Model Card on HF 🤗: https://huggingface.co/aboros98/merlin1.5
Merlin1.5 Benchmarks
ARC: 59.56 vs 96.7 (so35)-38.4%
HellaSwag: 74.63 vs 95.3 (gpt4)-21.7%
MMLU: 56.59 vs 88.3 (so35)-35.9%
TruthfulQA: 48.03 vs 59 (gpt4)-18.6%
WinoGrande: 74.66 vs 87.5 (gpt4)-14.7%
GSM8K: 50.27 vs 96.4 (so35)-47.9%
LLME Score: 0.20402
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)
Merlin1.5 Parameters and Internals
| LLM Name | Merlin1.5 |
| Repository 🤗 | https://huggingface.co/aboros98/merlin1.5 |
| Required VRAM | 5.6 GB |
| Updated | 2024-12-22 |
| Maintainer | aboros98 |
| Model Type | phi |
| Model Files | |
| Model Architecture | PhiForCausalLM |
| License | mit |
| Context Length | 2048 |
| Model Max Length | 2048 |
| Transformers Version | 4.37.0 |
| Tokenizer Class | CodeGenTokenizer |
| Padding Token | <|pad|> |
| Vocabulary Size | 51200 |
| Torch Data Type | float16 |
Best Alternatives to Merlin1.5
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| Phi1.5 Quantized 2 | 2K / GB | 9 | 0 |
| Merlin1.4 | 2K / 5.6 GB | 68 | 0 |
| Merlin1.2 | 2K / 5.6 GB | 72 | 0 |
| Merlin1.3 | 2K / 5.6 GB | 72 | 0 |
| Tinyllama Quant | 2K / 0.8 GB | 7 | 1 |
| Phi1.5 Update 4 | 2K / GB | 7 | 0 |
| Phi 2 Fp8 | 2K / 3.3 GB | 8 | 0 |
| Phi 2 Int8 Ov | 2K / 0 GB | 51 | 0 |
| Phi 2 Ov Quantized | 2K / 1.9 GB | 27 | 0 |
| Phi Bode 2 Ultraalpaca | 2K / 5.6 GB | 17 | 2 |
Note: green Score (e.g. "73.2") means that the model is better than aboros98/merlin1.5 .
Rank the Merlin1.5 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? 40123 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241217