Commit
•
0cab85d
1
Parent(s):
66f1bfd
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
This model has been quantized using [GPTQModel](https://github.com/ModelCloud/GPTQModel).
|
| 2 |
+
|
| 3 |
+
- **Bits**: 4
|
| 4 |
+
- **Group Size**: 128
|
| 5 |
+
- **Desc Act**: true
|
| 6 |
+
- **Static Groups**: false
|
| 7 |
+
- **Sym**: true
|
| 8 |
+
- **LM Head**: false
|
| 9 |
+
- **Damp Percent**: 0.01
|
| 10 |
+
- **True Sequential**: true
|
| 11 |
+
- **Model Name or Path**:
|
| 12 |
+
- **Model File Base Name**: model
|
| 13 |
+
- **Quant Method**: auto_round
|
| 14 |
+
- **Checkpoint Format**: gptq
|
| 15 |
+
- **Metadata**
|
| 16 |
+
- **Quantizer**: gptqmodel:0.9.8-dev0
|
| 17 |
+
- **Enable Full Range**: false
|
| 18 |
+
- **Batch Size**: 1
|
| 19 |
+
- **AMP**: true
|
| 20 |
+
- **LR Scheduler**: null
|
| 21 |
+
- **Enable Quanted Input**: true
|
| 22 |
+
- **Enable Minmax Tuning**: true
|
| 23 |
+
- **Learning Rate (LR)**: null
|
| 24 |
+
- **Minmax LR**: null
|
| 25 |
+
- **Low GPU Memory Usage**: true
|
| 26 |
+
- **Iterations (Iters)**: 200
|
| 27 |
+
- **Sequence Length (Seqlen)**: 2048
|
| 28 |
+
- **Number of Samples (Nsamples)**: 512
|
| 29 |
+
- **Sampler**: rand
|
| 30 |
+
- **Seed**: 42
|
| 31 |
+
- **Number of Blocks (Nblocks)**: 1
|
| 32 |
+
- **Gradient Accumulate Steps**: 1
|
| 33 |
+
- **Not Use Best MSE**: false
|
| 34 |
+
- **Dynamic Max Gap**: -1
|
| 35 |
+
- **Data Type**: int
|
| 36 |
+
- **Scale Data Type (Scale Dtype)**: fp16
|