Beta
Explore
Marketplace
Neural Labs
Playground
Wallet
Docs
ATIR – AI Model by Tung111 | AlphaNeural AI
You can deploy this model and start earning money today!
Tung111
/
ATIR
like
0
safetensors
us
Views
No views yet
Model card
Files and Versions
Community
API
Deploy
Fused LoRA (text hard negatives + CoQA hard negatives)
source_1: /data/zt/model/atir-model/stage1_model_lora_hard_text_1123/checkpoint-800
source_2: /data/zt/model/atir-model/stage1_model_lora_coqa_hard_negatives_1204_chunk3/best_lora_step_38
weights: w1=0.5, w2=0.5
strategy: concatenate A/B with sqrt(weights) -> r=64, alpha=64 so delta = w1
delta1 + w2
delta2
base: /data/zt/model/qwen2.5-omni-3b
