Instructions to use bespokelabs/Bespoke-Nimble-9B-v2 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use bespokelabs/Bespoke-Nimble-9B-v2 with PEFT:
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3.5-9B") model = PeftModel.from_pretrained(base_model, "bespokelabs/Bespoke-Nimble-9B-v2") - Transformers
How to use bespokelabs/Bespoke-Nimble-9B-v2 with Transformers:
# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("bespokelabs/Bespoke-Nimble-9B-v2", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download evaluation_summary.json from bespokelabs/Bespoke-Nimble-9B-v2: direct link, hf CLI and curl.
- Browser
- Download file 1.53 kB
-
https://huggingface.co/bespokelabs/Bespoke-Nimble-9B-v2/resolve/main/evaluation_summary.json
- Command line
-
hf download hf://bespokelabs/Bespoke-Nimble-9B-v2/evaluation_summary.json
-
curl -L -o evaluation_summary.json https://huggingface.co/bespokelabs/Bespoke-Nimble-9B-v2/resolve/main/evaluation_summary.json
1.53 kB
| { | |
| "bespoke_eval": { | |
| "n": 324, | |
| "accuracy": 0.8271604938271605, | |
| "nll": 0.7308274886209167, | |
| "brier_to_reference": 0.31239001056754634, | |
| "hard_label_ece10": 0.15284857863329301, | |
| "hard_label_n": 324, | |
| "soft_target_n": 0, | |
| "soft_target_tvd": null, | |
| "hard_label_nll": 0.7308274886209167, | |
| "hard_label_brier": 0.31239001056754634, | |
| "mean_confidence": 0.9726037056817401 | |
| }, | |
| "synthetic_jevbench_proxy": { | |
| "n": 534, | |
| "accuracy": 0.8258426966292135, | |
| "axes": { | |
| "intelligence": 78.5377656826666, | |
| "calibration": 80.4126946325609, | |
| "speed": 88.53288735436024, | |
| "cost": 42.600857357940114 | |
| }, | |
| "composite": 69.86047453614847, | |
| "invalid": 0, | |
| "latency": { | |
| "n": 242, | |
| "p50_seconds": 0.1018778674999794, | |
| "p95_seconds": 0.12314256449994894, | |
| "endpoint_kind": "gpu" | |
| }, | |
| "estimated_usd_per_1000": 0.08190393258426966, | |
| "environment": "Modal H100 SXM80GB; GPU-resident HTTP loopback client and native scorer; serial concurrency1; no internet transit; no KV/prefix cache; no quantization; no generative decoding; one excluded synthetic warmup", | |
| "reference_prices_per_million": [ | |
| 0.1, | |
| 0.15 | |
| ], | |
| "task_sha256": "1ed70789ec11f468e78e1da4151ebe69d509a46c11f4e69c2c27b1551bce4e78", | |
| "input_sha256": "af2dab21b44fae8ea791678f3a96fff1f4a3cd5dbc1ee0814fac21bc2331dbe7" | |
| }, | |
| "temperature": 1.0, | |
| "candidate_orders": 1, | |
| "official_jevbench": false, | |
| "probability_source": "native allowed-token logits and softmax" | |
| } | |