ELYZA Japanese Llama 2 7B Fast Instruct GPTQ is an open-source language model by dahara1. Features: 7b LLM, VRAM: 4.1GB, Context: 4K, License: llama2, Quantized, Instruction-Based, HF Score: 49.8, LLM Explorer Score: 0.14, Arc: 53.2, HellaSwag: 78.3, MMLU: 47.1, TruthfulQA: 39.1, WinoGrande: 73.2, GSM8K: 7.9.
Arxiv:2307.09288 4bit Autotrain compatible En Gptq Instruct Ja Llama Quantized Region:us Safetensors
Model Card on HF 🤗: https://huggingface.co/dahara1/ELYZA-japanese-Llama-2-7b-fast-instruct-GPTQ
ELYZA Japanese Llama 2 7B Fast Instruct GPTQ Benchmarks
ARC: 53.16 vs 96.7 (so35)-45%
HellaSwag: 78.25 vs 95.3 (gpt4)-17.9%
MMLU: 47.07 vs 88.3 (so35)-46.7%
TruthfulQA: 39.08 vs 59 (gpt4)-33.8%
WinoGrande: 73.24 vs 87.5 (gpt4)-16.3%
GSM8K: 7.88 vs 96.4 (so35)-91.8%
LLME Score: 0.14367
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)
ELYZA Japanese Llama 2 7B Fast Instruct GPTQ Parameters and Internals
| Model Type |
| |||
| Additional Notes |
| |||
| Supported Languages |
| |||
| Training Details |
| |||
| Input Output |
|
| LLM Name | ELYZA Japanese Llama 2 7B Fast Instruct GPTQ |
| Repository 🤗 | https://huggingface.co/dahara1/ELYZA-japanese-Llama-2-7b-fast-instruct-GPTQ |
| Base Model(s) | |
| Model Size | 7b |
| Required VRAM | 4.1 GB |
| Updated | 2025-09-23 |
| Maintainer | dahara1 |
| Model Type | llama |
| Instruction-Based | Yes |
| Model Files | |
| Supported Languages | ja en |
| GPTQ Quantization | Yes |
| Quantization Type | gptq|4bit |
| Model Architecture | LlamaForCausalLM |
| License | llama2 |
| Context Length | 4096 |
| Model Max Length | 4096 |
| Transformers Version | 4.35.0 |
| Tokenizer Class | LlamaTokenizer |
| Padding Token | </s> |
| Vocabulary Size | 45043 |
| Torch Data Type | float16 |
Best Alternatives to ELYZA Japanese Llama 2 7B Fast Instruct GPTQ
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| ... 7B 32K Instructions V4 Marlin | 32K / 4.1 GB | 8 | 0 |
| Llama 2 7B 32K Instruct GPTQ | 32K / 3.9 GB | 11 | 27 |
| CodeLlama 7B Instruct GPTQ | 16K / 3.9 GB | 1333 | 46 |
| ...a 7B Instruct GPTQ Calib Ja 1K | 16K / 3.9 GB | 6 | 0 |
| ...echless Tora Code 7B V1.0 GPTQ | 16K / 3.9 GB | 2 | 1 |
| CausalLM 7B GPTQ | 8K / 5.9 GB | 10 | 5 |
| Leo Hessianai 7B Chat GPTQ | 8K / 3.9 GB | 23 | 0 |
| ...ssianai 7B Chat Bilingual GPTQ | 8K / 3.9 GB | 12 | 0 |
| Xgen 7B 8K Open Instruct Gptq | 8K / 4.2 GB | 9 | 4 |
| Deepseek Math 7B Instruct GPTQ | 4K / 4.8 GB | 5 | 0 |
Note: green Score (e.g. "73.2") means that the model is better than dahara1/ELYZA-japanese-Llama-2-7b-fast-instruct-GPTQ .
Rank the ELYZA Japanese Llama 2 7B Fast Instruct GPTQ 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? 51638 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20241124