Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,47 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
|
| 3 |
+
```
|
| 4 |
+
slices:
|
| 5 |
+
- sources:
|
| 6 |
+
- model: Undi95/Mistral-11B-OpenOrcaPlatypus
|
| 7 |
+
layer_range: [0, 48]
|
| 8 |
+
- model: "/content/drive/MyDrive/Mistral-11B-CC-Zephyr"
|
| 9 |
+
layer_range: [0, 48]
|
| 10 |
+
merge_method: slerp
|
| 11 |
+
base_model: Undi95/Mistral-11B-OpenOrcaPlatypus
|
| 12 |
+
parameters:
|
| 13 |
+
t:
|
| 14 |
+
- filter: lm_head
|
| 15 |
+
value: [0.75]
|
| 16 |
+
- filter: embed_tokens
|
| 17 |
+
value: [0.75]
|
| 18 |
+
- filter: self_attn
|
| 19 |
+
value: [0.75, 0.25]
|
| 20 |
+
- filter: mlp
|
| 21 |
+
value: [0.25, 0.75]
|
| 22 |
+
- filter: layernorm
|
| 23 |
+
value: [0.5, 0.5]
|
| 24 |
+
- filter: modelnorm
|
| 25 |
+
value: [0.75]
|
| 26 |
+
- value: 0.5 # fallback for rest of tensors
|
| 27 |
+
dtype: float16
|
| 28 |
+
```
|
| 29 |
+
|
| 30 |
+
hf-causal-experimental (pretrained=/content/drive/MyDrive/Mistral-11B-Test), limit: None, provide_description: False, num_fewshot: 0, batch_size: 4
|
| 31 |
+
| Task |Version| Metric |Value | |Stderr|
|
| 32 |
+
|-------------|------:|--------|-----:|---|-----:|
|
| 33 |
+
|arc_challenge| 0|acc |0.5597|± |0.0145|
|
| 34 |
+
| | |acc_norm|0.5819|± |0.0144|
|
| 35 |
+
|arc_easy | 0|acc |0.8308|± |0.0077|
|
| 36 |
+
| | |acc_norm|0.8215|± |0.0079|
|
| 37 |
+
|hellaswag | 0|acc |0.6371|± |0.0048|
|
| 38 |
+
| | |acc_norm|0.8213|± |0.0038|
|
| 39 |
+
|piqa | 0|acc |0.8134|± |0.0091|
|
| 40 |
+
| | |acc_norm|0.8275|± |0.0088|
|
| 41 |
+
|truthfulqa_mc| 1|mc1 |0.3990|± |0.0171|
|
| 42 |
+
| | |mc2 |0.5685|± |0.0155|
|
| 43 |
+
|winogrande | 0|acc |0.7474|± |0.0122|
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+

|