apetersson commited on
Commit
ebbae5e
·
verified ·
1 Parent(s): 34c68e1

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +0 -6
README.md CHANGED
@@ -55,8 +55,6 @@ This run was trained on the CINECA Leonardo supercomputer.
55
  - **Scratch path:** `/leonardo_scratch/large/usertrain/a08trc02`
56
  - **Partition/reservation:** `boost_usr_prod`, `s_tra_ncc`
57
  - **Hardware:** 1 node with 4x NVIDIA A100-SXM-64GB
58
- - **Launcher:** `finetune/python/slurm/leonardo/train-diag-v2.slurm`
59
- - **Training entry point:** `finetune/python/training/sft_qwen36.py`
60
  - **Environment:** Python 3.11.7 virtualenv, torch 2.6-compatible stack,
61
  `transformers==5.9.0` Qwen3.6 text-only loading
62
 
@@ -79,10 +77,6 @@ environment, and dependencies were staged to `$SCRATCH` before training.
79
  The SFT samples were built from the ModuleWarden.com finetune-data corpus. The
80
  source workspace for this run was:
81
 
82
- ```text
83
- /Users/andreas/Nextcloud/Z1Hackathon/ZeroToOne_Data/finetune-data
84
- ```
85
-
86
  The case-sourcing pipeline starts from public software-security feeds and then
87
  normalizes them into package-version audit cases:
88
 
 
55
  - **Scratch path:** `/leonardo_scratch/large/usertrain/a08trc02`
56
  - **Partition/reservation:** `boost_usr_prod`, `s_tra_ncc`
57
  - **Hardware:** 1 node with 4x NVIDIA A100-SXM-64GB
 
 
58
  - **Environment:** Python 3.11.7 virtualenv, torch 2.6-compatible stack,
59
  `transformers==5.9.0` Qwen3.6 text-only loading
60
 
 
77
  The SFT samples were built from the ModuleWarden.com finetune-data corpus. The
78
  source workspace for this run was:
79
 
 
 
 
 
80
  The case-sourcing pipeline starts from public software-security feeds and then
81
  normalizes them into package-version audit cases:
82