Update README.md
Browse files
README.md
CHANGED
|
@@ -14,8 +14,6 @@ tags:
|
|
| 14 |
|
| 15 |
GPT-sZero is a text generation model meant to answer the question: "How strong of a model can Google Colab create?"
|
| 16 |
|
| 17 |
-
## Model Details
|
| 18 |
-
|
| 19 |
### Model Description
|
| 20 |
|
| 21 |
GPT-sZero has 16 heads, 24 layers, and 1280 embeddings, similar to GPT2-Large, with a context window of 16,384 tokens. This maximized GPU VRAM while balancing size and functionality.
|
|
@@ -25,87 +23,11 @@ GPT-sZero has 16 heads, 24 layers, and 1280 embeddings, similar to GPT2-Large, w
|
|
| 25 |
- **Language(s) (NLP):** English
|
| 26 |
- **License:** MIT
|
| 27 |
|
| 28 |
-
##
|
| 29 |
-
|
| 30 |
-
### Training Data
|
| 31 |
-
|
| 32 |
-
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
|
| 33 |
-
|
| 34 |
-
[More Information Needed]
|
| 35 |
-
|
| 36 |
-
### Training Procedure
|
| 37 |
-
|
| 38 |
-
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
| 39 |
-
|
| 40 |
-
#### Preprocessing [optional]
|
| 41 |
-
|
| 42 |
-
[More Information Needed]
|
| 43 |
-
|
| 44 |
-
|
| 45 |
-
#### Training Hyperparameters
|
| 46 |
-
|
| 47 |
-
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
|
| 48 |
-
|
| 49 |
-
#### Speeds, Sizes, Times [optional]
|
| 50 |
-
|
| 51 |
-
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
|
| 52 |
-
|
| 53 |
-
[More Information Needed]
|
| 54 |
-
|
| 55 |
-
## Evaluation
|
| 56 |
-
|
| 57 |
-
<!-- This section describes the evaluation protocols and provides the results. -->
|
| 58 |
-
|
| 59 |
-
### Testing Data, Factors & Metrics
|
| 60 |
-
|
| 61 |
-
#### Testing Data
|
| 62 |
-
|
| 63 |
-
<!-- This should link to a Dataset Card if possible. -->
|
| 64 |
-
|
| 65 |
-
[More Information Needed]
|
| 66 |
-
|
| 67 |
-
#### Factors
|
| 68 |
-
|
| 69 |
-
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
|
| 70 |
-
|
| 71 |
-
[More Information Needed]
|
| 72 |
-
|
| 73 |
-
#### Metrics
|
| 74 |
-
|
| 75 |
-
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
|
| 76 |
-
|
| 77 |
-
[More Information Needed]
|
| 78 |
-
|
| 79 |
-
### Results
|
| 80 |
-
|
| 81 |
-
[More Information Needed]
|
| 82 |
-
|
| 83 |
-
#### Summary
|
| 84 |
-
|
| 85 |
-
## Environmental Impact
|
| 86 |
-
- **Hardware Type:** Google Colab G4 GPU
|
| 87 |
-
- **Hours used:** 58
|
| 88 |
-
- **Cloud Provider:** Google
|
| 89 |
-
|
| 90 |
-
## Technical Specifications [optional]
|
| 91 |
-
|
| 92 |
-
### Model Architecture and Objective
|
| 93 |
| | |
|
| 94 |
|---------------------------------|----:|
|
| 95 |
|Layers |24|
|
| 96 |
|Heads |16|
|
| 97 |
|Embeddings |1280|
|
| 98 |
|Context Window |16384 tokens|
|
| 99 |
-
|Tokenizer |cl100k_base|
|
| 100 |
-
|
| 101 |
-
### Compute Infrastructure
|
| 102 |
-
|
| 103 |
-
[More Information Needed]
|
| 104 |
-
|
| 105 |
-
#### Hardware
|
| 106 |
-
|
| 107 |
-
[More Information Needed]
|
| 108 |
-
|
| 109 |
-
#### Software
|
| 110 |
-
|
| 111 |
-
[More Information Needed]
|
|
|
|
| 14 |
|
| 15 |
GPT-sZero is a text generation model meant to answer the question: "How strong of a model can Google Colab create?"
|
| 16 |
|
|
|
|
|
|
|
| 17 |
### Model Description
|
| 18 |
|
| 19 |
GPT-sZero has 16 heads, 24 layers, and 1280 embeddings, similar to GPT2-Large, with a context window of 16,384 tokens. This maximized GPU VRAM while balancing size and functionality.
|
|
|
|
| 23 |
- **Language(s) (NLP):** English
|
| 24 |
- **License:** MIT
|
| 25 |
|
| 26 |
+
### Model Architecture
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
| | |
|
| 28 |
|---------------------------------|----:|
|
| 29 |
|Layers |24|
|
| 30 |
|Heads |16|
|
| 31 |
|Embeddings |1280|
|
| 32 |
|Context Window |16384 tokens|
|
| 33 |
+
|Tokenizer |cl100k_base|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|