Hugging Face's logo

RichardErkhov
/
NLPinas_-_yi-bagel-2x34b-gguf

GGUF
Inference Endpoints
conversational
Model card Files Community
RichardErkhov commited on
Commit
04ffc5b
•
1 Parent(s): d3a1bbd

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +111 -0
README.md ADDED
@@ -0,0 +1,111 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ yi-bagel-2x34b - GGUF
11
+ - Model creator: https://huggingface.co/NLPinas/
12
+ - Original model: https://huggingface.co/NLPinas/yi-bagel-2x34b/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [yi-bagel-2x34b.Q2_K.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q2_K.gguf) | Q2_K | 11.94GB |
18
+ | [yi-bagel-2x34b.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.IQ3_XS.gguf) | IQ3_XS | 13.26GB |
19
+ | [yi-bagel-2x34b.IQ3_S.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.IQ3_S.gguf) | IQ3_S | 13.99GB |
20
+ | [yi-bagel-2x34b.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q3_K_S.gguf) | Q3_K_S | 13.93GB |
21
+ | [yi-bagel-2x34b.IQ3_M.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.IQ3_M.gguf) | IQ3_M | 14.5GB |
22
+ | [yi-bagel-2x34b.Q3_K.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q3_K.gguf) | Q3_K | 15.51GB |
23
+ | [yi-bagel-2x34b.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q3_K_M.gguf) | Q3_K_M | 15.51GB |
24
+ | [yi-bagel-2x34b.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q3_K_L.gguf) | Q3_K_L | 16.89GB |
25
+ | [yi-bagel-2x34b.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.IQ4_XS.gguf) | IQ4_XS | 17.36GB |
26
+ | [yi-bagel-2x34b.Q4_0.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q4_0.gguf) | Q4_0 | 18.13GB |
27
+ | [yi-bagel-2x34b.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.IQ4_NL.gguf) | IQ4_NL | 18.3GB |
28
+ | [yi-bagel-2x34b.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q4_K_S.gguf) | Q4_K_S | 18.25GB |
29
+ | [yi-bagel-2x34b.Q4_K.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q4_K.gguf) | Q4_K | 19.24GB |
30
+ | [yi-bagel-2x34b.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q4_K_M.gguf) | Q4_K_M | 12.05GB |
31
+ | [yi-bagel-2x34b.Q4_1.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q4_1.gguf) | Q4_1 | 17.51GB |
32
+ | [yi-bagel-2x34b.Q5_0.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q5_0.gguf) | Q5_0 | 17.49GB |
33
+ | [yi-bagel-2x34b.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q5_K_S.gguf) | Q5_K_S | 21.55GB |
34
+ | [yi-bagel-2x34b.Q5_K.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q5_K.gguf) | Q5_K | 22.65GB |
35
+ | [yi-bagel-2x34b.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q5_K_M.gguf) | Q5_K_M | 22.65GB |
36
+ | [yi-bagel-2x34b.Q5_1.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q5_1.gguf) | Q5_1 | 24.05GB |
37
+ | [yi-bagel-2x34b.Q6_K.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q6_K.gguf) | Q6_K | 26.28GB |
38
+ | [yi-bagel-2x34b.Q8_0.gguf](https://huggingface.co/RichardErkhov/NLPinas_-_yi-bagel-2x34b-gguf/blob/main/yi-bagel-2x34b.Q8_0.gguf) | Q8_0 | 34.03GB |
39
+
40
+
41
+
42
+
43
+ Original model description:
44
+ ---
45
+ base_model:
46
+ - jondurbin/bagel-dpo-34b-v0.2
47
+ - jondurbin/nontoxic-bagel-34b-v0.2
48
+ tags:
49
+ - mergekit
50
+ - merge
51
+ license: other
52
+ license_name: yi-license
53
+ license_link: https://huggingface.co/01-ai/Yi-34B-200K/blob/main/LICENSE
54
+ ---
55
+ # yi-bagel-2x34b
56
+
57
+ Released January 11, 2024
58
+
59
+ ![bagel-burger](bagel-burger.png)
60
+
61
+ This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). For more information, kindly refer to the model cards from jondurbin linked in the section below. This model debuted in the [leaderboard](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard) at rank #4 (January 11, 2024).
62
+
63
+ ## Merge Details
64
+ ### Merge Method
65
+
66
+ This model is an expertimental merge using the [linear](https://arxiv.org/abs/2203.05482) merge method. This is to assess the degree of which the DPO has an effect, in terms of censoring, as used in [jondurbin/bagel-dpo-34b-v0.2](https://huggingface.co/jondurbin/bagel-dpo-34b-v0.2).
67
+
68
+ ### Models Merged
69
+
70
+ The following models were included in the merge:
71
+ * [jondurbin/bagel-dpo-34b-v0.2](https://huggingface.co/jondurbin/bagel-dpo-34b-v0.2)
72
+ * [jondurbin/nontoxic-bagel-34b-v0.2](https://huggingface.co/jondurbin/nontoxic-bagel-34b-v0.2)
73
+
74
+ ## Open LLM Leaderboard Metrics (as of January 11, 2024)
75
+ | Metric | Value |
76
+ |-----------------------|-------|
77
+ | MMLU (5-shot) | 76.60 |
78
+ | ARC (25-shot) | 72.70 |
79
+ | HellaSwag (10-shot) | 85.44 |
80
+ | TruthfulQA (0-shot) | 71.42 |
81
+ | Winogrande (5-shot) | 82.72 |
82
+ | GSM8K (5-shot) | 60.73 |
83
+ | Average | 74.93 |
84
+
85
+ According to the leaderboard description, here are the benchmarks used for the evaluation:
86
+ - [MMLU](https://arxiv.org/abs/2009.03300) (5-shot) - a test to measure a text model’s multitask accuracy. The test covers 57 tasks including elementary mathematics, US history, computer science, law, and more.
87
+ - [AI2 Reasoning Challenge](https://arxiv.org/abs/1803.05457) -ARC- (25-shot) - a set of grade-school science questions.
88
+ - [HellaSwag](https://arxiv.org/abs/1905.07830) (10-shot) - a test of commonsense inference, which is easy for humans (~95%) but challenging for SOTA models.
89
+ - [TruthfulQA](https://arxiv.org/abs/2109.07958) (0-shot) - a test to measure a model’s propensity to reproduce falsehoods commonly found online.
90
+ - [Winogrande](https://arxiv.org/abs/1907.10641) (5-shot) - an adversarial and difficult Winograd benchmark at scale, for commonsense reasoning.
91
+ - [GSM8k](https://arxiv.org/abs/2110.14168) (5-shot) - diverse grade school math word problems to measure a model's ability to solve multi-step mathematical reasoning problems.
92
+
93
+ ### Configuration
94
+
95
+ The following YAML configuration was used to produce this model:
96
+
97
+ ```yaml
98
+ models:
99
+ - model: jondurbin/nontoxic-bagel-34b-v0.2
100
+ parameters:
101
+ weight: 0.5
102
+ - model: jondurbin/bagel-dpo-34b-v0.2
103
+ parameters:
104
+ weight: 0.5
105
+ merge_method: linear
106
+ dtype: float16
107
+ ```
108
+
109
+ ## Further Information
110
+ For additional information or inquiries about yi-bagel-2x34b, please contact the developer through email: [email protected].
111
+