Upload README.md with huggingface_hub
Browse files
README.md
ADDED
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model: amazingvince/zephyr-smol_llama-100m-sft-full
|
3 |
+
inference: false
|
4 |
+
license: apache-2.0
|
5 |
+
model-index:
|
6 |
+
- name: zephyr-smol_llama-100m-sft-full
|
7 |
+
results: []
|
8 |
+
model_creator: amazingvince
|
9 |
+
model_name: zephyr-smol_llama-100m-sft-full
|
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 |
+
# amazingvince/zephyr-smol_llama-100m-sft-full-GGUF
|
25 |
+
|
26 |
+
Quantized GGUF model files for [zephyr-smol_llama-100m-sft-full](https://huggingface.co/amazingvince/zephyr-smol_llama-100m-sft-full) from [amazingvince](https://huggingface.co/amazingvince)
|
27 |
+
|
28 |
+
|
29 |
+
| Name | Quant method | Size |
|
30 |
+
| ---- | ---- | ---- |
|
31 |
+
| [zephyr-smol_llama-100m-sft-full.fp16.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.fp16.gguf) | fp16 | 204.25 MB |
|
32 |
+
| [zephyr-smol_llama-100m-sft-full.q2_k.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.q2_k.gguf) | q2_k | 51.90 MB |
|
33 |
+
| [zephyr-smol_llama-100m-sft-full.q3_k_m.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.q3_k_m.gguf) | q3_k_m | 58.04 MB |
|
34 |
+
| [zephyr-smol_llama-100m-sft-full.q4_k_m.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.q4_k_m.gguf) | q4_k_m | 66.38 MB |
|
35 |
+
| [zephyr-smol_llama-100m-sft-full.q5_k_m.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.q5_k_m.gguf) | q5_k_m | 75.31 MB |
|
36 |
+
| [zephyr-smol_llama-100m-sft-full.q6_k.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.q6_k.gguf) | q6_k | 84.80 MB |
|
37 |
+
| [zephyr-smol_llama-100m-sft-full.q8_0.gguf](https://huggingface.co/afrideva/zephyr-smol_llama-100m-sft-full-GGUF/resolve/main/zephyr-smol_llama-100m-sft-full.q8_0.gguf) | q8_0 | 109.33 MB |
|
38 |
+
|
39 |
+
|
40 |
+
|
41 |
+
## Original Model Card:
|
42 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
43 |
+
should probably proofread and complete it, then remove this comment. -->
|
44 |
+
|
45 |
+
# zephyr-smol_llama-100m-sft-full
|
46 |
+
|
47 |
+
This model is a fine-tuned version of [BEE-spoke-data/smol_llama-101M-GQA](https://huggingface.co/BEE-spoke-data/smol_llama-101M-GQA) on an unknown dataset.
|
48 |
+
It achieves the following results on the evaluation set:
|
49 |
+
- Loss: 1.9579
|
50 |
+
|
51 |
+
## Model description
|
52 |
+
|
53 |
+
More information needed
|
54 |
+
|
55 |
+
## Intended uses & limitations
|
56 |
+
|
57 |
+
More information needed
|
58 |
+
|
59 |
+
## Training and evaluation data
|
60 |
+
|
61 |
+
More information needed
|
62 |
+
|
63 |
+
## Training procedure
|
64 |
+
|
65 |
+
### Training hyperparameters
|
66 |
+
|
67 |
+
The following hyperparameters were used during training:
|
68 |
+
- learning_rate: 2e-05
|
69 |
+
- train_batch_size: 16
|
70 |
+
- eval_batch_size: 16
|
71 |
+
- seed: 42
|
72 |
+
- distributed_type: multi-GPU
|
73 |
+
- num_devices: 2
|
74 |
+
- gradient_accumulation_steps: 4
|
75 |
+
- total_train_batch_size: 128
|
76 |
+
- total_eval_batch_size: 32
|
77 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
78 |
+
- lr_scheduler_type: cosine
|
79 |
+
- num_epochs: 1
|
80 |
+
|
81 |
+
### Training results
|
82 |
+
|
83 |
+
| Training Loss | Epoch | Step | Validation Loss |
|
84 |
+
|:-------------:|:-----:|:----:|:---------------:|
|
85 |
+
| 1.9642 | 0.7 | 1141 | 1.9578 |
|
86 |
+
|
87 |
+
|
88 |
+
### Framework versions
|
89 |
+
|
90 |
+
- Transformers 4.35.0
|
91 |
+
- Pytorch 2.1.0
|
92 |
+
- Datasets 2.14.6
|
93 |
+
- Tokenizers 0.14.1
|