End of training
Browse files- README.md +16 -10
- all_results.json +13 -0
- config.json +1 -1
- eval_results.json +16 -0
README.md
CHANGED
@@ -1,9 +1,15 @@
|
|
1 |
---
|
2 |
base_model: meta-llama/Llama-2-7b-hf
|
3 |
tags:
|
|
|
4 |
- trl
|
5 |
- dpo
|
6 |
- generated_from_trainer
|
|
|
|
|
|
|
|
|
|
|
7 |
model-index:
|
8 |
- name: llama2-7b-dpo-full-wo-kqa_silver_wogold-ep3
|
9 |
results: []
|
@@ -14,17 +20,17 @@ should probably proofread and complete it, then remove this comment. -->
|
|
14 |
|
15 |
# llama2-7b-dpo-full-wo-kqa_silver_wogold-ep3
|
16 |
|
17 |
-
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on
|
18 |
It achieves the following results on the evaluation set:
|
19 |
-
- Loss: 0.
|
20 |
-
- Rewards/chosen: 0.
|
21 |
-
- Rewards/rejected: -0.
|
22 |
-
- Rewards/accuracies: 0.
|
23 |
-
- Rewards/margins: 0.
|
24 |
-
- Logps/rejected: -1010.
|
25 |
-
- Logps/chosen: -408.
|
26 |
-
- Logits/rejected: -0.
|
27 |
-
- Logits/chosen: 0.
|
28 |
|
29 |
## Model description
|
30 |
|
|
|
1 |
---
|
2 |
base_model: meta-llama/Llama-2-7b-hf
|
3 |
tags:
|
4 |
+
- alignment-handbook
|
5 |
- trl
|
6 |
- dpo
|
7 |
- generated_from_trainer
|
8 |
+
- trl
|
9 |
+
- dpo
|
10 |
+
- generated_from_trainer
|
11 |
+
datasets:
|
12 |
+
- HuggingFaceH4/ultrafeedback_binarized
|
13 |
model-index:
|
14 |
- name: llama2-7b-dpo-full-wo-kqa_silver_wogold-ep3
|
15 |
results: []
|
|
|
20 |
|
21 |
# llama2-7b-dpo-full-wo-kqa_silver_wogold-ep3
|
22 |
|
23 |
+
This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on the HuggingFaceH4/ultrafeedback_binarized dataset.
|
24 |
It achieves the following results on the evaluation set:
|
25 |
+
- Loss: 0.6588
|
26 |
+
- Rewards/chosen: 0.0476
|
27 |
+
- Rewards/rejected: -0.0291
|
28 |
+
- Rewards/accuracies: 0.7912
|
29 |
+
- Rewards/margins: 0.0767
|
30 |
+
- Logps/rejected: -1010.3619
|
31 |
+
- Logps/chosen: -408.6368
|
32 |
+
- Logits/rejected: -0.5925
|
33 |
+
- Logits/chosen: 0.4749
|
34 |
|
35 |
## Model description
|
36 |
|
all_results.json
CHANGED
@@ -1,5 +1,18 @@
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
"train_loss": 0.6376642821327089,
|
4 |
"train_runtime": 2213.5067,
|
5 |
"train_samples": 8087,
|
|
|
1 |
{
|
2 |
"epoch": 1.0,
|
3 |
+
"eval_logits/chosen": 0.47494596242904663,
|
4 |
+
"eval_logits/rejected": -0.5924796462059021,
|
5 |
+
"eval_logps/chosen": -408.63677978515625,
|
6 |
+
"eval_logps/rejected": -1010.3619384765625,
|
7 |
+
"eval_loss": 0.6588297486305237,
|
8 |
+
"eval_rewards/accuracies": 0.7911764979362488,
|
9 |
+
"eval_rewards/chosen": 0.047554861754179,
|
10 |
+
"eval_rewards/margins": 0.07667416334152222,
|
11 |
+
"eval_rewards/rejected": -0.029119295999407768,
|
12 |
+
"eval_runtime": 302.3402,
|
13 |
+
"eval_samples": 2712,
|
14 |
+
"eval_samples_per_second": 8.97,
|
15 |
+
"eval_steps_per_second": 0.281,
|
16 |
"train_loss": 0.6376642821327089,
|
17 |
"train_runtime": 2213.5067,
|
18 |
"train_samples": 8087,
|
config.json
CHANGED
@@ -23,6 +23,6 @@
|
|
23 |
"tie_word_embeddings": false,
|
24 |
"torch_dtype": "bfloat16",
|
25 |
"transformers_version": "4.39.0.dev0",
|
26 |
-
"use_cache":
|
27 |
"vocab_size": 32000
|
28 |
}
|
|
|
23 |
"tie_word_embeddings": false,
|
24 |
"torch_dtype": "bfloat16",
|
25 |
"transformers_version": "4.39.0.dev0",
|
26 |
+
"use_cache": true,
|
27 |
"vocab_size": 32000
|
28 |
}
|
eval_results.json
ADDED
@@ -0,0 +1,16 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"epoch": 1.0,
|
3 |
+
"eval_logits/chosen": 0.47494596242904663,
|
4 |
+
"eval_logits/rejected": -0.5924796462059021,
|
5 |
+
"eval_logps/chosen": -408.63677978515625,
|
6 |
+
"eval_logps/rejected": -1010.3619384765625,
|
7 |
+
"eval_loss": 0.6588297486305237,
|
8 |
+
"eval_rewards/accuracies": 0.7911764979362488,
|
9 |
+
"eval_rewards/chosen": 0.047554861754179,
|
10 |
+
"eval_rewards/margins": 0.07667416334152222,
|
11 |
+
"eval_rewards/rejected": -0.029119295999407768,
|
12 |
+
"eval_runtime": 302.3402,
|
13 |
+
"eval_samples": 2712,
|
14 |
+
"eval_samples_per_second": 8.97,
|
15 |
+
"eval_steps_per_second": 0.281
|
16 |
+
}
|