internlm2-chat-7B-sft: benchmark results
InternLM team's internlm2-chat-7b-sft supervised fine-tuning checkpoint. Its card distinguishes this SFT release from the subsequently RLHF-aligned chat model, preserving the training-stage distinction. Provider: Shanghai AI Lab. Access: Open.
Aggregate Rating 1408 ± 40, rank #5717 of 10916 rated models, from 33 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Open PL LLM - KLEJ NER (generative, 5-shot) | 53.74 | Accuracy (%) | 74.8 |
| Open Portuguese LLM - FaQuAD NLI | 73.13 | Macro F1 (%) | 67.9 |
| Open PL LLM - DYK (generative, 5-shot) | 61.57 | Binary F1 (%) | 66 |
| Open PL LLM - PPC (multiple choice, 5-shot) | 74.5 | Accuracy (%) | 65 |
| Open PL LLM - PPC (generative, 5-shot) | 72.8 | Accuracy (%) | 64.9 |
| Open PL LLM - PolEmo2-OUT (multiple choice, 5-shot) | 71.26 | Accuracy (%) | 62.5 |
| Open PL LLM - KLEJ NER (multiple choice, 5-shot) | 49.42 | Accuracy (%) | 61.6 |
| Open PL LLM - PolQA Reranking (multiple choice, 5-shot) | 73.22 | Accuracy (%) | 61.6 |
| Open PL LLM - PolEmo2-OUT (generative, 5-shot) | 67.81 | Accuracy (%) | 56.5 |
| Open PL LLM - PSC (generative, 5-shot) | 90.58 | Binary F1 (%) | 56.2 |
| Open Portuguese LLM - ASSIN2 RTE | 90.21 | Macro F1 (%) | 55 |
| Open PL LLM - Generative | 49.55 | Average Generative Score (%) | 52.8 |
Interactive version: theaggregate.ai/model?slug=internlm2-chat-7b-sft · How It Works · Data refreshed daily, snapshot 2026-11-08.