Model Card on HF 🤗: https://huggingface.co/sumink/llamaft
Llamaft Benchmarks
MMLU Pro: 12.38
GPQA: 2.8
MUSR: 3.06
BBH: 12.95
IFEval: 16.09 vs 88 (so35)-81.7%
MATH Lvl 5: 1.66
LLME Score: 0.23034
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
What is the LLM Explorer Rank (Score)
Llamaft Parameters and Internals
| LLM Name | Llamaft |
| Repository 🤗 | https://huggingface.co/sumink/llamaft |
| Model Size | 3b |
| Required VRAM | 6.5 GB |
| Updated | 2025-09-01 |
| Maintainer | sumink |
| Model Type | llama |
| Model Files | |
| Model Architecture | LlamaForCausalLM |
| Context Length | 131072 |
| Model Max Length | 131072 |
| Transformers Version | 4.47.1 |
| Tokenizer Class | PreTrainedTokenizerFast |
| Padding Token | <|end_of_text|> |
| Vocabulary Size | 128256 |
| Torch Data Type | bfloat16 |
Best Alternatives to Llamaft
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| ISA 03 Mini 3B Hybrid Preview | 256K / 6.5 GB | 1138 | 4 |
| Llama 3.2 3B Instruct | 128K / 6.5 GB | 1749257 | 1686 |
| Llama 3.2 3B | 128K / 6.5 GB | 659224 | 626 |
| DeepSeek R1 Distill Llama 3B | 128K / 6.5 GB | 1411 | 16 |
| Hermes 3 Llama 3.2 3B | 128K / 6.5 GB | 7958 | 171 |
| Discord Micae Hermes 3 3B | 128K / 6.5 GB | 1261 | 7 |
| FuseChat Llama 3.2 3B Instruct | 128K / 6.5 GB | 839 | 7 |
| Llama 3.2 3B RP Toxic Fuse | 128K / 6.4 GB | 9 | 2 |
| Orpheus 3B 0.1 Ft | 128K / 6.6 GB | 33552 | 7 |
| Calme 3.1 Llamaloi 3B | 128K / 10.6 GB | 2440 | 1 |
Note: green Score (e.g. "73.2") means that the model is better than sumink/llamaft .
Rank the Llamaft Capabilities
🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟
What open-source LLMs or SLMs are you in search of? 51191 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241124