Model save
Browse files- README.md +16 -21
- all_results.json +3 -3
- config.json +29 -0
- model-00001-of-00004.safetensors +1 -1
- model-00002-of-00004.safetensors +1 -1
- model-00003-of-00004.safetensors +1 -1
- model-00004-of-00004.safetensors +1 -1
- train_results.json +3 -3
- trainer_state.json +0 -0
README.md
CHANGED
@@ -2,15 +2,10 @@
|
|
2 |
license: llama3
|
3 |
base_model: Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR
|
4 |
tags:
|
5 |
-
- alignment-handbook
|
6 |
-
- trl
|
7 |
-
- dpo
|
8 |
-
- generated_from_trainer
|
9 |
- trl
|
10 |
- dpo
|
|
|
11 |
- generated_from_trainer
|
12 |
-
datasets:
|
13 |
-
- princeton-nlp/llama3-ultrafeedback-armorm
|
14 |
model-index:
|
15 |
- name: Llama-3-8B-Magpi-Pro-MTR-UltraDPO-08
|
16 |
results: []
|
@@ -19,20 +14,20 @@ model-index:
|
|
19 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
20 |
should probably proofread and complete it, then remove this comment. -->
|
21 |
|
22 |
-
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/uw-nsl/huggingface/runs/
|
23 |
# Llama-3-8B-Magpi-Pro-MTR-UltraDPO-08
|
24 |
|
25 |
-
This model is a fine-tuned version of [Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR](https://huggingface.co/Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR) on
|
26 |
It achieves the following results on the evaluation set:
|
27 |
-
- Loss: 0.
|
28 |
-
- Rewards/chosen: -4.
|
29 |
-
- Rewards/rejected: -6.
|
30 |
-
- Rewards/accuracies: 0.
|
31 |
-
- Rewards/margins: 1.
|
32 |
-
- Logps/rejected: -
|
33 |
-
- Logps/chosen: -
|
34 |
-
- Logits/rejected: -1.
|
35 |
-
- Logits/chosen: -1.
|
36 |
|
37 |
## Model description
|
38 |
|
@@ -69,10 +64,10 @@ The following hyperparameters were used during training:
|
|
69 |
|
70 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
71 |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
72 |
-
| 0.
|
73 |
-
| 0.
|
74 |
-
| 0.
|
75 |
-
| 0.
|
76 |
|
77 |
|
78 |
### Framework versions
|
|
|
2 |
license: llama3
|
3 |
base_model: Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR
|
4 |
tags:
|
|
|
|
|
|
|
|
|
5 |
- trl
|
6 |
- dpo
|
7 |
+
- alignment-handbook
|
8 |
- generated_from_trainer
|
|
|
|
|
9 |
model-index:
|
10 |
- name: Llama-3-8B-Magpi-Pro-MTR-UltraDPO-08
|
11 |
results: []
|
|
|
14 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
15 |
should probably proofread and complete it, then remove this comment. -->
|
16 |
|
17 |
+
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/uw-nsl/huggingface/runs/22x1s5jw)
|
18 |
# Llama-3-8B-Magpi-Pro-MTR-UltraDPO-08
|
19 |
|
20 |
+
This model is a fine-tuned version of [Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR](https://huggingface.co/Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR) on an unknown dataset.
|
21 |
It achieves the following results on the evaluation set:
|
22 |
+
- Loss: 0.3841
|
23 |
+
- Rewards/chosen: -4.5601
|
24 |
+
- Rewards/rejected: -6.2488
|
25 |
+
- Rewards/accuracies: 0.8589
|
26 |
+
- Rewards/margins: 1.6887
|
27 |
+
- Logps/rejected: -898.4371
|
28 |
+
- Logps/chosen: -735.5136
|
29 |
+
- Logits/rejected: -1.3606
|
30 |
+
- Logits/chosen: -1.3413
|
31 |
|
32 |
## Model description
|
33 |
|
|
|
64 |
|
65 |
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
66 |
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
67 |
+
| 0.5793 | 0.2137 | 100 | 0.5183 | -2.6521 | -3.4319 | 0.7460 | 0.7798 | -616.7447 | -544.7192 | -1.2844 | -1.2700 |
|
68 |
+
| 0.5011 | 0.4275 | 200 | 0.4428 | -3.5015 | -4.6329 | 0.7903 | 1.1314 | -736.8406 | -629.6548 | -1.3147 | -1.2977 |
|
69 |
+
| 0.3663 | 0.6412 | 300 | 0.4012 | -3.8886 | -5.3500 | 0.8387 | 1.4613 | -808.5509 | -668.3669 | -1.3327 | -1.3138 |
|
70 |
+
| 0.3856 | 0.8549 | 400 | 0.3841 | -4.5601 | -6.2488 | 0.8589 | 1.6887 | -898.4371 | -735.5136 | -1.3606 | -1.3413 |
|
71 |
|
72 |
|
73 |
### Framework versions
|
all_results.json
CHANGED
@@ -14,9 +14,9 @@
|
|
14 |
"eval_samples_per_second": 15.467,
|
15 |
"eval_steps_per_second": 0.489,
|
16 |
"total_flos": 0.0,
|
17 |
-
"train_loss": 0.
