KarlLearnsAI Claude Sonnet 4.6 commited on
Commit
434c6b1
·
1 Parent(s): faad7f2

Add train.sh startup script and assets folder

Browse files

- train.sh: startup script for HF Spaces/Colab/Northflank that checks HF_TOKEN, installs train deps, and runs GRPO training
- assets/: placeholder folder for architecture diagram (architecture.jpg/png)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Files changed (2) hide show
  1. assets/README.md +1 -0
  2. train.sh +33 -0
assets/README.md ADDED
@@ -0,0 +1 @@
 
 
1
+ # Drop architecture.jpg or architecture.png here
train.sh ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/bin/bash
2
+ # ============================================================
3
+ # Training startup script for HF Spaces / Colab / Northflank
4
+ #
5
+ # Usage:
6
+ # ./train.sh # full run from config.yaml
7
+ # ./train.sh --steps 5 --episodes 3 # quick smoke test
8
+ # HF_TOKEN=hf_xxx ./train.sh # with inline token
9
+ # ============================================================
10
+
11
+ set -e
12
+
13
+ echo "============================================================"
14
+ echo " Nested RL Envs — GRPO Training"
15
+ echo " Team: Ludes Magnus"
16
+ echo "============================================================"
17
+
18
+ # Check HF_TOKEN
19
+ if [ -z "$HF_TOKEN" ]; then
20
+ echo "ERROR: HF_TOKEN environment variable is not set."
21
+ echo "Set it via: export HF_TOKEN=hf_xxx"
22
+ exit 1
23
+ fi
24
+
25
+ # Install training dependencies if not already installed
26
+ if ! python -c "import unsloth" 2>/dev/null; then
27
+ echo "Installing training dependencies..."
28
+ pip install -q -e ".[train]"
29
+ fi
30
+
31
+ # Run training
32
+ echo "Starting GRPO training..."
33
+ python -m layer1.train "$@"