JEV / code /reports

561 kB

Ctrl+K

1 contributor

Model card: tested on fresh, real-world data — independent gold-label benchmark vs TypeSafe Jev 1.13 (97% at 2 options, 90% at 16), HN/V2EX examples, 9B vs 27B guidance

1e25cd8 verified about 6 hours ago

review
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
b0_qwen35_9b.md

6.95 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
b0_qwen38_27b.md

5.4 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
data_audit.md

3.01 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
eval_27b_bundle.md

3.77 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
eval_s2_27b.md

8.64 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
eval_s2_9b.md

8.64 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
eval_s2_9b_epoch1.md

8.64 kB
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
eval_v08_bundle.md

3.77 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 19 hours ago
fanout_nopc.json

381 Bytes
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
fanout_pc.json

424 Bytes
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
humaneval_27b_base.json

54.3 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
humaneval_27b_bundle.json

54.3 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
humaneval_base.json

63 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 19 hours ago
humaneval_base_fp32norm.json

63.2 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 19 hours ago
humaneval_v08.json

63.4 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 19 hours ago
humaneval_v08_bundle.json

63.2 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 19 hours ago
humaneval_vllm_27b.json

54.1 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
humaneval_vllm_9b.json

62.8 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
m0_qwen35_9b.md

2.78 kB
Upload JEV v0.7.0 (Qwen3.5-9B distilled from Jev 1.13) 2 days ago
vllm_decisions_27b.json

1.16 kB
Model card: tested on fresh, real-world data — independent gold-label benchmark vs TypeSafe Jev 1.13 (97% at 2 options, 90% at 16), HN/V2EX examples, 9B vs 27B guidance about 6 hours ago
vllm_decisions_9b.json

1.33 kB
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
vllm_openai_9b.json

227 Bytes
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago
vllm_openai_9b_c256.json

131 Bytes
vLLM inference: adapter_vllm/ (decision head as lm_head LoRA), model card section "Inference with vLLM", OpenAI client, measured speed about 6 hours ago