Update .env.spark.example: TORCH_DTYPE now auto-detected

TORCH_DTYPE is automatically set from model config:
- codegen-350m → fp16
- devstral-small → bf16

Only uncomment to override the default.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <[email protected]>

.env.spark.example

@@ -17,8 +17,10 @@ HF_TOKEN=<your-hf-token>
 # Model Settings
 MAX_CONTEXT=8192
 BATCH_SIZE=1
-TORCH_DTYPE=fp16
-# TORCH_DTYPE=bf16               # Use bf16 for Devstral (Phase 3)
+# TORCH_DTYPE is auto-detected from model config (codegen→fp16, devstral→bf16)
+# Uncomment to override:
+# TORCH_DTYPE=fp16
+# TORCH_DTYPE=bf16
 
 # Device Override (set to 'cpu' if GPU not supported yet)
 # DEVICE=cuda                    # Default: auto-detect