JEV / code /src /jev_judge

89.2 kB

Ctrl+K

1 contributor

vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed

b16c3a6 verified about 1 hour ago

__init__.py

108 Bytes
v0.8.0: continued training to 0.93 epoch with LR annealing — t30k KL 0.0276→0.0210, choice top-1 0.884→0.898, score MAE 0.119→0.103, ECE 0.0014→0.0007 2 days ago
calibration.py

4.22 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
checkpointing.py

2.73 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
data.py

16 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 1 hour ago
head_init.py

2.95 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
infer.py

2.74 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
losses.py

2.09 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
metrics.py

6.46 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
model.py

15.4 kB
v0.8.0 dual-head bundle: pristine Qwen3.5-9B backbone + lm_head (bit-identical, HumanEval 70.7% = base) + unmerged LoRA decision adapter + 24-slot head; card: one backbone two heads about 14 hours ago
server.py

10.4 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
template.py

4.38 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
train_loop.py

16.4 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 1 hour ago
vllm_engine.py

5.33 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 1 hour ago