Bram Vanroy
commited on
Commit
•
6a37317
1
Parent(s):
351f9fe
add geitje
Browse files- evals/arc/arc_nl_GEITje-7B-chat.json +23 -0
- evals/arc/arc_nl_GEITje-7B.json +23 -0
- evals/hellaswag/hellaswag_nl_GEITje-7B-chat.json +23 -0
- evals/hellaswag/hellaswag_nl_GEITje-7B.json +23 -0
- evals/models.json +18 -2
- evals/truthfulqa/truthfulqa_nl_GEITje-7B-chat.json +23 -0
- evals/truthfulqa/truthfulqa_nl_GEITje-7B.json +23 -0
- evals/truthfulqa/truthfulqa_nl_Mixtral-8x7B-v0.1.json +23 -0
evals/arc/arc_nl_GEITje-7B-chat.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"arc_nl": {
|
4 |
+
"acc": 0.41488451668092385,
|
5 |
+
"acc_stderr": 0.014416604608618254,
|
6 |
+
"acc_norm": 0.42001710863986313,
|
7 |
+
"acc_norm_stderr": 0.014441744604939771
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"arc_nl": 0
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=Rijgersberg/GEITje-7B-chat,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
|
16 |
+
"batch_size": 4,
|
17 |
+
"device": "cuda",
|
18 |
+
"no_cache": false,
|
19 |
+
"limit": null,
|
20 |
+
"bootstrap_iters": 100000,
|
21 |
+
"description_dict": {}
|
22 |
+
}
|
23 |
+
}
|
evals/arc/arc_nl_GEITje-7B.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"arc_nl": {
|
4 |
+
"acc": 0.3857998289136014,
|
5 |
+
"acc_stderr": 0.014243429178521566,
|
6 |
+
"acc_norm": 0.38152266894781867,
|
7 |
+
"acc_norm_stderr": 0.014213487288956106
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"arc_nl": 0
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=Rijgersberg/GEITje-7B,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
|
16 |
+
"batch_size": 4,
|
17 |
+
"device": "cuda",
|
18 |
+
"no_cache": false,
|
19 |
+
"limit": null,
|
20 |
+
"bootstrap_iters": 100000,
|
21 |
+
"description_dict": {}
|
22 |
+
}
|
23 |
+
}
|
evals/hellaswag/hellaswag_nl_GEITje-7B-chat.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"hellaswag_nl": {
|
4 |
+
"acc": 0.5085806799784134,
|
5 |
+
"acc_stderr": 0.005194057692409769,
|
6 |
+
"acc_norm": 0.6658391797085806,
|
7 |
+
"acc_norm_stderr": 0.004900756947736894
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"hellaswag_nl": 1
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=Rijgersberg/GEITje-7B-chat,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
|
16 |
+
"batch_size": 4,
|
17 |
+
"device": "cuda",
|
18 |
+
"no_cache": false,
|
19 |
+
"limit": null,
|
20 |
+
"bootstrap_iters": 100000,
|
21 |
+
"description_dict": {}
|
22 |
+
}
|
23 |
+
}
|
evals/hellaswag/hellaswag_nl_GEITje-7B.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"hellaswag_nl": {
|
4 |
+
"acc": 0.4991905018888289,
|
5 |
+
"acc_stderr": 0.005194815910065167,
|
6 |
+
"acc_norm": 0.6545062061521857,
|
7 |
+
"acc_norm_stderr": 0.004940577731224625
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"hellaswag_nl": 1
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=Rijgersberg/GEITje-7B,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
|
16 |
+
"batch_size": 4,
|
17 |
+
"device": "cuda",
|
18 |
+
"no_cache": false,
|
19 |
+
"limit": null,
|
20 |
+
"bootstrap_iters": 100000,
|
21 |
+
"description_dict": {}
|
22 |
+
}
|
23 |
+
}
|
evals/models.json
CHANGED
@@ -1,4 +1,20 @@
|
|
1 |
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
"gpt-neo-1.3b-dutch": {
|
3 |
"compute_dtype": "bfloat16",
|
4 |
"dutch_coverage": "pretrained",
