Upload folder using huggingface_hub
Browse files- README.md +23 -56
- pytorch_model.bin +1 -1
README.md
CHANGED
@@ -4,53 +4,10 @@ base_model: croissantllm/CroissantLLMBase
|
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
model-index:
|
7 |
-
- name: out_translation
|
8 |
results: []
|
9 |
---
|
10 |
|
11 |
-
|
12 |
-
### Usage
|
13 |
-
|
14 |
-
```python
|
15 |
-
>>> chat_input = "<|im_start|> system\nYou are a helpful assistant.<|im_end|> \n<|im_start|> user\nTraduit ce texte en anglais : \nEn 1975, la localité comptait 90 habitants, des Guiziga et lors du recensement de 2005, on y a dénombré x habitants.<|im_end|> \n<|im_start|> assistant\n"
|
16 |
-
|
17 |
-
>>> inputs = tokenizer(chat_input, return_tensors="pt").to(model.device)
|
18 |
-
|
19 |
-
>>> tokens = model.generate(**inputs, **generation_args)
|
20 |
-
Setting `pad_token_id` to `eos_token_id`:2 for open-end generation.
|
21 |
-
|
22 |
-
>>> print(tokenizer.decode(tokens[0]))
|
23 |
-
|
24 |
-
<s><|im_start|> system
|
25 |
-
You are a helpful assistant.<|im_end|>
|
26 |
-
<|im_start|> user
|
27 |
-
Traduit ce texte en anglais :
|
28 |
-
En 1975, la localité comptait 90 habitants, des Guiziga et lors du recensement de 2005, on y a dénombré x habitants.<|im_end|>
|
29 |
-
<|im_start|> assistant
|
30 |
-
When the town had 90 inhabitants in 1975, it was called Guizaga and during the census of 2005, there were x inhabitants.<|im_end|>
|
31 |
-
</s>
|
32 |
-
|
33 |
-
|
34 |
-
>>> chat_input = "<|im_start|> system\nYou are a helpful assistant.<|im_end|> \n<|im_start|> user\nCorrige les fautes dans ce texte : \nEn 1975, la localité comptait 90 habitant, des Guiziga et lors du recensement de 2005, on y a dénombrer 56 habitants.<|im_end|> \n<|im_start|> assistant\n"
|
35 |
-
|
36 |
-
>>> inputs = tokenizer(chat_input, return_tensors="pt").to(model.device)
|
37 |
-
|
38 |
-
>>> tokens = model.generate(**inputs, **generation_args)
|
39 |
-
Setting `pad_token_id` to `eos_token_id`:2 for open-end generation.
|
40 |
-
|
41 |
-
>>> print(tokenizer.decode(tokens[0]))
|
42 |
-
<s><|im_start|> system
|
43 |
-
You are a helpful assistant.<|im_end|>
|
44 |
-
<|im_start|> user
|
45 |
-
Corrige les fautes dans ce texte :
|
46 |
-
En 1975, la localité comptait 90 habitant, des Guiziga et lors du recensement de 2005, on y a dénombrer 56 habitants.<|im_end|>
|
47 |
-
|
48 |
-
<|im_start|> assistant
|
49 |
-
En 1975, la commune comptait 90 habitants dont des Guizigas et au recensement de 2005, elle en compte 56.<|im_end|>
|
50 |
-
</s>
|
51 |
-
>>>
|
52 |
-
```
|
53 |
-
|
54 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
55 |
should probably proofread and complete it, then remove this comment. -->
|
56 |
|
@@ -83,11 +40,11 @@ datasets:
|
|
83 |
type: sharegpt
|
84 |
|
85 |
chat_template: "chatml"
|
86 |
-
|
87 |
|
88 |
-
dataset_prepared_path:
|
89 |
val_set_size: 0.05
|
90 |
-
output_dir:
|
91 |
|
92 |
sequence_len: 2048
|
93 |
sample_packing: false
|
@@ -107,9 +64,9 @@ wandb_watch:
|
|
107 |
wandb_name:
|
108 |
wandb_log_model:
|
109 |
|
110 |
-
gradient_accumulation_steps:
|
111 |
micro_batch_size: 16
|
112 |
-
num_epochs:
|
113 |
optimizer: adamw_bnb_8bit
|
114 |
lr_scheduler: cosine
|
115 |
learning_rate: 0.00003
|
@@ -146,11 +103,11 @@ fsdp_config:
|
|
146 |
|
147 |
</details><br>
|
148 |
|
149 |
-
# out_translation
|
150 |
|
151 |
This model is a fine-tuned version of [croissantllm/CroissantLLMBase](https://huggingface.co/croissantllm/CroissantLLMBase) on the None dataset.
