4bit Autotrain compatible Base model:quantized:unsloth/m... Base model:unsloth/mistral-7b-... Conversational Dataset:arcee-ai/evolkit-20k Dataset:cognitivecomputations/... Dataset:kingnish/reasoning-bas... En Endpoints compatible Gguf Instruct Lora Mistral Model-index Quantized Region:us Safetensors Sft Sharded Tensorflow Theprint Trl Unsloth
Model Card on HF 🤗: https://huggingface.co/theprint/ReWiz-7B
ReWiz 7B Benchmarks
MMLU Pro: 18.56
GPQA: 3.36
MUSR: 16.74
BBH: 23.5
IFEval: 40.48 vs 88 (so35)-54%
MATH Lvl 5: 2.95
LLME Score: 0.30229
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)
ReWiz 7B Parameters and Internals
| LLM Name | ReWiz 7B |
| Repository 🤗 | https://huggingface.co/theprint/ReWiz-7B |
| Base Model(s) | |
| Model Size | 7b |
| Required VRAM | 14.5 GB |
| Updated | 2024-12-21 |
| Maintainer | theprint |
| Instruction-Based | Yes |
| Model Files | |
| Supported Languages | en |
| GGUF Quantization | Yes |
| Quantization Type | 4bit|gguf |
| Model Architecture | AutoModelForCausalLM |
| License | apache-2.0 |
| Model Max Length | 32768 |
| Is Biased | none |
| Tokenizer Class | LlamaTokenizer |
| Padding Token | [control_768] |
| PEFT Type | LORA |
| LoRA Model | Yes |
| PEFT Target Modules | o_proj|k_proj|v_proj|down_proj|q_proj|gate_proj|up_proj |
| LoRA Alpha | 16 |
| LoRA Dropout | 0 |
| R Param | 8 |
Best Alternatives to ReWiz 7B
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| ...wen Math 7b 24 1 100 1 Nonmath | 0K / 15.2 GB | 306 | 0 |
| ShoriRP V0.75d | 0K / 0.2 GB | 49 | 32 |
| Mistral 7B Instruct 4bit | 0K / 0.2 GB | 18 | 1 |
| Mistral 7B Instruct V0.3 | 32K / 14.5 GB | 3431010 | 1196 |
| Mistral 7B Instruct V0.3 | 32K / 14.5 GB | 7432 | 5 |
| Mistral 7B Instruct V0.3 | 32K / 14.5 GB | 3243 | 3 |
| ...ralai Mistral 7B Instruct V0.3 | 32K / 14.5 GB | 58 | 1 |
| Mistral 7B Instruct V0.3 | 32K / 14.5 GB | 7 | 0 |
| Full V4 Astromistral Final | 32K / 4.5 GB | 15 | 1 |
| Mistral 7B Instruct Uz | 0K / 14.5 GB | 361 | 9 |
Note: green Score (e.g. "73.2") means that the model is better than theprint/ReWiz-7B .
Rank the ReWiz 7B 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? 40066 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241217