metadata

language:
  - en
  - ru
license: llama2
tags:
  - merge
  - mergekit
  - nsfw
  - not-for-all-audiences
model-index:
  - name: Gembo-v1-70b
    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: 71.25
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ChuckMcSneed/Gembo-v1-70b
          name: Open LLM Leaderboard
      - 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: 86.98
            name: normalized accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ChuckMcSneed/Gembo-v1-70b
          name: Open LLM Leaderboard
      - 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: 70.85
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ChuckMcSneed/Gembo-v1-70b
          name: Open LLM Leaderboard
      - 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: 63.25
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ChuckMcSneed/Gembo-v1-70b
          name: Open LLM Leaderboard
      - 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: 80.51
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ChuckMcSneed/Gembo-v1-70b
          name: Open LLM Leaderboard
      - 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: 50.19
            name: accuracy
        source:
          url: >-
            https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=ChuckMcSneed/Gembo-v1-70b
          name: Open LLM Leaderboard

This is my first "serious"(with practical use cases) experimental merge. Judge harshly. Mainly made for RP, but should be okay as an assistant. Turned out quite good, considering the amount of LORAs I merged into it.

Observations

GPTisms and repetition: put temperature and rep. pen. higher, make GPTisms stop sequences
A bit different than the ususal stuff; I'd say that it has so much slop in it that it unslops itself
Lightly censored
Fairly neutral, can be violent if you ask it really good, Goliath is a bit better at it
Has a bit of optimism baked in, but it's not very severe
Doesn't know when to stop, can be quite verbose or just stop almost immediately(maybe wants to use LimaRP settings idk)
Sometimes can't handle '
Second model that tried to be funny unprompted to me(First one was Goliath)
Moderately intelligent
Quite creative

Naming

Internal name of this model was euryale-guano-saiga-med-janboros-kim-lima-wiz-tony-d30-s40, but I decided to keep it short, and since it was iteration G in my files, I called it "Gembo".

Quants

Thanks for GGUF quants, @Artefact2!

GGUF

Prompt format

Alpaca. You can also try some other formats, I'm pretty sure it has a lot of them from all those merges.

### Instruction:
{instruction}

### Response:

Settings

As I already mentioned, high temperature and rep.pen. works great. For RP try something like this:

temperature=5
MinP=0.10
rep.pen.=1.15

Adjust to match your needs.

How it was created

I took Sao10K/Euryale-1.3-L2-70B (Good base model) and added

Mikael110/llama-2-70b-guanaco-qlora (Creativity+assistant)
IlyaGusev/saiga2_70b_lora (Creativity+assistant)
s1ghhh/medllama-2-70b-qlora-1.1 (More data)
v2ray/Airoboros-2.1-Jannie-70B-QLoRA (Creativity+assistant)
Chat-Error/fiction.live-Kimiko-V2-70B (Creativity)
Doctor-Shotgun/limarpv3-llama2-70b-qlora (Creativity)
v2ray/LLaMA-2-Wizard-70B-QLoRA (Creativity+assistant)
v2ray/TonyGPT-70B-QLoRA (Special spice)

Then I SLERP-merged it with cognitivecomputations/dolphin-2.2-70b (Needed to bridge the gap between this wonderful mess and Smaxxxer, otherwise it's quality is low) with 0.3t and then SLERP-merged it again with ChuckMcSneed/SMaxxxer-v1-70b (Creativity) with 0.4t. For SLERP-merges I used https://github.com/arcee-ai/mergekit.

Benchmarks (Do they even mean anything anymore?)

NeoEvalPlusN_benchmark

My meme benchmark.

Test name	Gembo
B	2.5
C	1.5
D	3
S	7.5
P	5.25
Total	19.75

Absurdly high. That's what happens when you optimize the merges for a benchmark.

WolframRavenwolf

Benchmark by @wolfram

Artefact2/Gembo-v1-70b-GGUF GGUF Q5_K_M, 4K context, Alpaca format:

✅ Gave correct answers to all 18/18 multiple choice questions! Just the questions, no previous information, gave correct answers: 16/18
✅ Consistently acknowledged all data input with "OK".
➖ Did NOT follow instructions to answer with just a single letter or more than just a single letter.

This shows that this model can be used for real world use cases as an assistant.

Open LLM Leaderboard Evaluation Results

Leaderboard on Huggingface

Model	Average	ARC	HellaSwag	MMLU	TruthfulQA	Winogrande	GSM8K
ChuckMcSneed/Gembo-v1-70b	70.51	71.25	86.98	70.85	63.25	80.51	50.19
ChuckMcSneed/SMaxxxer-v1-70b	72.23	70.65	88.02	70.55	60.7	82.87	60.58

Looks like adding a shitton of RP stuff decreased HellaSwag, WinoGrande and GSM8K, but increased TruthfulQA, MMLU and ARC. Interesting. To be hosnest, I'm a bit surprised that it didn't do that much worse.

Detailed results can be found here

Metric	Value
Avg.	70.51
AI2 Reasoning Challenge (25-Shot)	71.25
HellaSwag (10-Shot)	86.98
MMLU (5-Shot)	70.85
TruthfulQA (0-shot)	63.25
Winogrande (5-shot)	80.51
GSM8k (5-shot)	50.19