Downloads · 30 days
83
24% of all-time downloads
AfkaraLP/CoralGPT-Supra-50M-merge-70
CoralGPT-Supra-50M-merge-70 is a machine learning model from AfkaraLP. Use it for the machine learning task on the model card, and read the license before you ship it in a product. The card lists the license as apache-2.0.
Fine-tuned from SupraLabs/Supra-1.5-50M-Base-exp on Minecraft Bedwars chat from the CoralMC server.
Downloads · 30 days
83
24% of all-time downloads
All-time downloads
342
Public
Parameters
51.8M
367 MB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors207 MB · 56%
From the Hugging Face model README
Fine-tuned from SupraLabs/Supra-1.5-50M-Base-exp on Minecraft Bedwars chat from the CoralMC server.
from transformers import AutoModelForCausalLM, AutoTokenizer
model = AutoModelForCausalLM.from_pretrained("AfkaraLP/CoralGPT-Supra-50M-merge-70")
tokenizer = AutoTokenizer.from_pretrained("AfkaraLP/CoralGPT-Supra-50M-merge-70")
prompt = "[Afkara]: 1 x duo\n[8hi]: io\n[Afkara]: ws?\n[8hi]: si 12\n[Afkara]:"
inputs = tokenizer(prompt, return_tensors="pt")
outputs = model.generate(**inputs, max_new_tokens=80, do_sample=True, temperature=0.7)
print(tokenizer.decode(outputs[0]))
./llama-cli -m CoralGPT-Supra-50M-merge-70.f16.gguf \
-p "[Afkara]: 1 x duo\n[8hi]: io\n[Afkara]: ws?\n[8hi]: si 12\n[Afkara]:" \
-n 80 -t 0.7
Ranked #1 in evaluation (see full summary):
This model was created by merging best-models-1/checkpoint-10000 (full fine-tune) with supra-chat-model-filtered/checkpoint-200 (filtered fine-tune) using linear weight interpolation with alpha=0.7, giving more weight to the filtered checkpoint.
A converted, ready-to-run CoralGPT-Supra-50M-merge-70.litertlm file is included in this repo. It was produced with Google's litert-torch HuggingFace exporter (bundle_litert_lm=True, dynamic weight-only int8 quantization, 3.8x smaller than fp32).
To run on Android:
.litertlm file.</s>).The model is a plain continuation model (no chat template) - feed it raw CoralMC-style chat lines like the widget examples above.