Reflection Model Benchmarks
MMLU Pro: 25.68
GPQA: 6.71
MUSR: 5.75
BBH: 25.21
IFEval: 15.99 vs 88 (so35)-81.8%
MATH Lvl 5: 4.31
LLME Score: 0.29568
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)
Reflection Model Parameters and Internals
| LLM Name | Reflection Model |
| Repository 🤗 | https://huggingface.co/olabs-ai/reflection_model |
| Required VRAM | 0.2 GB |
| Updated | 2024-09-18 |
| Maintainer | olabs-ai |
| Model Files | |
| Model Architecture | Adapter |
| License | apache-2.0 |
| Model Max Length | 131072 |
| Is Biased | none |
| Tokenizer Class | PreTrainedTokenizerFast |
| Padding Token | <|finetune_right_pad_id|> |
| PEFT Type | LORA |
| LoRA Model | Yes |
| PEFT Target Modules | q_proj|k_proj|gate_proj|v_proj|down_proj|o_proj|up_proj |
| LoRA Alpha | 16 |
| LoRA Dropout | 0 |
| R Param | 16 |
Best Alternatives to Reflection Model
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| Tot Llama Update | 0K / 0 GB | 2308 | 0 |
| Ll3 C3 Lora New | 0K / 0 GB | 7 | 0 |
| Zephyr Phi 1 5 Sft Qlora | 0K / 0 GB | 5 | 0 |
| Phi Openllm Lb Test | 0K / 0 GB | 5 | 0 |
| Hua V0.1 | 0K / 0 GB | 6 | 0 |
| CodeHAWK | 0K / 0 GB | 8 | 0 |
| ....001 0.5ultrafeedback LowLr I1 | 0K / 0 GB | 7 | 0 |
| ...ew2 B0.001 0.5ultrafeedback I1 | 0K / 0 GB | 5 | 0 |
| Phi 2 Gpo Renew2 B0.001 V4 I1 | 0K / 0.2 GB | 5 | 0 |
| Absa 10 Domains Large | 0K / 0.1 GB | 6 | 0 |
Rank the Reflection Model 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? 36026 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Release v2024072803