|
152 |
It achieves the following results on the evaluation set:
|
153 |
-
- Loss: 0.
|
154 |
|
155 |
## Model description
|
156 |
|
@@ -173,19 +130,29 @@ The following hyperparameters were used during training:
|
|
173 |
- train_batch_size: 16
|
174 |
- eval_batch_size: 16
|
175 |
- seed: 42
|
|
|
|
|
176 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
177 |
- lr_scheduler_type: cosine
|
178 |
- lr_scheduler_warmup_steps: 100
|
179 |
-
- num_epochs:
|
180 |
|
181 |
### Training results
|
182 |
|
183 |
| Training Loss | Epoch | Step | Validation Loss |
|
184 |
|:-------------:|:-----:|:----:|:---------------:|
|
185 |
-
|
|
186 |
-
| 0.
|
187 |
-
| 0.
|
188 |
-
| 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
189 |
|
190 |
|
191 |
### Framework versions
|
|
|
4 |
tags:
|
5 |
- generated_from_trainer
|
6 |
model-index:
|
7 |
+
- name: gpfs/workdir/fayssema/models/out_translation
|
8 |
results: []
|
9 |
---
|
10 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
11 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
12 |
should probably proofread and complete it, then remove this comment. -->
|
13 |
|
|
|
40 |
type: sharegpt
|
41 |
|
42 |
chat_template: "chatml"
|
43 |
+
default_system_message: ""
|
44 |
|
45 |
+
dataset_prepared_path: new_pii
|
46 |
val_set_size: 0.05
|
47 |
+
output_dir: /gpfs/workdir/fayssema/models/out_translation
|
48 |
|
49 |
sequence_len: 2048
|
50 |
sample_packing: false
|
|
|
64 |
wandb_name:
|
65 |
wandb_log_model:
|
66 |
|
67 |
+
gradient_accumulation_steps: 2
|
68 |
micro_batch_size: 16
|
69 |
+
num_epochs: 3
|
70 |
optimizer: adamw_bnb_8bit
|
71 |
lr_scheduler: cosine
|
72 |
learning_rate: 0.00003
|
|
|
103 |
|
104 |
</details><br>
|
105 |
|
106 |
+
# gpfs/workdir/fayssema/models/out_translation
|
107 |
|
108 |
This model is a fine-tuned version of [croissantllm/CroissantLLMBase](https://huggingface.co/croissantllm/CroissantLLMBase) on the None dataset.
|
109 |
It achieves the following results on the evaluation set:
|
110 |
+
- Loss: 0.0098
|
111 |
|
112 |
## Model description
|
113 |
|
|
|
130 |
- train_batch_size: 16
|
131 |
- eval_batch_size: 16
|
132 |
- seed: 42
|
133 |
+
- gradient_accumulation_steps: 2
|
134 |
+
- total_train_batch_size: 32
|
135 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
136 |
- lr_scheduler_type: cosine
|
137 |
- lr_scheduler_warmup_steps: 100
|
138 |
+
- num_epochs: 3
|
139 |
|
140 |
### Training results
|
141 |
|
142 |
| Training Loss | Epoch | Step | Validation Loss |
|
143 |
|:-------------:|:-----:|:----:|:---------------:|
|
144 |
+
| 2.6652 | 0.0 | 1 | 2.0261 |
|
145 |
+
| 0.2986 | 0.25 | 73 | 0.0199 |
|
146 |
+
| 0.19 | 0.5 | 146 | 0.0136 |
|
147 |
+
| 0.3032 | 0.76 | 219 | 0.0158 |
|
148 |
+
| 0.1343 | 1.01 | 292 | 0.0125 |
|
149 |
+
| 0.12 | 1.26 | 365 | 0.0117 |
|
150 |
+
| 0.2266 | 1.51 | 438 | 0.0113 |
|
151 |
+
| 0.1924 | 1.77 | 511 | 0.0097 |
|
152 |
+
| 0.1448 | 2.02 | 584 | 0.0095 |
|
153 |
+
| 0.0718 | 2.27 | 657 | 0.0098 |
|
154 |
+
| 0.1184 | 2.52 | 730 | 0.0097 |
|
155 |
+
| 0.1124 | 2.77 | 803 | 0.0098 |
|
156 |
|
157 |
|
158 |
### Framework versions
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2690937142
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e89e76e2740576ca3dab415a5c722d2fdf0d12f0e2b71f451413f4786723afd7
|
3 |
size 2690937142
|