Upload subspace_ar_bsf071226_v11nl_1x_comp/run_config.yaml with huggingface_hub

subspace_ar_bsf071226_v11nl_1x_comp/run_config.yaml

@@ -0,0 +1,65 @@
+ar_cosmag: false
+ar_cosmag_lambda: 0.3
+ar_num_layers: 25
+ar_raw_target: false
+ar_relmse: false
+attn_implementation: sdpa
+av_ortho_penalty: 0.0
+base_ckpt: Qwen/Qwen3-8B
+batch_size: 12
+bullet_min_chars: 1
+comp_mean_bias_npy: /workspace/asherps/nla_analysis/ckpt_snap/free8_minloo075_s900/comp_mean_bias.npy
+compositional: true
+device_map: single
+eval_parquet: null
+eval_skip_rows: 35000
+fixed_bullets: false
+freeze_backbone: false
+full_ft_dtype: fp32
+gradient_accumulation_steps: 1
+gradient_checkpointing: true
+heldout_every: 100
+heldout_parquet: null
+heldout_rows: 1000
+hf_upload: true
+hf_upload_min_steps: 1000
+hf_upload_private: false
+hf_upload_repo: null
+judge_concurrency: 32
+judge_key_env: ANTHROPIC_API_KEY_FALLBACK
+lora_alpha: 16
+lora_r: 128
+lr: 2.0e-05
+lr_warmup_steps: 50
+max_gpu_mem: 0
+max_grad_norm: 1.0
+max_len: 1024
+max_rows: null
+min_lr: 2.0e-06
+mode: ar
+n_samples: 4
+no_wandb: false
+num_bullets: 13
+num_steps: 1298
+paged_optim: false
+parquet: /workspace/asherps/nla-data/subspace_bsf071226_halfchat_v11nl_1x_grounded/ar_sft.parquet
+quant: none
+reward_subspace_npz: null
+sample_every: 2000
+sample_max_new_tokens: 256
+save_dir: /workspace/asherps/nla-ckpts/subspace_ar_bsf071226_v11nl_1x_comp
+save_every: 2000
+seed: 0
+sft_eval_every: 250
+sft_eval_n: 24
+sft_evals: ''
+sidecar: /workspace/asherps/nla-data/subspace_bsf071226_halfchat_v11nl_1x_grounded/ar_sft.parquet
+strip_final_norm: true
+think_column: null
+think_unmasked: false
+uniform_l2_penalty: 0.0
+use_lora: false
+wandb_group: warmstart
+wandb_name: subspace_ar_bsf071226_v11nl_1x_comp
+wandb_project: nla
+wandb_tags: subspace_nla,bsf,v11nl_1x,comp