Model Card on HF 🤗: https://huggingface.co/KoboldAI/fairseq-dense-6.7B
Fairseq Dense 6.7B Benchmarks
ARC: 39.42 vs 96.7 (so35)-59.2%
HellaSwag: 71.26 vs 95.3 (gpt4)-25.2%
MMLU: 26.91 vs 88.3 (so35)-69.5%
TruthfulQA: 32.73 vs 59 (gpt4)-44.5%
WinoGrande: 65.27 vs 87.5 (gpt4)-25.4%
LLME Score: 0.04538
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)
Fairseq Dense 6.7B Parameters and Internals
| Model Type |
| |
| Additional Notes |
|
| LLM Name | Fairseq Dense 6.7B |
| Repository 🤗 | https://huggingface.co/KoboldAI/fairseq-dense-6.7B |
| Model Size | 6.7b |
| Required VRAM | 13.7 GB |
| Updated | 2025-08-19 |
| Maintainer | KoboldAI |
| Model Type | xglm |
| Model Files | |
| Supported Languages | en |
| Model Architecture | XGLMForCausalLM |
| Context Length | 2048 |
| Model Max Length | 2048 |
| Transformers Version | 4.17.0 |
| Tokenizer Class | GPT2Tokenizer |
| Vocabulary Size | 50261 |
| Activation Function | gelu |
Best Alternatives to Fairseq Dense 6.7B
Best Alternatives |
Context / RAM |
Downloads |
Likes |
|---|---|---|---|
| Fairseq Dense 6.7B Shinen | 2K / 13.3 GB | 172 | 1 |
| Fairseq Dense 6.7B Janeway | 2K / 13.3 GB | 27 | 4 |
Rank the Fairseq Dense 6.7B 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? 50751 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Release v20241124