Commit
·
56f2ae6
1
Parent(s):
53aeb62
Testing model card updates
Browse files
README.md
CHANGED
|
@@ -1,7 +1,44 @@
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
---
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
license: mit
|
| 3 |
---
|
| 4 |
+
- en
|
| 5 |
+
pipeline_tag: text-generation
|
| 6 |
+
tags:
|
| 7 |
+
- phi
|
| 8 |
+
- nlp
|
| 9 |
+
- math
|
| 10 |
+
- code
|
| 11 |
+
- chat
|
| 12 |
+
- conversational
|
| 13 |
+
inference:
|
| 14 |
+
parameters:
|
| 15 |
+
temperature: 0
|
| 16 |
+
widget:
|
| 17 |
+
- messages:
|
| 18 |
+
- role: user
|
| 19 |
+
content: How should I explain the Internet?
|
| 20 |
+
library_name: transformers
|
| 21 |
---
|
| 22 |
+
|
| 23 |
+
# Phi-4 Model Card
|
| 24 |
+
|
| 25 |
+
[Phi-4 Technical Report](https://arxiv.org/pdf/2412.08905)
|
| 26 |
+
|
| 27 |
+
## Model Summary
|
| 28 |
+
|
| 29 |
+
| | |
|
| 30 |
+
|-------------------------|-------------------------------------------------------------------------------|
|
| 31 |
+
| **Developers** | Microsoft Research |
|
| 32 |
+
| **Description** | `phi-4` is a state-of-the-art open model built upon a blend of synthetic datasets, data from filtered public domain websites, and acquired academic books and Q&A datasets. The goal of this approach was to ensure that small capable models were trained with data focused on high quality and advanced reasoning.<br><br>`phi-4` underwent a rigorous enhancement and alignment process, incorporating both supervised fine-tuning and direct preference optimization to ensure precise instruction adherence and robust safety measures |
|
| 33 |
+
| **Architecture** | 14B parameters, dense decoder-only Transformer model |
|
| 34 |
+
| **Inputs** | Text, best suited for prompts in the chat format |
|
| 35 |
+
| **Context length** | 16K tokens |
|
| 36 |
+
| **GPUs** | 1920 H100-80G |
|
| 37 |
+
| **Training time** | 21 days |
|
| 38 |
+
| **Training data** | 9.8T tokens |
|
| 39 |
+
| **Outputs** | Generated text in response to input |
|
| 40 |
+
| **Dates** | October 2024 – November 2024 |
|
| 41 |
+
| **Status** | Static model trained on an offline dataset with cutoff dates of June 2024 and earlier for publicly available data |
|
| 42 |
+
| **Release date** | December 12, 2024 |
|
| 43 |
+
| **License** | MIT |
|
| 44 |
+
|