metadata
license: apache-2.0
model-index:
- name: OpenHermes-2.5-neural-chat-v3-3-Slerp
results:
- task:
type: text-generation
name: Text Generation
dataset:
name: AI2 Reasoning Challenge (25-Shot)
type: ai2_arc
config: ARC-Challenge
split: test
args:
num_few_shot: 25
metrics:
- type: acc_norm
value: 68.09
name: normalized accuracy
- task:
type: text-generation
name: Text Generation
dataset:
name: HellaSwag (10-Shot)
type: hellaswag
split: validation
args:
num_few_shot: 10
metrics:
- type: acc_norm
value: 86.2
name: normalized accuracy
- task:
type: text-generation
name: Text Generation
dataset:
name: MMLU (5-Shot)
type: cais/mmlu
config: all
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 64.26
name: accuracy
- task:
type: text-generation
name: Text Generation
dataset:
name: TruthfulQA (0-shot)
type: truthful_qa
config: multiple_choice
split: validation
args:
num_few_shot: 0
metrics:
- type: mc2
value: 62.78
- task:
type: text-generation
name: Text Generation
dataset:
name: Winogrande (5-shot)
type: winogrande
config: winogrande_xl
split: validation
args:
num_few_shot: 5
metrics:
- type: acc
value: 79.16
name: accuracy
- task:
type: text-generation
name: Text Generation
dataset:
name: GSM8k (5-shot)
type: gsm8k
config: main
split: test
args:
num_few_shot: 5
metrics:
- type: acc
value: 67.78
name: accuracy
tags:
- merge
OpenHermes-2.5-neural-chat-v3-3-Slerp
This is the model for OpenHermes-2.5-neural-chat-v3-3-Slerp. I used mergekit to merge models.
Prompt Templates
You can use these prompt templates, but I recommend using ChatML.
ChatML (OpenHermes-2.5-Mistral-7B):
<|im_start|>system
{system}<|im_end|>
<|im_start|>user
{user}<|im_end|>
<|im_start|>assistant
{asistant}<|im_end|>
neural-chat-7b-v3-3:
### System:
{system}
### User:
{user}
### Assistant:
Yaml Config to reproduce
slices:
- sources:
- model: teknium/OpenHermes-2.5-Mistral-7B
layer_range: [0, 32]
- model: Intel/neural-chat-7b-v3-3
layer_range: [0, 32]
merge_method: slerp
base_model: mistralai/Mistral-7B-v0.1
parameters:
t:
- filter: self_attn
value: [0, 0.5, 0.3, 0.7, 1]
- filter: mlp
value: [1, 0.5, 0.7, 0.3, 0]
- value: 0.5 # fallback for rest of tensors
dtype: bfloat16
Quantizationed versions
Quantizationed versions of this model is available thanks to TheBloke.
GPTQ
GGUF
AWQ
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 71.38 |
| ARC (25-shot) | 68.09 |
| HellaSwag (10-shot) | 86.2 |
| MMLU (5-shot) | 64.26 |
| TruthfulQA (0-shot) | 62.78 |
| Winogrande (5-shot) | 79.16 |
| GSM8K (5-shot) | 67.78 |