IceMedovukhaRP-7b

IceMedovukhaRP-7b (Ice0.101-20.03-RP)

mistral v0.2 base

The words mead and medovukha are closely related and go back to the Proto-Indo-European word médʰu (honey). Produced in Eastern Europe since pagan times, it remained popular well into the 19th century (unlike in Western Europe, where by the Middle Ages mead had already been mostly replaced by wine and beer).

[!IMPORTANT] ST settings, rules-lorebook look here (work in proggres.. look in discord for test variants or try from previous models or edit IceLemonMedovukhaRP-7b to have without reasoning)

[!NOTE] IceLemonMedovukhaRP-7b Same model but with light GRPO training to use <reasoning>

[!TIP] Get last version of rules and ST settings preset, or ask me a questions you can here. on my AI related discord server for feedback, questions and other stuff.

[!NOTE] In general Alpaca format will work, but I will recommend to try given ST settings preset and rules-lorebook.

ko-fi To buy sweets for my cat :3

The model has a context limit of 32k tokens. However, the quality of responses from any model begins to decline after 16k tokens, with a more rapid degradation beyond 21k tokens. I recommend using 21k tokens as the maximum for optimal performance.

Exl2 Quants

[!WARNING]

• 4.2bpw-exl2
• 6.5bpw-exl2
• 8bpw-exl2

Thx mradermacher for GGUF

[!WARNING]

• GGUF
• i1-GGUF

Download

I recommend using the huggingface-hub Python library:

[!TIP]

pip3 install huggingface-hub

To download the main branch to a folder called IceMedovukhaRP-7b:

[!TIP]

mkdir IceMedovukhaRP-7b
huggingface-cli download icefog72/IceMedovukhaRP-7b --local-dir IceMedovukhaRP-7b --local-dir-use-symlinks False

<details> <summary>More advanced huggingface-cli download usage</summary>

If you remove the --local-dir-use-symlinks False parameter, the files will instead be stored in the central Hugging Face cache directory (default location on Linux is: ~/.cache/huggingface), and symlinks will be added to the specified --local-dir, pointing to their real location in the cache. This allows for interrupted downloads to be resumed, and allows you to quickly clone the repo to multiple places on disk without triggering a download again. The downside, and the reason why I don't list that as the default option, is that the files are then hidden away in a cache folder and it's harder to know where your disk space is being used, and to clear it up if/when you want to remove a download model.

The cache location can be changed with the HF_HOME environment variable, and/or the --cache-dir parameter to huggingface-cli.

For more documentation on downloading with huggingface-cli, please see: HF -> Hub Python Library -> Download files -> Download from the CLI.

To accelerate downloads on fast connections (1Gbit/s or higher), install hf_transfer:

pip3 install hf_transfer

And set environment variable HF_HUB_ENABLE_HF_TRANSFER to 1:

mkdir FOLDERNAME
HF_HUB_ENABLE_HF_TRANSFER=1 huggingface-cli download MODEL --local-dir FOLDERNAME --local-dir-use-symlinks False

Windows Command Line users: You can set the environment variable by running set HF_HUB_ENABLE_HF_TRANSFER=1 before the download command.

</details>

Models Merged

The following models were included in the merge:

• E:\FModels\Ice0.100-20.03-RP
• E:\FModels\Ice0.99-20.03-RP

Configuration

The following YAML configuration was used to produce this model:

slices:
  - sources:
      - model: E:\FModels\Ice0.100-20.03-RP
        layer_range: [0, 32]
      - model: E:\FModels\Ice0.99-20.03-RP
        layer_range: [0, 32]

merge_method: slerp
base_model: E:\FModels\Ice0.100-20.03-RP
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