Upload README.md with huggingface_hub
Browse files
README.md
CHANGED
|
@@ -11,9 +11,9 @@ tags:
|
|
| 11 |
- **Source**: `https://huggingface.co/LifuWang/DistillT5`
|
| 12 |
- **File**: `model.safetensors`
|
| 13 |
- **FP8 Format**: `E5M2`
|
| 14 |
-
- **Architecture**:
|
| 15 |
- **Precision Recovery Type**: LoRA
|
| 16 |
-
- **Precision Recovery File**: `model-lora-r64-
|
| 17 |
- **FP8 File**: `model-fp8-e5m2.safetensors`
|
| 18 |
|
| 19 |
## Usage (Inference)
|
|
@@ -26,8 +26,8 @@ fp8_state = load_file("model-fp8-e5m2.safetensors")
|
|
| 26 |
|
| 27 |
# Load precision recovery file if available
|
| 28 |
recovery_state = {}
|
| 29 |
-
if "model-lora-r64-
|
| 30 |
-
recovery_state = load_file("model-lora-r64-
|
| 31 |
|
| 32 |
# Reconstruct high-precision weights
|
| 33 |
reconstructed = {}
|
|
|
|
| 11 |
- **Source**: `https://huggingface.co/LifuWang/DistillT5`
|
| 12 |
- **File**: `model.safetensors`
|
| 13 |
- **FP8 Format**: `E5M2`
|
| 14 |
+
- **Architecture**: all
|
| 15 |
- **Precision Recovery Type**: LoRA
|
| 16 |
+
- **Precision Recovery File**: `model-lora-r64-all.safetensors` if available
|
| 17 |
- **FP8 File**: `model-fp8-e5m2.safetensors`
|
| 18 |
|
| 19 |
## Usage (Inference)
|
|
|
|
| 26 |
|
| 27 |
# Load precision recovery file if available
|
| 28 |
recovery_state = {}
|
| 29 |
+
if "model-lora-r64-all.safetensors":
|
| 30 |
+
recovery_state = load_file("model-lora-r64-all.safetensors")
|
| 31 |
|
| 32 |
# Reconstruct high-precision weights
|
| 33 |
reconstructed = {}
|