elichen3051
commited on
Commit
•
fda2c0b
1
Parent(s):
912d391
End of training
Browse files- README.md +7 -1
- all_results.json +13 -0
- config.json +1 -1
- eval_results.json +16 -0
- runs/Jun08_09-35-24_40fc03db7258/events.out.tfevents.1717858594.40fc03db7258.8499.1 +3 -0
README.md
CHANGED
@@ -2,9 +2,15 @@
|
|
2 |
license: llama2
|
3 |
base_model: elichen3051/llama2-7b-sft-chat-no-template
|
4 |
tags:
|
|
|
5 |
- trl
|
6 |
- dpo
|
7 |
- generated_from_trainer
|
|
|
|
|
|
|
|
|
|
|
8 |
model-index:
|
9 |
- name: llama2-7b-sft-chat-no-template-dpo
|
10 |
results: []
|
@@ -15,7 +21,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
15 |
|
16 |
# llama2-7b-sft-chat-no-template-dpo
|
17 |
|
18 |
-
This model is a fine-tuned version of [elichen3051/llama2-7b-sft-chat-no-template](https://huggingface.co/elichen3051/llama2-7b-sft-chat-no-template) on
|
19 |
It achieves the following results on the evaluation set:
|
20 |
- Loss: 0.6026
|
21 |
- Rewards/chosen: -0.9396
|
|
|
2 |
license: llama2
|
3 |
base_model: elichen3051/llama2-7b-sft-chat-no-template
|
4 |
tags:
|
5 |
+
- alignment-handbook
|
6 |
- trl
|
7 |
- dpo
|
8 |
- generated_from_trainer
|
9 |
+
- trl
|
10 |
+
- dpo
|
11 |
+
- generated_from_trainer
|
12 |
+
datasets:
|
13 |
+
- HuggingFaceH4/ultrafeedback_binarized
|
14 |
model-index:
|
15 |
- name: llama2-7b-sft-chat-no-template-dpo
|
16 |
results: []
|
|
|
21 |
|
22 |
# llama2-7b-sft-chat-no-template-dpo
|
23 |
|
24 |
+
This model is a fine-tuned version of [elichen3051/llama2-7b-sft-chat-no-template](https://huggingface.co/elichen3051/llama2-7b-sft-chat-no-template) on the HuggingFaceH4/ultrafeedback_binarized dataset.
|
25 |
It achieves the following results on the evaluation set:
|
26 |
- Loss: 0.6026
|
27 |
- Rewards/chosen: -0.9396
|
all_results.json
CHANGED
@@ -1,5 +1,18 @@
|
|
1 |
{
|
2 |
"epoch": 0.998691442030882,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
"total_flos": 0.0,
|
4 |
"train_loss": 0.6240661440405456,
|
5 |
"train_runtime": 18073.4146,
|
|
|
1 |
{
|
2 |
"epoch": 0.998691442030882,
|
3 |
+
"eval_logits/chosen": -1.0320225954055786,
|
4 |
+
"eval_logits/rejected": -1.019870400428772,
|
5 |
+
"eval_logps/chosen": -371.21112060546875,
|
6 |
+
"eval_logps/rejected": -372.5772399902344,
|
7 |
+
"eval_loss": 0.6026462316513062,
|
8 |
+
"eval_rewards/accuracies": 0.656000018119812,
|
9 |
+
"eval_rewards/chosen": -0.9395894408226013,
|
10 |
+
"eval_rewards/margins": 0.3643726706504822,
|
11 |
+
"eval_rewards/rejected": -1.3039621114730835,
|
12 |
+
"eval_runtime": 224.7608,
|
13 |
+
"eval_samples": 2000,
|
14 |
+
"eval_samples_per_second": 8.898,
|
15 |
+
"eval_steps_per_second": 0.556,
|
16 |
"total_flos": 0.0,
|
17 |
"train_loss": 0.6240661440405456,
|
18 |
"train_runtime": 18073.4146,
|
config.json
CHANGED
@@ -25,6 +25,6 @@
|
|
25 |
"tie_word_embeddings": false,
|
26 |
"torch_dtype": "bfloat16",
|
27 |
"transformers_version": "4.41.2",
|
28 |
-
"use_cache":
|
29 |
"vocab_size": 32002
|
30 |
}
|
|
|
25 |
"tie_word_embeddings": false,
|
26 |
"torch_dtype": "bfloat16",
|
27 |
"transformers_version": "4.41.2",
|
28 |
+
"use_cache": true,
|
29 |
"vocab_size": 32002
|
30 |
}
|
eval_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 0.998691442030882,
|
3 |
+
"eval_logits/chosen": -1.0320225954055786,
|
4 |
+
"eval_logits/rejected": -1.019870400428772,
|
5 |
+
"eval_logps/chosen": -371.21112060546875,
|
6 |
+
"eval_logps/rejected": -372.5772399902344,
|
7 |
+
"eval_loss": 0.6026462316513062,
|
8 |
+
"eval_rewards/accuracies": 0.656000018119812,
|
9 |
+
"eval_rewards/chosen": -0.9395894408226013,
|
10 |
+
"eval_rewards/margins": 0.3643726706504822,
|
11 |
+
"eval_rewards/rejected": -1.3039621114730835,
|
12 |
+
"eval_runtime": 224.7608,
|
13 |
+
"eval_samples": 2000,
|
14 |
+
"eval_samples_per_second": 8.898,
|
15 |
+
"eval_steps_per_second": 0.556
|
16 |
+
}
|
runs/Jun08_09-35-24_40fc03db7258/events.out.tfevents.1717858594.40fc03db7258.8499.1
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ae9bc5cbf7002a7e939c88c6fdfabe2147648ab51ea2dbd550cedc0cd3a37773
|
3 |
+
size 828
|