Language Modeling on WikiText-2 stride (test)
3.504bpwMistral-7B-Instruct-v0.3 (QAM-W-3.5)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Mistral-7B-Instruct-v0.3 (QAM-W-3.5)model=Mistral-7B-Instruct-v0.3, config=QAM-W-3.52026.05 | 3.504 | 4.985 | 1.98 | 0.0192 | |
| TinyLlama-1.1B-Chat (QAM-W-3.5-raw)model=TinyLlama-1.1B-Chat, config=QAM-W-3.5-raw2026.05 | 3.506 | 7.6935 | 7.6 | 0.0724 | |
| Qwen2.5-3B-Instruct (QAM-W-3.5-raw)model=Qwen2.5-3B-Instruct, config=QAM-W-3.5-raw2026.05 | 3.506 | 7.7226 | 13.9 | 0.1322 | |
| Qwen2.5-3B-Instruct (QAM-W-3.5)model=Qwen2.5-3B-Instruct, config=QAM-W-3.52026.05 | 3.509 | 7.2887 | 7.5 | 0.0665 | |
| TinyLlama-1.1B-Chat (QAM-W-3.5)model=TinyLlama-1.1B-Chat, config=QAM-W-3.52026.05 | 3.511 | 7.4396 | 4.1 | 0.037 | |
| TinyLlama-1.1B-Chat (BF16)model=TinyLlama-1.1B-Chat, config=BF162026.05 | 16 | 7.1499 | — | — | |
| Qwen2.5-3B-Instruct (BF16)model=Qwen2.5-3B-Instruct, config=BF162026.05 | 16 | 6.7826 | — | — | |
| Mistral-7B-Instruct-v0.3 (BF16)model=Mistral-7B-Instruct-v0.3, config=BF162026.05 | 16 | 4.8883 | — | — |