Ramadhirra
commited on
Commit
•
19fb3be
1
Parent(s):
76894ea
Update README.md
Browse files
README.md
CHANGED
@@ -1 +1,42 @@
|
|
1 |
-
## ExLlamaV2 quantizations for [ChaoticNeutrals/Bepis_9B](https://huggingface.co/ChaoticNeutrals/Bepis_9B/).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
## ExLlamaV2 quantizations for [ChaoticNeutrals/Bepis_9B](https://huggingface.co/ChaoticNeutrals/Bepis_9B/).
|
2 |
+
|
3 |
+
All credits belong to [ChaoticNeutrals](https://huggingface.co/ChaoticNeutrals).
|
4 |
+
|
5 |
+
## Original model information:
|
6 |
+
|
7 |
+
# Bepis
|
8 |
+
|
9 |
+
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/626dfb8786671a29c715f8a9/H0_oJhrIEGBIwogB77p5w.jpeg)
|
10 |
+
|
11 |
+
A new 9B model from jeiku. This one is smart, proficient at markdown, knows when to stop talking, and is quite soulful. The merge was an equal 3 way split between https://huggingface.co/ChaoticNeutrals/Prodigy_7B, https://huggingface.co/Test157t/Prima-LelantaclesV6-7b, and https://huggingface.co/cgato/Thespis-CurtainCall-7b-v0.2.1
|
12 |
+
|
13 |
+
If there's any 7B to 11B merge or finetune you'd like to see, feel free to leave a message.
|
14 |
+
|
15 |
+
The following YAML configuration was used to produce this model:
|
16 |
+
|
17 |
+
```yaml
|
18 |
+
slices:
|
19 |
+
- sources:
|
20 |
+
- model: primathespis
|
21 |
+
layer_range: [0, 20]
|
22 |
+
- sources:
|
23 |
+
- model: prodigalthespis
|
24 |
+
layer_range: [12, 32]
|
25 |
+
merge_method: passthrough
|
26 |
+
dtype: float16
|
27 |
+
|
28 |
+
```
|
29 |
+
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
30 |
+
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_ChaoticNeutrals__Bepis_9B)
|
31 |
+
|
32 |
+
| Metric |Value|
|
33 |
+
|---------------------------------|----:|
|
34 |
+
|Avg. |62.40|
|
35 |
+
|AI2 Reasoning Challenge (25-Shot)|62.54|
|
36 |
+
|HellaSwag (10-Shot) |80.12|
|
37 |
+
|MMLU (5-Shot) |62.84|
|
38 |
+
|TruthfulQA (0-shot) |53.30|
|
39 |
+
|Winogrande (5-shot) |76.48|
|
40 |
+
|GSM8k (5-shot) |39.12|
|
41 |
+
|
42 |
+
|