Add revisions and fixes
Browse files- Add details and links to overview
- Add vocab information
- Fix usage instructions
- Add MistralAI to citations for sentencepiece mistral model vocab
- Fix Citations and other references
README.md
CHANGED
@@ -15,29 +15,28 @@ tags:
|
|
15 |
|
16 |
## Valerie v0.1 Model Card
|
17 |
|
18 |
-
|
19 |
|
20 |
-
Valerie v0.1 is a custom language model created using `llama.cpp` (commit: 532c173) with a context length of 256 tokens, embedding length of 256, 8 heads, and
|
21 |
-
16 layers. This model was pretrained on a dataset consisting of female V's dialog from Cyberpunk 2077, extracted using the [Voice Over Subtitle
|
22 |
-
Map](https://www.nexusmods.com/cyberpunk2077/mods/2045) mod.
|
23 |
|
24 |
-
The `ggml-valerie-v0.1-256x32-f32-LATEST.gguf` release represents a single pass through all 51443 samples, completing one iteration over the entire dataset,
|
25 |
-
and took approximately 3 hours for training.
|
26 |
|
27 |
-
|
28 |
|
29 |
-
| Model name | Adam iteration | Model filename |
|
30 |
-
| ----------------------- | -------------- | ---------------------------------------- |
|
31 |
-
| Valerie v0.1 Checkpoint | 950 | chk-valerie-v0.1-256x32-950.gguf |
|
32 |
-
| Valerie v0.1 Model | 1700 | ggml-valerie-v0.1-256x32-f32-LATEST.gguf |
|
33 |
|
34 |
### Files and versions
|
35 |
|
|
|
36 |
- ggml-valerie-v0.1-256x32-f32-950.gguf: The pretrained model checkpoint version 950.
|
37 |
- ggml-valerie-v0.1-256x32-f32-LATEST.gguf: The latest pretrained model checkpoint.
|
38 |
|
39 |
-
|
40 |
|
|
|
41 |
- Context length: 256 tokens
|
42 |
- Embedding length: 256
|
43 |
- Heads: 8
|
@@ -46,25 +45,34 @@ and took approximately 3 hours for training.
|
|
46 |
- Seed: 1
|
47 |
- Saved checkpoint every 50 iterations
|
48 |
|
49 |
-
|
50 |
|
51 |
To use Valerie v0.1, follow these steps:
|
52 |
|
53 |
-
1.
|
54 |
|
55 |
-
|
56 |
-
|
57 |
-
|
|
|
|
|
|
|
|
|
58 |
|
59 |
-
|
|
|
|
|
|
|
|
|
60 |
|
61 |
-
2. Download
|
62 |
|
63 |
-
|
64 |
-
|
65 |
-
|
|
|
66 |
|
67 |
-
|
68 |
|
69 |
3. Perform inference with the latest model checkpoint using the provided command:
|
70 |
|
@@ -72,13 +80,14 @@ To use Valerie v0.1, follow these steps:
|
|
72 |
./main -m models/valerie/v0.1/ggml-valerie-v0.1-256x32-f32-LATEST.gguf --color -e -s 1 -c 4096
|
73 |
```
|
74 |
|
75 |
-
|
76 |
|
77 |
When using Valerie v0.1 in your research, please remember to cite the following:
|
78 |
|
79 |
-
- Aberrio. (
|
80 |
-
-
|
81 |
-
-
|
|
|
82 |
|
83 |
### Contributors
|
84 |
|
@@ -86,10 +95,8 @@ Austin Berrio (teleprint-me) - Created and trained Valerie v0.1 using `llama.cpp
|
|
86 |
|
87 |
### Community
|
88 |
|
89 |
-
Join the community of fellow language model enthusiasts and researchers by sharing your knowledge, asking questions, and collaborating on projects related to
|
90 |
-
creating custom models using `llama.cpp`.
|
91 |
|
92 |
### License
|
93 |
|
94 |
-
Valerie v0.1 is released under the CC-BY-NC-SA-3.0 license. You are free to use, modify, and redistribute this model for non-commercial purposes, but you must
|
95 |
-
provide attribution to the original authors and release any derived works under the same license.
|
|
|
15 |
|
16 |
## Valerie v0.1 Model Card
|
17 |
|
18 |
+
## Overview
|
19 |
|
20 |
+
Valerie v0.1 is a custom language model created using `llama.cpp` (commit: 532c173) with a context length of 256 tokens, embedding length of 256, 8 heads, and 16 layers. This model was pretrained on a dataset consisting of [female V's](https://cyberpunk.fandom.com/wiki/V_(character)) dialog from [Cyberpunk 2077](https://cyberpunk.fandom.com/wiki/Cyberpunk_Wiki), extracted using the [Voice Over Subtitle Map](https://www.nexusmods.com/cyberpunk2077/mods/2045) mod.
