magnum-32b-v2 - EXL2 4.7bpw rpcal mk2

This is a 4.7bpw EXL2 quant of anthracite-org/magnum-32b-v2

This quant was made using exllamav2-0.1.8 with Bluemoon-light dataset for RP.

I tested this quant shortly in some random RPs (including ones over 8k context) and it seems to work fine.

Prompt Templates

Uses ChatML format like mentioned below.

Original readme below

This is the third in a series of models designed to replicate the prose quality of the Claude 3 models, specifically Sonnet and Opus. This model is fine-tuned on top of Qwen1.5 32B.

Prompting

Model has been Instruct tuned with the ChatML formatting. A typical input would look like this:

"""<|im_start|>user
Hi there!<|im_end|>
<|im_start|>assistant
Nice to meet you!<|im_end|>
<|im_start|>user
Can I ask a question?<|im_end|>
<|im_start|>assistant
"""

Credits

Stheno dataset (filtered)
NobodyExistsOnTheInternet/claude_3.5s_single_turn_unslop_filtered
NobodyExistsOnTheInternet/PhiloGlanSharegpt
NobodyExistsOnTheInternet/Magpie-Reasoning-Medium-Subset
kalomaze/Opus_Instruct_25k
Nopm/Opus_WritingStruct
Gryphe/Sonnet3.5-SlimOrcaDedupCleaned (A ~16k rows subset)

This model has been a team effort, and the credits goes to all members of Anthracite.

Training

The training was done for 2 epochs. We used 8x NVIDIA H100 Tensor Core GPUs for the full-parameter fine-tuning of the model.

Safety

...