Add .gguf to lfs to track
Browse files- .gitattributes +1 -0
- README.md +68 -0
.gitattributes
CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
*.gguf filter=lfs diff=lfs merge=lfs -text
|
README.md
CHANGED
@@ -1,3 +1,71 @@
|
|
1 |
---
|
2 |
license: mit
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: mit
|
3 |
+
language:
|
4 |
+
- en
|
5 |
---
|
6 |
+
|
7 |
+
# On-chain llama.cpp - Internet Computer
|
8 |
+
|
9 |
+
|
10 |
+
- These models were created for Internet Computer canisters deployed with [onicai/llama_cpp_canister](https://github.com/onicai/llama_cpp_canister)
|
11 |
+
- They are used in the LLM canisters of [ICGPT](https://icgpt.icpp.world/)
|
12 |
+
- The models were created with the training procedure outlined in [karpathy/llama2.c](https://github.com/karpathy/llama2.c)
|
13 |
+
- You can run them local too, as described in [karpathy/llama2.c](https://github.com/karpathy/llama2.c)
|
14 |
+
|
15 |
+
## Setup git
|
16 |
+
|
17 |
+
See: [Getting Started: set-up](https://huggingface.co/docs/hub/repositories-getting-started#set-up)
|
18 |
+
|
19 |
+
```bash
|
20 |
+
pip install huggingface-hub
|
21 |
+
|
22 |
+
git clone <this-repo>
|
23 |
+
cd <this-repo>
|
24 |
+
|
25 |
+
git lfs install
|
26 |
+
git lfs track "*.gguf"
|
27 |
+
huggingface-cli lfs-enable-largefiles .
|
28 |
+
|
29 |
+
# add & push as usual with git
|
30 |
+
git add <file-name>
|
31 |
+
git commit -m "Adding <file-name>"
|
32 |
+
git push -u origin main
|
33 |
+
```
|
34 |
+
|
35 |
+
## TinyStories models
|
36 |
+
|
37 |
+
| model | notes |
|
38 |
+
|-------|-------|
|
39 |
+
| stories260Ktok512.guff | Use this for development & debugging |
|
40 |
+
| stories15Mtok4096.guff | Fits in canister & works well ! |
|
41 |
+
| stories42Mtok4096.guff | As of April 28, hits instruction limit of canister |
|
42 |
+
| stories42Mtok32000.guff (*) | As of April 28, hits instruction limit of canister |
|
43 |
+
| stories110Mtok32000.guff (*) | As of April 28, hits instruction limit of canister |
|
44 |
+
|
45 |
+
|
46 |
+
We used [convert-llama2c-to-ggml](https://github.com/ggerganov/llama.cpp/tree/32c8486e1f0297393cb22ac0a0d26a6b17ad4d54/examples/convert-llama2c-to-ggml) to convert the llama2.c model+tokenizer to llama.cpp gguf format.
|
47 |
+
- Good read: [lama : add support for llama2.c models](https://github.com/ggerganov/llama.cpp/issues/2379)
|
48 |
+
|
49 |
+
For example:
|
50 |
+
```bash
|
51 |
+
# From llama.cpp root folder
|
52 |
+
|
53 |
+
# Build everything
|
54 |
+
make -j
|
55 |
+
|
56 |
+
# Convert a llama2c model+tokenizer to gguf
|
57 |
+
convert-llama2c-to-ggml --llama2c-model stories260Ktok512.bin --copy-vocab-from-model tok512.bin --llama2c-output-model stories260Ktok512.gguf
|
58 |
+
convert-llama2c-to-ggml --llama2c-model stories15Mtok4096.bin --copy-vocab-from-model tok4096.bin --llama2c-output-model stories15Mtok4096.gguf
|
59 |
+
convert-llama2c-to-ggml --llama2c-model stories42Mtok4096.bin --copy-vocab-from-model tok4096.bin --llama2c-output-model stories42Mtok4096.gguf
|
60 |
+
convert-llama2c-to-ggml --llama2c-model stories110Mtok32000.bin --copy-vocab-from-model models/ggml-vocab-llama.gguf --llama2c-output-model stories110Mtok32000.gguf
|
61 |
+
convert-llama2c-to-ggml --llama2c-model stories42Mtok32000.bin --copy-vocab-from-model models/ggml-vocab-llama.gguf --llama2c-output-model stories42Mtok32000.gguf
|
62 |
+
|
63 |
+
# Run it local, like this
|
64 |
+
main -m stories15Mtok4096.gguf -p "Joe loves writing stories" -n 600 -c 128
|
65 |
+
|
66 |
+
# Quantization
|
67 |
+
#
|
68 |
+
```
|
69 |
+
|
70 |
+
(*) Files with asterix behind them were not trained by us, but simply copied from [karpathy/tinyllamas](https://huggingface.co/karpathy/tinyllamas/tree/main) and renamed. We are providing them here under a different name for clarity and ease-of-access.
|
71 |
+
|