Beta
Explore
Marketplace
Neural Labs
Chat
Wallet
Docs
gerbil-qwen3-coder-30b-bf16 – AI Model by jaimef21 | AlphaNeural AI
You can deploy this model and start earning money today!
jaimef21
/
gerbil-qwen3-coder-30b-bf16
like
0
safetensors
qwen3_moe
gerbil
scheme
lora
code
dpo
en
Qwen/Qwen3-Coder-30B-A3B-Instruct
adapter
apache-2.0
us
Views
No views yet
Model card
Files and Versions
Community
API
Deploy
gerbil-qwen3-coder-30b-bf16
Qwen3-Coder-30B-A3B-Instruct fine-tuned for
Gerbil Scheme
generation.
Training pipeline and tooling:
https://github.com/ober/gerbil-lora
Training pipeline
Three-stage LoRA fine-tune (r=32, α=64, fused-MoE expert targets):
CPT
— Continued pre-training on Gerbil source corpus (lr 2e-5, 2 epochs)
SFT
— Supervised fine-tune on instruction/response pairs (lr 1e-4, 2 epochs)
DPO
— Direct preference optimization on wrong→right pairs (lr 5e-6, 3 epochs)
DPO eval (vs base Qwen3-Coder-30B-A3B-Instruct)
Metric
Base
Trained
Δ
Holdout task score
31
39
+8
Anti-idioms hit
1
0
-1
Code blocks wrapped
9
14
+5
tok_lean_sum (P(chosen) > P(rejected))
-4.17
+4.03
+8.19
wins chosen / rejected (n=66)
47 / 19
52 / 13
+5 / -6
Weights
BF16 merged weights, ~57 GB across 13 safetensors shards.