|
@@ -93,7 +109,7 @@
|
|
93 |
"model_name": "mistralai/Mixtral-8x7B-v0.1",
|
94 |
"model_type": "not-given",
|
95 |
"num_parameters": 46702792704,
|
96 |
-
"quantization":
|
97 |
},
|
98 |
"neural-chat-7b-v3-1": {
|
99 |
"compute_dtype": "bfloat16",
|
@@ -127,4 +143,4 @@
|
|
127 |
"num_parameters": 7241732096,
|
128 |
"quantization": "8-bit"
|
129 |
}
|
130 |
-
}
|
|
|
1 |
{
|
2 |
+
"geitje-7b": {
|
3 |
+
"compute_dtype": "auto",
|
4 |
+
"dutch_coverage": "fine-tuned",
|
5 |
+
"model_name": "Rijgersberg/GEITje-7B",
|
6 |
+
"model_type": "fine-tuned",
|
7 |
+
"num_parameters": 7241732096,
|
8 |
+
"quantization": "8-bit"
|
9 |
+
},
|
10 |
+
"geitje-7b-chat": {
|
11 |
+
"compute_dtype": "auto",
|
12 |
+
"dutch_coverage": "fine-tuned",
|
13 |
+
"model_name": "Rijgersberg/GEITje-7B-chat",
|
14 |
+
"model_type": "instruction-tuned",
|
15 |
+
"num_parameters": 7241732096,
|
16 |
+
"quantization": "8-bit"
|
17 |
+
},
|
18 |
"gpt-neo-1.3b-dutch": {
|
19 |
"compute_dtype": "bfloat16",
|
20 |
"dutch_coverage": "pretrained",
|
|
|
109 |
"model_name": "mistralai/Mixtral-8x7B-v0.1",
|
110 |
"model_type": "not-given",
|
111 |
"num_parameters": 46702792704,
|
112 |
+
"quantization": "8-bit"
|
113 |
},
|
114 |
"neural-chat-7b-v3-1": {
|
115 |
"compute_dtype": "bfloat16",
|
|
|
143 |
"num_parameters": 7241732096,
|
144 |
"quantization": "8-bit"
|
145 |
}
|
146 |
+
}
|
evals/truthfulqa/truthfulqa_nl_GEITje-7B-chat.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"truthfulqa_nl": {
|
4 |
+
"mc1": 0.2802547770700637,
|
5 |
+
"mc1_stderr": 0.016040129721386853,
|
6 |
+
"mc2": 0.46430395880030356,
|
7 |
+
"mc2_stderr": 0.015383562770588449
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"truthfulqa_nl": 1
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=Rijgersberg/GEITje-7B-chat,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
|
16 |
+
"batch_size": 4,
|
17 |
+
"device": "cuda",
|
18 |
+
"no_cache": false,
|
19 |
+
"limit": null,
|
20 |
+
"bootstrap_iters": 100000,
|
21 |
+
"description_dict": {}
|
22 |
+
}
|
23 |
+
}
|
evals/truthfulqa/truthfulqa_nl_GEITje-7B.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"truthfulqa_nl": {
|
4 |
+
"mc1": 0.27388535031847133,
|
5 |
+
"mc1_stderr": 0.01592681625497211,
|
6 |
+
"mc2": 0.4321017012137338,
|
7 |
+
"mc2_stderr": 0.01473465985240878
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"truthfulqa_nl": 1
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=Rijgersberg/GEITje-7B,use_accelerate=True,device_map_option=auto,dtype=auto,load_in_8bit=True",
|
16 |
+
"batch_size": 4,
|
17 |
+
"device": "cuda",
|
18 |
+
"no_cache": false,
|
19 |
+
"limit": null,
|
20 |
+
"bootstrap_iters": 100000,
|
21 |
+
"description_dict": {}
|
22 |
+
}
|
23 |
+
}
|
evals/truthfulqa/truthfulqa_nl_Mixtral-8x7B-v0.1.json
ADDED
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"results": {
|
3 |
+
"truthfulqa_nl": {
|
4 |
+
"mc1": 0.310828025477707,
|
5 |
+
"mc1_stderr": 0.016529733724696277,
|
6 |
+
"mc2": 0.4629846929417042,
|
7 |
+
"mc2_stderr": 0.01508660067127546
|
8 |
+
}
|
9 |
+
},
|
10 |
+
"versions": {
|
11 |
+
"truthfulqa_nl": 1
|
12 |
+
},
|
13 |
+
"config": {
|
14 |
+
"model": "hf-auto",
|
15 |
+
"model_args": "pretrained=mistralai/Mixtral-8x7B-v0.1,use_accelerate=True,device_map_option=auto,dtype=auto,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 |
+
}
|