MIXED_V7 - Mixed (CRD+SRD) Model (V7)

Dataset

  • Source: mixed_crd_cot_10k_unsloth.jsonl
  • Examples: 10,000
  • Format: messages[] chat format (mixture of structured clinical + general reasoning)

Training Configuration

Parameter Value
Learning Rate 0.00012
LoRA Rank 48
LoRA Alpha 96
LoRA Dropout 0.025
Target Modules All (MLP + Attention)
Epochs 3
Batch Size (effective) 16
Warmup 4%
RSLoRA Enabled

Training Results

  • Training Time: 2.57 hours
  • Final Loss: 0.7364
Downloads last month
51
Safetensors
Model size
8B params
Tensor type
BF16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for kinzakhan1/MIXED_V7

Finetuned
(2197)
this model