logo
Browse files- README.md +61 -0
- misc/logo.png +3 -0
README.md
CHANGED
@@ -1,3 +1,64 @@
|
|
1 |
---
|
2 |
license: apache-2.0
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
license: apache-2.0
|
3 |
+
pipeline_tag: text-generation
|
4 |
+
library_name: transformers
|
5 |
+
language: [
|
6 |
+
'am', 'ar', 'as', 'az', 'be', 'bg', 'bn', 'bn_rom', 'br',
|
7 |
+
'bs', 'ca', 'cs', 'cy', 'da', 'de', 'el', 'en', 'eo', 'es',
|
8 |
+
'et', 'eu', 'fa', 'ff', 'fi', 'fr', 'fy', 'ga', 'gd', 'gl',
|
9 |
+
'gn', 'gu', 'ha', 'he', 'hi', 'hi_rom', 'hr', 'ht', 'hu',
|
10 |
+
'hy', 'id', 'ig', 'is', 'it', 'ja', 'jv', 'ka', 'kk', 'km',
|
11 |
+
'kn', 'ko', 'ku', 'ky', 'la', 'lg', 'li', 'ln', 'lo', 'lt',
|
12 |
+
'lv', 'mg', 'mk', 'ml', 'mn', 'mr', 'ms', 'my', 'my_zaw',
|
13 |
+
'ne', 'nl', 'no', 'ns', 'om', 'or', 'pa', 'pl', 'ps', 'pt',
|
14 |
+
'qu', 'rm', 'ro', 'ru', 'sa', 'si', 'sc', 'sd', 'sk', 'sl',
|
15 |
+
'so', 'sq', 'sr', 'ss', 'su', 'sv', 'sw', 'ta', 'ta_rom',
|
16 |
+
'te', 'te_rom', 'th', 'tl', 'tn', 'tr', 'ug', 'uk', 'ur',
|
17 |
+
'ur_rom', 'uz', 'vi', 'wo', 'xh', 'yi', 'yo',
|
18 |
+
'zh-Hans', 'zh-Hant', 'zu',
|
19 |
+
]
|
20 |
+
datasets: [
|
21 |
+
'yahma/alpaca-cleaned',
|
22 |
+
'gbharti/wealth-alpaca_lora',
|
23 |
+
'saillab/taco-datasets',
|
24 |
+
'xu-song/cc100-samples',
|
25 |
+
'ontocord/fineweb-permissive-multilingual-2m',
|
26 |
+
'MuskumPillerum/General-Knowledge',
|
27 |
+
'yirenc/general_knowledge_boolean',
|
28 |
+
'nampdn-ai/tiny-textbooks',
|
29 |
+
'nampdn-ai/tiny-codes',
|
30 |
+
'bigcode/the-stack-smol-xs',
|
31 |
+
'm-a-p/CodeFeedback-Filtered-Instruction',
|
32 |
+
'jtatman/python-code-dataset-500k',
|
33 |
+
'iamtarun/python_code_instructions_18k_alpaca',
|
34 |
+
'HuggingFaceH4/CodeAlpaca_20K',
|
35 |
+
'gair-prox/open-web-math-pro',
|
36 |
+
'rvv-karma/Math-QA',
|
37 |
+
'ajibawa-2023/Maths-College',
|
38 |
+
'microsoft/orca-math-word-problems-200k',
|
39 |
+
'fblgit/simple-math',
|
40 |
+
'SkunkworksAI/reasoning-0.01',
|
41 |
+
'badrex/llm-emoji-dataset',
|
42 |
+
]
|
43 |
+
tags:
|
44 |
+
- litgpt
|
45 |
+
- litdata
|
46 |
---
|
47 |
+
|
48 |
+
# tangled-llama-58m-32k-base-v0.1
|
49 |
+
|
50 |
+
![logo](./misc/logo.png)
|
51 |
+
|
52 |
+
A pretrained language model based on the Llama model with about **58M** parameters. This model has been trained on **11.4B** (`11,422,750,857`) tokens from more than **0.8M** (`796,399`) dataset rows.
|
53 |
+
|
54 |
+
This model **isn't** designed for immediate use but rather for Continued Pretraining and Finetuning on a downstream task. While it can handle a context length of up to **128K** (`131,072`) tokens, it was pretrained with sequences of **2K** (`2048`) tokens.
|
55 |
+
|
56 |
+
The objective is to streamline the cognitive or reasoning core, eliminating any redundant knowledge from the model.
|
57 |
+
|
58 |
+
[loss, val_loss](https://api.wandb.ai/links/mtasic85/0i3wqsmb)
|
59 |
+
|
60 |
+
[val_ppl](https://api.wandb.ai/links/mtasic85/vz1jgu3v)
|
61 |
+
|
62 |
+
[epoch](https://api.wandb.ai/links/mtasic85/qltqthjr)
|
63 |
+
|
64 |
+
[learning_rate](https://api.wandb.ai/links/mtasic85/eqpiton4)
|
misc/logo.png
ADDED
Git LFS Details
|