|
|
|
|
|
21 |
|
22 |
+
The `ggml-valerie-v0.1-256x32-f32-LATEST.gguf` release represents a single pass through all 51443 samples, completing one iteration over the entire dataset, and took approximately 3 hours for training.
|
|
|
23 |
|
24 |
+
## Model Information
|
25 |
|
26 |
+
| Model name | Adam iteration | Model filename | Vocabulary size |
|
27 |
+
| ----------------------- | -------------- | ---------------------------------------- | --------------- |
|
28 |
+
| Valerie v0.1 Checkpoint | 950 | chk-valerie-v0.1-256x32-950.gguf | 32,000 |
|
29 |
+
| Valerie v0.1 Model | 1700 | ggml-valerie-v0.1-256x32-f32-LATEST.gguf | 32,000 |
|
30 |
|
31 |
### Files and versions
|
32 |
|
33 |
+
- ggml-vocab-mistral.gguf: Extracted Mistral 7B model vocabulary
|
34 |
- ggml-valerie-v0.1-256x32-f32-950.gguf: The pretrained model checkpoint version 950.
|
35 |
- ggml-valerie-v0.1-256x32-f32-LATEST.gguf: The latest pretrained model checkpoint.
|
36 |
|
37 |
+
## Settings
|
38 |
|
39 |
+
- Vocabulary size: 32,000
|
40 |
- Context length: 256 tokens
|
41 |
- Embedding length: 256
|
42 |
- Heads: 8
|
|
|
45 |
- Seed: 1
|
46 |
- Saved checkpoint every 50 iterations
|
47 |
|
48 |
+
## Usage
|
49 |
|
50 |
To use Valerie v0.1, follow these steps:
|
51 |
|
52 |
+
1. Clone the `llama.cpp` library
|
53 |
|
54 |
+
```sh
|
55 |
+
git clone https://github.com/ggerganov/llama.cpp
|
56 |
+
```
|
57 |
+
|
58 |
+
Reference the `llama.cpp` [README.md](https://github.com/ggerganov/llama.cpp/blob/master/README.md) for more information about building. You can build using raw CPU or even OpenBLAS. CUDA, ROCm, Vulkan, and other backends are also available.
|
59 |
+
|
60 |
+
Arch Linux Example:
|
61 |
|
62 |
+
```sh
|
63 |
+
# CPU build using BLAS backend on Arch Linux
|
64 |
+
sudo pacman -S openblas openblas64
|
65 |
+
make LLAMA_OPENBLAS=1
|
66 |
+
```
|
67 |
|
68 |
+
2. Download the latest model.
|
69 |
|
70 |
+
```sh
|
71 |
+
wget https://huggingface.co/teleprint-me/cyberpunk-valerie-v0.1/resolve/main/ggml-valerie-v0.1-256x32-f32-LATEST.gguf?download=true -O
|
72 |
+
ggml-valerie-v0.1-256x32-f32-LATEST.gguf
|
73 |
+
```
|
74 |
|
75 |
+
This will download the latest available base model.
|
76 |
|
77 |
3. Perform inference with the latest model checkpoint using the provided command:
|
78 |
|
|
|
80 |
./main -m models/valerie/v0.1/ggml-valerie-v0.1-256x32-f32-LATEST.gguf --color -e -s 1 -c 4096
|
81 |
```
|
82 |
|
83 |
+
## Citations
|
84 |
|
85 |
When using Valerie v0.1 in your research, please remember to cite the following:
|
86 |
|
87 |
+
- Aberrio. (2024). Valerie v0.1: A custom language model for female V's dialog from Cyberpunk 2077. <https://huggingface.co/teleprint-me/cyberpunk-valerie-v0.1>
|
88 |
+
- GGML team. (2023). `llama.cpp` version `532c173`. Georgi Gerganov Machine Learning Library. <https://github.com/ggerganov/llama.cpp>
|
89 |
+
- MistralAI (2023). Extracted sentencepiece model vocabulary: <https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2>
|
90 |
+
- julieisdead (2021). Voice Over Subtitle Map: Two files that contain the IDs for, Voice Over files the other Subtitles. <https://www.nexusmods.com/cyberpunk2077/mods/2045>
|
91 |
|
92 |
### Contributors
|
93 |
|
|
|
95 |
|
96 |
### Community
|
97 |
|
98 |
+
Join the community of fellow language model enthusiasts and researchers by sharing your knowledge, asking questions, and collaborating on projects related to creating custom models using `llama.cpp`.
|
|
|
99 |
|
100 |
### License
|
101 |
|
102 |
+
Valerie v0.1 is released under the CC-BY-NC-SA-3.0 license. You are free to use, modify, and redistribute this model for non-commercial purposes, but you must provide attribution to the original authors and release any derived works under the same license.
|
|