TearGosling
commited on
Commit
•
282cb76
1
Parent(s):
0e0bdbd
End of training
Browse files- README.md +146 -0
- pytorch_model.bin +2 -2
README.md
ADDED
@@ -0,0 +1,146 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
library_name: transformers
|
3 |
+
license: apache-2.0
|
4 |
+
base_model: mistralai/Mistral-Nemo-Base-2407
|
5 |
+
tags:
|
6 |
+
- axolotl
|
7 |
+
- generated_from_trainer
|
8 |
+
model-index:
|
9 |
+
- name: pyg3v1-nemo-3ep-ckpts
|
10 |
+
results: []
|
11 |
+
---
|
12 |
+
|
13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
14 |
+
should probably proofread and complete it, then remove this comment. -->
|
15 |
+
|
16 |
+
[<img src="https://raw.githubusercontent.com/axolotl-ai-cloud/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/axolotl-ai-cloud/axolotl)
|
17 |
+
<details><summary>See axolotl config</summary>
|
18 |
+
|
19 |
+
axolotl version: `0.4.1`
|
20 |
+
```yaml
|
21 |
+
base_model: mistralai/Mistral-Nemo-Base-2407
|
22 |
+
model_type: AutoModelForCausalLM
|
23 |
+
tokenizer_type: AutoTokenizer
|
24 |
+
|
25 |
+
load_in_8bit: false
|
26 |
+
load_in_4bit: false
|
27 |
+
strict: false
|
28 |
+
|
29 |
+
plugins:
|
30 |
+
- axolotl.integrations.liger.LigerPlugin
|
31 |
+
liger_rope: true
|
32 |
+
liger_rms_norm: true
|
33 |
+
liger_swiglu: true
|
34 |
+
liger_fused_linear_cross_entropy: true
|
35 |
+
|
36 |
+
chat_template: chatml
|
37 |
+
|
38 |
+
datasets:
|
39 |
+
- path: PygTesting/pyg3v1
|
40 |
+
type: sharegpt
|
41 |
+
conversation: chatml
|
42 |
+
|
43 |
+
hub_model_id: PygTesting/pyg3v1-nemo-3ep-ckpts
|
44 |
+
hub_strategy: every_save
|
45 |
+
hf_use_auth_token: true
|
46 |
+
|
47 |
+
dataset_prepared_path: ./data/pyg3v1-data/tokenized
|
48 |
+
val_set_size: 0.0
|
49 |
+
output_dir: ./data/pyg3v1-nemo-2eps-out
|
50 |
+
|
51 |
+
sequence_len: 8192
|
52 |
+
sample_packing: true
|
53 |
+
#eval_sample_packing: false
|
54 |
+
pad_to_sequence_len: true
|
55 |
+
|
56 |
+
wandb_project: pyg3v1-nemo
|
57 |
+
wandb_entity:
|
58 |
+
wandb_watch:
|
59 |
+
wandb_name: more_eps_lower_lr
|
60 |
+
wandb_log_model:
|
61 |
+
|
62 |
+
#unsloth_cross_entropy_loss: true
|
63 |
+
|
64 |
+
gradient_accumulation_steps: 4
|
65 |
+
micro_batch_size: 4
|
66 |
+
num_epochs: 3
|
67 |
+
optimizer: adamw_torch_fused
|
68 |
+
lr_scheduler: cosine
|
69 |
+
learning_rate: 0.0000075
|
70 |
+
|
71 |
+
train_on_inputs: false
|
72 |
+
group_by_length: false
|
73 |
+
bf16: auto
|
74 |
+
fp16:
|
75 |
+
tf32: false
|
76 |
+
|
77 |
+
gradient_checkpointing: true
|
78 |
+
gradient_checkpointing_kwargs:
|
79 |
+
use_reentrant: false
|
80 |
+
early_stopping_patience:
|
81 |
+
resume_from_checkpoint:
|
82 |
+
logging_steps: 1
|
83 |
+
xformers_attention:
|
84 |
+
flash_attention: true
|
85 |
+
|
86 |
+
warmup_ratio: 0.03
|
87 |
+
evals_per_epoch: 0
|
88 |
+
eval_table_size:
|
89 |
+
saves_per_epoch: 3
|
90 |
+
debug:
|
91 |
+
deepspeed: deepspeed_configs/zero1.json
|
92 |
+
weight_decay: 0.01
|
93 |
+
fsdp:
|
94 |
+
fsdp_config:
|
95 |
+
special_tokens:
|
96 |
+
pad_token: <pad>
|
97 |
+
|
98 |
+
```
|
99 |
+
|
100 |
+
</details><br>
|
101 |
+
|
102 |
+
# pyg3v1-nemo-3ep-ckpts
|
103 |
+
|
104 |
+
This model is a fine-tuned version of [mistralai/Mistral-Nemo-Base-2407](https://huggingface.co/mistralai/Mistral-Nemo-Base-2407) on the None dataset.
|
105 |
+
|
106 |
+
## Model description
|
107 |
+
|
108 |
+
More information needed
|
109 |
+
|
110 |
+
## Intended uses & limitations
|
111 |
+
|
112 |
+
More information needed
|
113 |
+
|
114 |
+
## Training and evaluation data
|
115 |
+
|
116 |
+
More information needed
|
117 |
+
|
118 |
+
## Training procedure
|
119 |
+
|
120 |
+
### Training hyperparameters
|
121 |
+
|
122 |
+
The following hyperparameters were used during training:
|
123 |
+
- learning_rate: 7.5e-06
|
124 |
+
- train_batch_size: 4
|
125 |
+
- eval_batch_size: 4
|
126 |
+
- seed: 42
|
127 |
+
- distributed_type: multi-GPU
|
128 |
+
- num_devices: 8
|
129 |
+
- gradient_accumulation_steps: 4
|
130 |
+
- total_train_batch_size: 128
|
131 |
+
- total_eval_batch_size: 32
|
132 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
133 |
+
- lr_scheduler_type: cosine
|
134 |
+
- lr_scheduler_warmup_steps: 29
|
135 |
+
- num_epochs: 3
|
136 |
+
|
137 |
+
### Training results
|
138 |
+
|
139 |
+
|
140 |
+
|
141 |
+
### Framework versions
|
142 |
+
|
143 |
+
- Transformers 4.45.0.dev0
|
144 |
+
- Pytorch 2.4.0+rocm6.1
|
145 |
+
- Datasets 2.21.0
|
146 |
+
- Tokenizers 0.19.1
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
-
size
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:1b5ac10dd3afaa3661ced2591c15e36fb7b8ed288066e5b66386150d71339785
|
3 |
+
size 24495615534
|