Bram Vanroy commited on
Commit
9c8bb6d
1 Parent(s): 0b1977f

add evaluations

Browse files
evals/truthfulqa/truthfulqa_nl_Mistral-7B-v0.1.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": {
3
+ "truthfulqa_nl": {
4
+ "mc1": 0.30828025477707005,
5
+ "mc1_stderr": 0.01649225000783665,
6
+ "mc2": 0.4516092295846247,
7
+ "mc2_stderr": 0.015055722136057215
8
+ }
9
+ },
10
+ "versions": {
11
+ "truthfulqa_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": 2,
17
+ "device": "cuda",
18
+ "no_cache": false,
19
+ "limit": null,
20
+ "bootstrap_iters": 100000,
21
+ "description_dict": {}
22
+ }
23
+ }
evals/truthfulqa/truthfulqa_nl_neural-chat-7b-v3-1.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": {
3
+ "truthfulqa_nl": {
4
+ "mc1": 0.3668789808917197,
5
+ "mc1_stderr": 0.017212610850337793,
6
+ "mc2": 0.5149795684141129,
7
+ "mc2_stderr": 0.016374624397865214
8
+ }
9
+ },
10
+ "versions": {
11
+ "truthfulqa_nl": 1
12
+ },
13
+ "config": {
14
+ "model": "hf-auto",
15
+ "model_args": "pretrained=Intel/neural-chat-7b-v3-1,use_accelerate=True,device_map_option=auto,dtype=bfloat16,load_in_8bit=True",
16
+ "batch_size": 2,
17
+ "device": "cuda",
18
+ "no_cache": false,
19
+ "limit": null,
20
+ "bootstrap_iters": 100000,
21
+ "description_dict": {}
22
+ }
23
+ }
evals/truthfulqa/truthfulqa_nl_zephyr-7b-beta.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "results": {
3
+ "truthfulqa_nl": {
4
+ "mc1": 0.3643312101910828,
5
+ "mc1_stderr": 0.017187218623737063,
6
+ "mc2": 0.5276394671698873,
7
+ "mc2_stderr": 0.016211241700768243
8
+ }
9
+ },
10
+ "versions": {
11
+ "truthfulqa_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": 2,
17
+ "device": "cuda",
18
+ "no_cache": false,
19
+ "limit": null,
20
+ "bootstrap_iters": 100000,
21
+ "description_dict": {}
22
+ }
23
+ }