Update README.md
Browse files
README.md
CHANGED
|
@@ -5,7 +5,7 @@ language:
|
|
| 5 |
- en
|
| 6 |
base_model:
|
| 7 |
- Qwen/Qwen3-Omni-30B-A3B-Instruct
|
| 8 |
-
|
| 9 |
|
| 10 |
# LAT-Audio
|
| 11 |
|
|
@@ -21,7 +21,7 @@ During reasoning, LAT-Audio iteratively incorporates audio evidence through a **
|
|
| 21 |
- temporal hallucination (invalid timestamps)
|
| 22 |
- timestamp drift (progressive misalignment over time)
|
| 23 |
|
| 24 |
-
|
| 25 |
|
| 26 |
## Model Description
|
| 27 |
|
|
@@ -38,7 +38,7 @@ LAT-Audio formulates long-form audio understanding as a structured reasoning pro
|
|
| 38 |
|
| 39 |
This design enables robust temporal reasoning under long-context settings, where conventional direct modeling approaches often fail.
|
| 40 |
|
| 41 |
-
|
| 42 |
|
| 43 |
## Model Variants
|
| 44 |
|
|
@@ -48,7 +48,7 @@ We provide two model variants:
|
|
| 48 |
| **LAT-Audio** | ✅ Yes | LAT-Chronicle | Tool-augmented multi-step reasoning model with global-to-local temporal inference |
|
| 49 |
| **LAT-Audio-Base** | ❌ No | LAT-Chronicle + in-house | Direct modeling baseline fine-tuned from Qwen3-Omni with more in-house data, offering faster and simpler inference |
|
| 50 |
|
| 51 |
-
|
| 52 |
|
| 53 |
## Quick Start
|
| 54 |
|
|
@@ -58,10 +58,11 @@ pip install -U "huggingface_hub[cli]"
|
|
| 58 |
huggingface-cli download mcshao/LAT-Audio --local-dir ./LAT-Audio
|
| 59 |
huggingface-cli download mcshao/LAT-Audio-Base --local-dir ./LAT-Audio-Base
|
| 60 |
|
|
|
|
| 61 |
For detailed inference methods and examples, please refer to the official repository:
|
| 62 |
👉 https://github.com/alanshaoTT/LAT-Audio-Repo
|
| 63 |
|
| 64 |
-
|
| 65 |
|
| 66 |
## Contact
|
| 67 |
|
|
|
|
| 5 |
- en
|
| 6 |
base_model:
|
| 7 |
- Qwen/Qwen3-Omni-30B-A3B-Instruct
|
| 8 |
+
|
| 9 |
|
| 10 |
# LAT-Audio
|
| 11 |
|
|
|
|
| 21 |
- temporal hallucination (invalid timestamps)
|
| 22 |
- timestamp drift (progressive misalignment over time)
|
| 23 |
|
| 24 |
+
|
| 25 |
|
| 26 |
## Model Description
|
| 27 |
|
|
|
|
| 38 |
|
| 39 |
This design enables robust temporal reasoning under long-context settings, where conventional direct modeling approaches often fail.
|
| 40 |
|
| 41 |
+
|
| 42 |
|
| 43 |
## Model Variants
|
| 44 |
|
|
|
|
| 48 |
| **LAT-Audio** | ✅ Yes | LAT-Chronicle | Tool-augmented multi-step reasoning model with global-to-local temporal inference |
|
| 49 |
| **LAT-Audio-Base** | ❌ No | LAT-Chronicle + in-house | Direct modeling baseline fine-tuned from Qwen3-Omni with more in-house data, offering faster and simpler inference |
|
| 50 |
|
| 51 |
+
|
| 52 |
|
| 53 |
## Quick Start
|
| 54 |
|
|
|
|
| 58 |
huggingface-cli download mcshao/LAT-Audio --local-dir ./LAT-Audio
|
| 59 |
huggingface-cli download mcshao/LAT-Audio-Base --local-dir ./LAT-Audio-Base
|
| 60 |
|
| 61 |
+
|
| 62 |
For detailed inference methods and examples, please refer to the official repository:
|
| 63 |
👉 https://github.com/alanshaoTT/LAT-Audio-Repo
|
| 64 |
|
| 65 |
+
|
| 66 |
|
| 67 |
## Contact
|
| 68 |
|