|
18 |
-
"train_runtime":
|
19 |
"train_samples": 59875,
|
20 |
-
"train_samples_per_second":
|
21 |
"train_steps_per_second": 0.047
|
22 |
}
|
|
|
14 |
"eval_samples_per_second": 15.467,
|
15 |
"eval_steps_per_second": 0.489,
|
16 |
"total_flos": 0.0,
|
17 |
+
"train_loss": 0.47420934423389477,
|
18 |
+
"train_runtime": 9989.1556,
|
19 |
"train_samples": 59875,
|
20 |
+
"train_samples_per_second": 5.994,
|
21 |
"train_steps_per_second": 0.047
|
22 |
}
|
config.json
CHANGED
@@ -0,0 +1,29 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_name_or_path": "Magpie-Align/Llama-3-8B-Magpie-Mix-300KMT-150KR",
|
3 |
+
"architectures": [
|
4 |
+
"LlamaForCausalLM"
|
5 |
+
],
|
6 |
+
"attention_bias": false,
|
7 |
+
"attention_dropout": 0.0,
|
8 |
+
"bos_token_id": 128000,
|
9 |
+
"eos_token_id": 128001,
|
10 |
+
"hidden_act": "silu",
|
11 |
+
"hidden_size": 4096,
|
12 |
+
"initializer_range": 0.02,
|
13 |
+
"intermediate_size": 14336,
|
14 |
+
"max_position_embeddings": 8192,
|
15 |
+
"mlp_bias": false,
|
16 |
+
"model_type": "llama",
|
17 |
+
"num_attention_heads": 32,
|
18 |
+
"num_hidden_layers": 32,
|
19 |
+
"num_key_value_heads": 8,
|
20 |
+
"pretraining_tp": 1,
|
21 |
+
"rms_norm_eps": 1e-05,
|
22 |
+
"rope_scaling": null,
|
23 |
+
"rope_theta": 500000.0,
|
24 |
+
"tie_word_embeddings": false,
|
25 |
+
"torch_dtype": "bfloat16",
|
26 |
+
"transformers_version": "4.42.3",
|
27 |
+
"use_cache": false,
|
28 |
+
"vocab_size": 128256
|
29 |
+
}
|
model-00001-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4976698672
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:742781e710746aeb5a909ceceb213512ba3fada72e6c45065ac9aa4fa9e0a55f
|
3 |
size 4976698672
|
model-00002-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4999802720
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:60ca27216afadb3bc4bf3d21f47e481a20c699a0916f3bf2f4c645860f7b8e6d
|
3 |
size 4999802720
|
model-00003-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4915916176
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:85b544a7380dcc0f7b18864ff1c5849da4076f9f1782c671f58a3eae25da9acb
|
3 |
size 4915916176
|
model-00004-of-00004.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1168138808
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7a2258cf1cd5f0cf0a7300092eb45d6562e0732a7eace0d3fdd1933ad5254e55
|
3 |
size 1168138808
|
train_results.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"epoch": 0.9981298423724285,
|
3 |
"total_flos": 0.0,
|
4 |
-
"train_loss": 0.
|
5 |
-
"train_runtime":
|
6 |
"train_samples": 59875,
|
7 |
-
"train_samples_per_second":
|
8 |
"train_steps_per_second": 0.047
|
9 |
}
|
|
|
1 |
{
|
2 |
"epoch": 0.9981298423724285,
|
3 |
"total_flos": 0.0,
|
4 |
+
"train_loss": 0.47420934423389477,
|
5 |
+
"train_runtime": 9989.1556,
|
6 |
"train_samples": 59875,
|
7 |
+
"train_samples_per_second": 5.994,
|
8 |
"train_steps_per_second": 0.047
|
9 |
}
|
trainer_state.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|