File size: 1,480 Bytes
07b7148
 
 
 
 
 
 
 
 
 
 
 
 
 
 
ee75e96
07b7148
ee75e96
 
 
07b7148
ee75e96
07b7148
ee75e96
 
07b7148
ee75e96
 
 
 
07b7148
 
 
 
 
 
f35bab9
07b7148
 
 
 
cc3e943
07b7148
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
---
license: other
base_model: microsoft/phi-1_5
tags:
- generated_from_trainer
model-index:
- name: salesGPT_v1
  results: []
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# salesGPT_v1

# salesGPT_v1_1

Extension of salesGPT_v1: https://huggingface.co/goendalf666/salesGPT_v1
In this version different LoRa configurations are tested.
The backbone of this model is phi1_5: https://huggingface.co/microsoft/phi-1_5

## References

The main model is salesGPT_v2: https://huggingface.co/goendalf666/salesGPT_v2
To see the code for the generation, inference and dataset generation see the github repo: https://github.com/tom813/salesGPT_foundation

The datasets that are used to fine tune this model are:
https://huggingface.co/datasets/goendalf666/sales-conversations-instruction-base
https://huggingface.co/datasets/goendalf666/sales-textbook_for_convincing_and_selling
https://huggingface.co/datasets/goendalf666/sales-conversations-2


### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- training_steps: 80000


### Framework versions

- Transformers 4.32.1
- Pytorch 2.1.0.dev20230829+cu121
- Datasets 2.14.5
- Tokenizers 0.13.3