RichardErkhov commited on
Commit
401f5b6
1 Parent(s): 8fdff1c

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +78 -0
README.md ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ TinyAiroboros-2.2.1 - AWQ
11
+ - Model creator: https://huggingface.co/aloobun/
12
+ - Original model: https://huggingface.co/aloobun/TinyAiroboros-2.2.1/
13
+
14
+
15
+
16
+
17
+ Original model description:
18
+ ---
19
+ datasets:
20
+ - jondurbin/airoboros-2.2.1
21
+ tags:
22
+ - airoboros
23
+ - tinyllama
24
+ ---
25
+
26
+ This model is a fine-tuned version of PY007/TinyLlama-1.1B-Chat-v0.3 (finetuned on 15k rows of airoboros-2.2.1 dataset)
27
+
28
+ ## lm-eval
29
+
30
+ ```
31
+ | Task |Version| Metric |Value | |Stderr|
32
+ |-------------|------:|--------|-----:|---|-----:|
33
+ |arc_challenge| 0|acc |0.2671|± |0.0129|
34
+ | | |acc_norm|0.2850|± |0.0132|
35
+ |arc_easy | 0|acc |0.5673|± |0.0102|
36
+ | | |acc_norm|0.5109|± |0.0103|
37
+ |boolq | 1|acc |0.6040|± |0.0086|
38
+ |hellaswag | 0|acc |0.4155|± |0.0049|
39
+ | | |acc_norm|0.5420|± |0.0050|
40
+ |openbookqa | 0|acc |0.2200|± |0.0185|
41
+ | | |acc_norm|0.3420|± |0.0212|
42
+ |piqa | 0|acc |0.7057|± |0.0106|
43
+ | | |acc_norm|0.6970|± |0.0107|
44
+ |winogrande | 0|acc |0.5714|± |0.0139|
45
+
46
+ ```
47
+ ## Usage:
48
+
49
+ ```
50
+ from transformers import AutoTokenizer
51
+ import transformers
52
+ import torch
53
+
54
+ model = "aloobun/TinyAiroboros-2.2.1"
55
+
56
+ tokenizer = AutoTokenizer.from_pretrained(model)
57
+ pipeline = transformers.pipeline(
58
+ "text-generation",
59
+ model=model,
60
+ torch_dtype=torch.float16,
61
+ device_map="auto",
62
+ )
63
+
64
+ prompt = "Write a short story about a dystopian society."
65
+
66
+ sequences = pipeline(
67
+ f'[INST] {prompt} [/INST]',
68
+ do_sample=True,
69
+ top_k=10,
70
+ num_return_sequences=1,
71
+ eos_token_id=tokenizer.eos_token_id,
72
+ max_length=1024,
73
+ )
74
+ for seq in sequences:
75
+ print(f"Result: {seq['generated_text']}")
76
+ ```
77
+
78
+