Bram Vanroy commited on
Commit
3b151fc
1 Parent(s): 6087d40
evals/hellaswag/hellaswag_nl_Mistral-7B-v0.1.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": {
3
+ "hellaswag_nl": {
4
+ "acc": 0.44079870480302213,
5
+ "acc_stderr": 0.005158280633507224,
6
+ "acc_norm": 0.5840259039395574,
7
+ "acc_norm_stderr": 0.005120942804814836
8
+ }
9
+ },
10
+ "versions": {
11
+ "hellaswag_nl": 1
12
+ },
13
+ "config": {
14
+ "model": "hf-auto",
15
+ "model_args": "pretrained=mistralai/Mistral-7B-v0.1,use_accelerate=True,device_map_option=auto,dtype=bfloat16,load_in_8bit=True",
16
+ "batch_size": "auto",
17
+ "device": "cuda",
18
+ "no_cache": false,
19
+ "limit": null,
20
+ "bootstrap_iters": 100000,
21
+ "description_dict": {}
22
+ }
23
+ }
evals/hellaswag/hellaswag_nl_zephyr-7b-beta.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": {
3
+ "hellaswag_nl": {
4
+ "acc": 0.443712898003238,
5
+ "acc_stderr": 0.005161800895769491,
6
+ "acc_norm": 0.5756071235833783,
7
+ "acc_norm_stderr": 0.005135087516086676
8
+ }
9
+ },
10
+ "versions": {
11
+ "hellaswag_nl": 1
12
+ },
13
+ "config": {
14
+ "model": "hf-auto",
15
+ "model_args": "pretrained=HuggingFaceH4/zephyr-7b-beta,use_accelerate=True,device_map_option=auto,dtype=bfloat16,load_in_8bit=True",
16
+ "batch_size": "auto",
17
+ "device": "cuda",
18
+ "no_cache": false,
19
+ "limit": null,
20
+ "bootstrap_iters": 100000,
21
+ "description_dict": {}
22
+ }
23
+ }
evals/mmlu/mmlu_nl_zephyr-7b-beta.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": {
3
+ "mmlu_nl": {
4
+ "acc": 0.4840251954162556,
5
+ "acc_stderr": 0.00435367934077066,
6
+ "acc_norm": 0.4326477954010776,
7
+ "acc_norm_stderr": 0.00431620265785813
8
+ }
9
+ },
10
+ "versions": {
11
+ "mmlu_nl": 0
12
+ },
13
+ "config": {
14
+ "model": "hf-auto",
15
+ "model_args": "pretrained=HuggingFaceH4/zephyr-7b-beta,use_accelerate=True,device_map_option=auto,dtype=bfloat16,load_in_8bit=True",
16
+ "batch_size": "auto",
17
+ "device": "cuda",
18
+ "no_cache": false,
19
+ "limit": null,
20
+ "bootstrap_iters": 100000,
21
+ "description_dict": {}
22
+ }
23
+ }