Quantization made by Richard Erkhov.

L3-Nymeria-8B - GGUF

Model creator: https://huggingface.co/tannedbum/
Original model: https://huggingface.co/tannedbum/L3-Nymeria-8B/

Name	Quant method	Size
L3-Nymeria-8B.Q2_K.gguf	Q2_K	2.96GB
L3-Nymeria-8B.IQ3_XS.gguf	IQ3_XS	3.28GB
L3-Nymeria-8B.IQ3_S.gguf	IQ3_S	3.43GB
L3-Nymeria-8B.Q3_K_S.gguf	Q3_K_S	3.41GB
L3-Nymeria-8B.IQ3_M.gguf	IQ3_M	3.52GB
L3-Nymeria-8B.Q3_K.gguf	Q3_K	3.74GB
L3-Nymeria-8B.Q3_K_M.gguf	Q3_K_M	3.74GB
L3-Nymeria-8B.Q3_K_L.gguf	Q3_K_L	4.03GB
L3-Nymeria-8B.IQ4_XS.gguf	IQ4_XS	4.18GB
L3-Nymeria-8B.Q4_0.gguf	Q4_0	4.34GB
L3-Nymeria-8B.IQ4_NL.gguf	IQ4_NL	4.38GB
L3-Nymeria-8B.Q4_K_S.gguf	Q4_K_S	4.37GB
L3-Nymeria-8B.Q4_K.gguf	Q4_K	4.58GB
L3-Nymeria-8B.Q4_K_M.gguf	Q4_K_M	4.58GB
L3-Nymeria-8B.Q4_1.gguf	Q4_1	4.78GB
L3-Nymeria-8B.Q5_0.gguf	Q5_0	5.21GB
L3-Nymeria-8B.Q5_K_S.gguf	Q5_K_S	5.21GB
L3-Nymeria-8B.Q5_K.gguf	Q5_K	5.34GB
L3-Nymeria-8B.Q5_K_M.gguf	Q5_K_M	5.34GB
L3-Nymeria-8B.Q5_1.gguf	Q5_1	5.65GB
L3-Nymeria-8B.Q6_K.gguf	Q6_K	6.14GB
L3-Nymeria-8B.Q8_0.gguf	Q8_0	7.95GB

Original model description:

base_model: - princeton-nlp/Llama-3-Instruct-8B-SimPO - Sao10K/L3-8B-Stheno-v3.2 library_name: transformers tags: - mergekit - merge - roleplay - sillytavern - llama3 - not-for-all-audiences license: cc-by-nc-4.0 language: - en

The smartest L3 8B model combined with high-end RP model. What could go wrong.

The idea was to fuse a bit of SimPO's realism with Stheno. It took a few days to come up with a balanced slerp configuration, but I'm more than satisfied with the end result.

SillyTavern

Text Completion presets

temp 0.9
top_k 30
top_p 0.75
min_p 0.2
rep_pen 1.1
smooth_factor 0.25
smooth_curve 1

Advanced Formatting

Context & Instruct preset by Virt-io

Instruct Mode: Enabled

merge

This is a merge of pre-trained language models created using mergekit.

This model was merged using the slerp merge method.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:


slices:
  - sources:
      - model: Sao10K/L3-8B-Stheno-v3.2
        layer_range: [0, 32]
      - model: princeton-nlp/Llama-3-Instruct-8B-SimPO
        layer_range: [0, 32]
merge_method: slerp
base_model: Sao10K/L3-8B-Stheno-v3.2
parameters:
  t:
    - filter: self_attn
      value: [0.4, 0.5, 0.6, 0.4, 0.6]
    - filter: mlp
      value: [0.6, 0.5, 0.4, 0.6, 0.4]
    - value: 0.5
dtype: bfloat16

Original model information:

Model: Sao10K/L3-8B-Stheno-v3.2

Stheno-v3.2-Zeta

Changes compared to v3.1
- Included a mix of SFW and NSFW Storywriting Data, thanks to Gryphe
- Included More Instruct / Assistant-Style Data
- Further cleaned up Roleplaying Samples from c2 Logs -> A few terrible, really bad samples escaped heavy filtering. Manual pass fixed it.
- Hyperparameter tinkering for training, resulting in lower loss levels.

Testing Notes - Compared to v3.1
- Handles SFW / NSFW seperately better. Not as overly excessive with NSFW now. Kinda balanced.
- Better at Storywriting / Narration.
- Better at Assistant-type Tasks.
- Better Multi-Turn Coherency -> Reduced Issues?
- Slightly less creative? A worthy tradeoff. Still creative.
- Better prompt / instruction adherence.

Want to support my work ? My Ko-fi page: https://ko-fi.com/tannedbum