Hugging Face's logo

afrideva
/
Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF

Text Generation
GGUF
Generated from Trainer
ggml
quantized
q2_k
q3_k_m
q4_k_m
q5_k_m
q6_k
q8_0
Model card Files Community
1
afrideva commited on
Commit
f531cf7
•
1 Parent(s): fbd2759

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +94 -0
README.md ADDED
@@ -0,0 +1,94 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: Tensoic/Tiny-Llama-openhermes-1.1B-step-715k-1.5T
3
+ inference: false
4
+ license: apache-2.0
5
+ model-index:
6
+ - name: out
7
+ results: []
8
+ model_creator: Tensoic
9
+ model_name: Tiny-Llama-openhermes-1.1B-step-715k-1.5T
10
+ pipeline_tag: text-generation
11
+ quantized_by: afrideva
12
+ tags:
13
+ - generated_from_trainer
14
+ - gguf
15
+ - ggml
16
+ - quantized
17
+ - q2_k
18
+ - q3_k_m
19
+ - q4_k_m
20
+ - q5_k_m
21
+ - q6_k
22
+ - q8_0
23
+ ---
24
+ # Tensoic/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF
25
+
26
+ Quantized GGUF model files for [Tiny-Llama-openhermes-1.1B-step-715k-1.5T](https://huggingface.co/Tensoic/Tiny-Llama-openhermes-1.1B-step-715k-1.5T) from [Tensoic](https://huggingface.co/Tensoic)
27
+
28
+
29
+ | Name | Quant method | Size |
30
+ | ---- | ---- | ---- |
31
+ | [tiny-llama-openhermes-1.1b-step-715k-1.5t.q2_k.gguf](https://huggingface.co/afrideva/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF/resolve/main/tiny-llama-openhermes-1.1b-step-715k-1.5t.q2_k.gguf) | q2_k | 482.14 MB |
32
+ | [tiny-llama-openhermes-1.1b-step-715k-1.5t.q3_k_m.gguf](https://huggingface.co/afrideva/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF/resolve/main/tiny-llama-openhermes-1.1b-step-715k-1.5t.q3_k_m.gguf) | q3_k_m | 549.85 MB |
33
+ | [tiny-llama-openhermes-1.1b-step-715k-1.5t.q4_k_m.gguf](https://huggingface.co/afrideva/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF/resolve/main/tiny-llama-openhermes-1.1b-step-715k-1.5t.q4_k_m.gguf) | q4_k_m | 667.81 MB |
34
+ | [tiny-llama-openhermes-1.1b-step-715k-1.5t.q5_k_m.gguf](https://huggingface.co/afrideva/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF/resolve/main/tiny-llama-openhermes-1.1b-step-715k-1.5t.q5_k_m.gguf) | q5_k_m | 782.04 MB |
35
+ | [tiny-llama-openhermes-1.1b-step-715k-1.5t.q6_k.gguf](https://huggingface.co/afrideva/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF/resolve/main/tiny-llama-openhermes-1.1b-step-715k-1.5t.q6_k.gguf) | q6_k | 903.41 MB |
36
+ | [tiny-llama-openhermes-1.1b-step-715k-1.5t.q8_0.gguf](https://huggingface.co/afrideva/Tiny-Llama-openhermes-1.1B-step-715k-1.5T-GGUF/resolve/main/tiny-llama-openhermes-1.1b-step-715k-1.5t.q8_0.gguf) | q8_0 | 1.17 GB |
37
+
38
+
39
+
40
+ ## Original Model Card:
41
+ This model is a fine-tuned version of [PY007/TinyLlama-1.1B-intermediate-step-715k-1.5T](https://huggingface.co/PY007/TinyLlama-1.1B-intermediate-step-715k-1.5T) on the [openhermes](https://huggingface.co/datasets/teknium/openhermes) dataset.
42
+ It achieves the following results on the evaluation set:
43
+ - Loss: 1.2355
44
+
45
+ ## Training procedure
46
+
47
+ ### Training hyperparameters
48
+
49
+ The following hyperparameters were used during training:
50
+ - learning_rate: 0.0002
51
+ - train_batch_size: 1
52
+ - eval_batch_size: 1
53
+ - seed: 42
54
+ - distributed_type: multi-GPU
55
+ - num_devices: 8
56
+ - total_train_batch_size: 8
57
+ - total_eval_batch_size: 8
58
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
59
+ - lr_scheduler_type: cosine
60
+ - lr_scheduler_warmup_steps: 100
61
+ - num_epochs: 1
62
+
63
+ ### Training results
64
+
65
+ | Training Loss | Epoch | Step | Validation Loss |
66
+ |:-------------:|:-----:|:-----:|:---------------:|
67
+ | 1.4654 | 0.0 | 1 | 3.5326 |
68
+ | 1.2162 | 0.05 | 1503 | 1.9335 |
69
+ | 1.1918 | 0.1 | 3006 | 1.7391 |
70
+ | 1.4188 | 0.15 | 4509 | 1.7574 |
71
+ | 1.8281 | 0.2 | 6012 | 1.6704 |
72
+ | 0.8639 | 0.25 | 7515 | 1.7459 |
73
+ | 1.3764 | 0.3 | 9018 | 1.6832 |
74
+ | 2.1172 | 0.35 | 10521 | 1.6398 |
75
+ | 1.1855 | 0.4 | 12024 | 1.6007 |
76
+ | 1.5604 | 0.45 | 13527 | 1.5256 |
77
+ | 1.0224 | 0.5 | 15030 | 1.4891 |
78
+ | 1.5582 | 0.55 | 16533 | 1.4903 |
79
+ | 0.9489 | 0.6 | 18036 | 1.4179 |
80
+ | 1.67 | 0.65 | 19539 | 1.4585 |
81
+ | 0.8542 | 0.7 | 21042 | 1.3810 |
82
+ | 1.5301 | 0.75 | 22545 | 1.3645 |
83
+ | 0.951 | 0.8 | 24048 | 1.3087 |
84
+ | 1.1791 | 0.85 | 25551 | 1.3018 |
85
+ | 1.3342 | 0.9 | 27054 | 1.2595 |
86
+ | 1.1221 | 0.95 | 28557 | 1.2355 |
87
+
88
+
89
+ ### Framework versions
90
+
91
+ - Transformers 4.34.0
92
+ - Pytorch 2.0.1+cu117
93
+ - Datasets 2.14.6
94
+ - Tokenizers 0.14.1