whatmodel.sh / Llama3-Instruct-8B-WPO-HB-v2
Llama3-Instruct-8B-WPO-HB-v2
↗ share
+ contribute
Llama3-Instruct-8B-WPO-HB-v2 is an open-weights language model from wzhouad (WPO), released July 2024. Its published size is 8.03B parameters.
Catalogue identity notes
Maker repo wzhouad/Llama3-Instruct-8B-WPO-HB-v2 @62b1483: no licence tag on the card. 8,030,261,248 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2024-07: month the maker's Hugging Face repo was created (createdAt 2024-07-24).
$ how it compares
The LLMs ranked either side of Llama3-Instruct-8B-WPO-HB-v2, out of 1132 with a score, then more from wzhouad (WPO).
| # | model | score | where | ||
|---|---|---|---|---|---|
| 399 | Yehia 7B Reasoning (preview) | 56±4 | local | ||
| 400 |  Magistral Medium 1.0 (2506) | 56±1 | comparison only | ||
| 401 | Llama3-Instruct-8B-WPO-HB-v2 | 56±2 | local | ||
| 402 |  Gemma-2-9B-It-SPPO-Iter3 | 56±2 | local | ||
| 403 |  GPT-4 Turbo (2024-04-09) | 55±0 | api $15.00/M | ||
| more from wzhouad (WPO) | |||||
| 193 | gemma-2-9b-it-WPO-HB | 65±2 | local | ||
$ what it's good at
chat & writing▁▁▁▄▆▅▆█▆▆▆▂ #212 of 466 * · 99% of its score · 1 test
Beats Granite 4.2 8B, Qwen3-VL 30B-A3B Thinking, Mistral Small 4 (119B-A6.5B) here, though they rank higher overall. Trails LFM2.5 1.2B Instruct, LFM2.5 1.2B Thinking, Gemma 3 12B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| AlpacaEval 2 length-controlled | 53.4% ✎ | ▁▂▂▃▆▆█▆▄▄▁▁ beats 111/122 | 99% |
$ where to run it
Other ways to run it
- Run it as published · needs ~32.12 GB · from huggingface.co/wzhouad (official)
uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='wzhouad/Llama3-Instruct-8B-WPO-HB-v2', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"
Official weights and documentation: huggingface.co/wzhouad ↗
$ how the score was worked out · AlpacaEval 1.0 and 2.0 leader… 99%
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
| source | results | share of the score |
|---|---|---|
 AlpacaEval 1.0 and 2.0 leaderboards (tatsu-lab/alpaca_eval repository) | 1 | 99% |
| prior · pull toward the middle | 1% | |
| weighted average | = 55.5 |
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json