRING | 1 | general.version | 'v0.0' |
19
- | 9 | STRING | 1 | general.description | 'This gguf is ported from a first version of Maykeye attempt ' |
20
- | 10 | STRING | 1 | general.quantized_by | 'Mofosyne' |
21
- | 11 | STRING | 1 | general.size_label | '4.6M' |
22
- | 12 | STRING | 1 | general.license | 'apache-2.0' |
23
- | 13 | STRING | 1 | general.url | 'https://huggingface.co/mofosyne/TinyLLama-v0-llamafile' |
24
- | 14 | STRING | 1 | general.source.url | 'https://huggingface.co/Maykeye/TinyLLama-v0' |
25
- | 15 | [STRING] | 5 | general.tags | [ 'tiny ', '\n\x00\x00\x00\x00', 'tiny', '\x04\x00\x00\x00\x00', 'llama', ... ] |
26
- | 16 | [STRING] | 1 | general.languages | [ 'en' ] |
27
- | 17 | [STRING] | 2 | general.datasets | [ 'https', ']\x00\x00\x00\x00', ... ] |
28
  | 18 | UINT32 | 1 | llama.block_count | 8 |
29
  | 19 | UINT32 | 1 | llama.context_length | 2048 |
30
  | 20 | UINT32 | 1 | llama.embedding_length | 64 |
@@ -34,11 +34,11 @@ There are 36 key-value pairs in this file
34
  | 24 | UINT32 | 1 | general.file_type | 1 |
35
  | 25 | UINT32 | 1 | llama.vocab_size | 32000 |
36
  | 26 | UINT32 | 1 | llama.rope.dimension_count | 4 |
37
- | 27 | STRING | 1 | tokenizer.ggml.model | 'llama' |
38
- | 28 | STRING | 1 | tokenizer.ggml.pre | 'default' |
39
- | 29 | [STRING] | 32000 | tokenizer.ggml.tokens | [ 'А', '\x02\x00\x00\x00\x00', 'š', '\x02\x00\x00\x00\x00', 'α', ... ] |
40
- | 30 | [FLOAT32] | 32000 | tokenizer.ggml.scores | [ -31740.0, -31739.0, -31738.0, -31737.0, -31736.0, -31735.0, -31734.0, ... ] |
41
- | 31 | [INT32] | 32000 | tokenizer.ggml.token_type | [ 1, 1, 1, 1, 1, 1, 1, ... ] |
42
  | 32 | UINT32 | 1 | tokenizer.ggml.bos_token_id | 1 |
43
  | 33 | UINT32 | 1 | tokenizer.ggml.eos_token_id | 2 |
44
  | 34 | UINT32 | 1 | tokenizer.ggml.unknown_token_id | 0 |
 
11
  | 1 | UINT32 | 1 | GGUF.version | 3 |
12
  | 2 | UINT64 | 1 | GGUF.tensor_count | 75 |
13
  | 3 | UINT64 | 1 | GGUF.kv_count | 33 |
14
+ | 4 | STRING | 1 | general.architecture | `llama` |
15
+ | 5 | STRING | 1 | general.type | `model` |
16
+ | 6 | STRING | 1 | general.name | `TinyLLama` |
17
+ | 7 | STRING | 1 | general.author | `Maykeye` |
18
+ | 8 | STRING | 1 | general.version | `v0.0` |
19
+ | 9 | STRING | 1 | general.description | `This gguf is ported from a fir`...`M but using Llama architecture` |
20
+ | 10 | STRING | 1 | general.quantized_by | `Mofosyne` |
21
+ | 11 | STRING | 1 | general.size_label | `4.6M` |
22
+ | 12 | STRING | 1 | general.license | `apache-2.0` |
23
+ | 13 | STRING | 1 | general.url | `https://huggingface.co/mofosyne/TinyLLama-v0-llamafile` |
24
+ | 14 | STRING | 1 | general.source.url | `https://huggingface.co/Maykeye/TinyLLama-v0` |
25
+ | 15 | [STRING] | 5 | general.tags | [ `text generation`, `transformer`, `llama`, `tiny`, `tiny model` ] |
26
+ | 16 | [STRING] | 1 | general.languages | [ `en` ] |
27
+ | 17 | [STRING] | 2 | general.datasets | [ `https://hugging`...`-GPT4-train.txt`, `https://hugging`...`-GPT4-valid.txt` ] |
28
  | 18 | UINT32 | 1 | llama.block_count | 8 |
29
  | 19 | UINT32 | 1 | llama.context_length | 2048 |
30
  | 20 | UINT32 | 1 | llama.embedding_length | 64 |
 
