Hugging Face's logo

RichardErkhov
/
MiniLLM_-_MiniPLM-Qwen-1.2B-gguf

GGUF
Inference Endpoints
conversational
Model card Files Community
RichardErkhov commited on
Commit
dead74d
•
1 Parent(s): 8df9a2d

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +88 -0
README.md ADDED
@@ -0,0 +1,88 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Quantization made by Richard Erkhov.
2
+
3
+ [Github](https://github.com/RichardErkhov)
4
+
5
+ [Discord](https://discord.gg/pvy7H8DZMG)
6
+
7
+ [Request more models](https://github.com/RichardErkhov/quant_request)
8
+
9
+
10
+ MiniPLM-Qwen-1.2B - GGUF
11
+ - Model creator: https://huggingface.co/MiniLLM/
12
+ - Original model: https://huggingface.co/MiniLLM/MiniPLM-Qwen-1.2B/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [MiniPLM-Qwen-1.2B.Q2_K.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q2_K.gguf) | Q2_K | 0.51GB |
18
+ | [MiniPLM-Qwen-1.2B.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q3_K_S.gguf) | Q3_K_S | 0.57GB |
19
+ | [MiniPLM-Qwen-1.2B.Q3_K.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q3_K.gguf) | Q3_K | 0.61GB |
20
+ | [MiniPLM-Qwen-1.2B.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q3_K_M.gguf) | Q3_K_M | 0.61GB |
21
+ | [MiniPLM-Qwen-1.2B.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q3_K_L.gguf) | Q3_K_L | 0.63GB |
22
+ | [MiniPLM-Qwen-1.2B.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.IQ4_XS.gguf) | IQ4_XS | 0.65GB |
23
+ | [MiniPLM-Qwen-1.2B.Q4_0.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q4_0.gguf) | Q4_0 | 0.67GB |
24
+ | [MiniPLM-Qwen-1.2B.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.IQ4_NL.gguf) | IQ4_NL | 0.67GB |
25
+ | [MiniPLM-Qwen-1.2B.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q4_K_S.gguf) | Q4_K_S | 0.69GB |
26
+ | [MiniPLM-Qwen-1.2B.Q4_K.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q4_K.gguf) | Q4_K | 0.72GB |
27
+ | [MiniPLM-Qwen-1.2B.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q4_K_M.gguf) | Q4_K_M | 0.72GB |
28
+ | [MiniPLM-Qwen-1.2B.Q4_1.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q4_1.gguf) | Q4_1 | 0.72GB |
29
+ | [MiniPLM-Qwen-1.2B.Q5_0.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q5_0.gguf) | Q5_0 | 0.78GB |
30
+ | [MiniPLM-Qwen-1.2B.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q5_K_S.gguf) | Q5_K_S | 0.79GB |
31
+ | [MiniPLM-Qwen-1.2B.Q5_K.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q5_K.gguf) | Q5_K | 0.81GB |
32
+ | [MiniPLM-Qwen-1.2B.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q5_K_M.gguf) | Q5_K_M | 0.81GB |
33
+ | [MiniPLM-Qwen-1.2B.Q5_1.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q5_1.gguf) | Q5_1 | 0.83GB |
34
+ | [MiniPLM-Qwen-1.2B.Q6_K.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q6_K.gguf) | Q6_K | 0.93GB |
35
+ | [MiniPLM-Qwen-1.2B.Q8_0.gguf](https://huggingface.co/RichardErkhov/MiniLLM_-_MiniPLM-Qwen-1.2B-gguf/blob/main/MiniPLM-Qwen-1.2B.Q8_0.gguf) | Q8_0 | 1.15GB |
36
+
37
+
38
+
39
+
40
+ Original model description:
41
+ ---
42
+ library_name: transformers
43
+ license: apache-2.0
44
+ datasets:
45
+ - monology/pile-uncopyrighted
46
+ - MiniLLM/pile-diff_samp-qwen_1.8B-qwen_104M-r0.5
47
+ language:
48
+ - en
49
+ metrics:
50
+ - accuracy
51
+ pipeline_tag: text-generation
52
+ ---
53
+
54
+ # MinPLM-Qwen-1.2B
55
+
56
+ [paper](https://arxiv.org/abs/2410.17215) | [code](https://github.com/thu-coai/MiniPLM)
57
+
58
+ **MiniPLM-Qwen-1.2B** is a 1.2B model with Qwen achitecture pre-trained from scratch on [the Pile](https://huggingface.co/datasets/monology/pile-uncopyrighted) using the MiniPLM knowledge distillation framework with the [offcial QWen1.5-1.8B](https://huggingface.co/Qwen/Qwen1.5-1.8B) as the teacher model.
59
+
60
+ We also open-source the [pre-training corpus](https://huggingface.co/datasets/MiniLLM/pile-diff_samp-qwen_1.8B-qwen_104M-r0.5) refined by Difference Sampling in MiniPLM for reproducibility.
61
+
62
+ <p align='left'>
63
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/624ac662102fcdff87be51b9/2BqT0NgkmIXYlktovw9kG.png" width="1000">
64
+ </p>
65
+
66
+ ## Evaluation
67
+
68
+ MiniPLM models achieves better performance given the same computation and scales well across model sizes:
69
+
70
+ <p align='left'>
71
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/624ac662102fcdff87be51b9/EOYzajQcwQFT5PobqL3j0.png" width="1000">
72
+ </p>
73
+
74
+ ## Baseline Models
75
+ + [Conventional Pre-Training](https://huggingface.co/MiniLLM/Pretrain-Qwen-1.2B)
76
+ + [VanillaKD](https://huggingface.co/MiniLLM/VanillaKD-Pretrain-Qwen-1.2B)
77
+
78
+ ## Citation
79
+
80
+ ```bibtext
81
+ @article{miniplm,
82
+ title={MiniPLM: Knowledge Distillation for Pre-Training Language Models},
83
+ author={Yuxian Gu and Hao Zhou and Fandong Meng and Jie Zhou and Minlie Huang},
84
+ journal={arXiv preprint arXiv:2410.17215},
85
+ year={2024}
86
+ }
87
+ ```
88
+