34
  | 24 | UINT32 | 1 | general.file_type | 1 |
35
  | 25 | UINT32 | 1 | llama.vocab_size | 32000 |
36
  | 26 | UINT32 | 1 | llama.rope.dimension_count | 4 |
37
+ | 27 | STRING | 1 | tokenizer.ggml.model | `llama` |
38
+ | 28 | STRING | 1 | tokenizer.ggml.pre | `default` |
39
+ | 29 | [STRING] | 32000 | tokenizer.ggml.tokens | [ `<unk>`, `<s>`, `</s>`, `<0x00>`, `<0x01>`, ... ] |
40
+ | 30 | [FLOAT32] | 32000 | tokenizer.ggml.scores | [ 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, ... ] |
41
+ | 31 | [INT32] | 32000 | tokenizer.ggml.token_type | [ 2, 3, 3, 6, 6, 6, 6, ... ] |
42
  | 32 | UINT32 | 1 | tokenizer.ggml.bos_token_id | 1 |
43
  | 33 | UINT32 | 1 | tokenizer.ggml.eos_token_id | 2 |
44
  | 34 | UINT32 | 1 | tokenizer.ggml.unknown_token_id | 0 |
llama.cpp CHANGED
@@ -1 +1 @@
1
- Subproject commit f299aa98ecc19cbc574e9d698e03999e89de3d3d
 
1
+ Subproject commit c3776cacabce2ee35f172fb72be7a519752125fa
llamafile-creation.sh CHANGED
@@ -6,7 +6,7 @@ METADATA_FILE="maykeye_tinyllama-metadata.json"
6
  ###############################################################################
7
  # Pull both model folder, llamafile (for the engine) and llama.cpp (for the conversion script)
8
  echo == Prep Enviroment ==
9
- git submodule update --init
10
 
11
  ###############################################################################
12
  echo == Build and prep the llamafile engine execuable ==
@@ -37,6 +37,9 @@ echo == Convert from safetensor to gguf ==
37
  ./llama.cpp/convert_hf_to_gguf.py ${MODEL_DIR} --metadata ${METADATA_FILE} --outtype f16 --verbose
38
  mv ${MODEL_DIR}/${OUTFILE}.gguf ${OUTFILE}.gguf
39
 
 
 
 
40
  ###############################################################################
41
  echo == Generating Llamafile ==
42
  cp ./llamafile/o/llama.cpp/main/main ${OUTFILE}.llamafile
@@ -53,7 +56,3 @@ EOF
53
  ###############################################################################
54
  echo == Test Output ./${OUTFILE}.llamafile ==
55
  ./${OUTFILE}.llamafile --cli -p "hello world the gruff man said"
56
-
57
- ###############################################################################
58
- echo == Useful GGUF Technical Dump ==
59
- ./llama.cpp/gguf-py/scripts/gguf_dump.py --markdown ${OUTFILE}.gguf > ${OUTFILE}.md
 
6
  ###############################################################################
7
  # Pull both model folder, llamafile (for the engine) and llama.cpp (for the conversion script)
8
  echo == Prep Enviroment ==
9
+ #git submodule update --init
10
 
11
  ###############################################################################
12
  echo == Build and prep the llamafile engine execuable ==
 
37
  ./llama.cpp/convert_hf_to_gguf.py ${MODEL_DIR} --metadata ${METADATA_FILE} --outtype f16 --verbose
38
  mv ${MODEL_DIR}/${OUTFILE}.gguf ${OUTFILE}.gguf
39
 
40
+ # Generate Diagnostics Dumpfile
41
+ ./llama.cpp/gguf-py/scripts/gguf_dump.py --markdown ${OUTFILE}.gguf > ${OUTFILE}.dump.md
42
+
43
  ###############################################################################
44
  echo == Generating Llamafile ==
45
  cp ./llamafile/o/llama.cpp/main/main ${OUTFILE}.llamafile
 
56
  ###############################################################################
57
  echo == Test Output ./${OUTFILE}.llamafile ==
58
  ./${OUTFILE}.llamafile --cli -p "hello world the gruff man said"