diff --git a/vs1/ckpt/Gclean/step_100/heads.pt b/vs1/ckpt/Gclean/step_100/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..bae4f9bae06aaa66755e25b6c8053eab7d01a903 --- /dev/null +++ b/vs1/ckpt/Gclean/step_100/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8b557e9a99606685e0122cb3d35469288e2eff9cf0e569f2d0f4c47ae7c83a61 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_100/meta.json b/vs1/ckpt/Gclean/step_100/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..6d51d44927782f1c805b34d78f9ccc7149bc767b --- /dev/null +++ b/vs1/ckpt/Gclean/step_100/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 100, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 02:04:07 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_100/new_rows.pt b/vs1/ckpt/Gclean/step_100/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..8baf3bce5edd7242303d4dbf4696dc672622ceaf --- /dev/null +++ b/vs1/ckpt/Gclean/step_100/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a565e803e26fe5e9b265bb982932e14c23fba79d34f7dbc749a6ba4ffd8b8eb3 +size 17072 diff --git a/vs1/ckpt/Gclean/step_100/optim_state.pt b/vs1/ckpt/Gclean/step_100/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..55898c38e7602507eaf20b0a1c278ab3c8251377 --- /dev/null +++ b/vs1/ckpt/Gclean/step_100/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bed5e4fa3a42af68db7f6ca0842a2c1b1c66d29a35a5aaa6f3a975b5e6552cb6 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_100/trainable_params.pt b/vs1/ckpt/Gclean/step_100/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..1cf5cea36fcf6bb276bb69aec841f66b161a8f5b --- /dev/null +++ b/vs1/ckpt/Gclean/step_100/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52a38a231d824827b4aa4a603938e8a6e4864d773872bf1b9335afd029837d0c +size 132339780 diff --git a/vs1/ckpt/Gclean/step_125/heads.pt b/vs1/ckpt/Gclean/step_125/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..872e66114e7736b2a832a3a516b211b86e2655a3 --- /dev/null +++ b/vs1/ckpt/Gclean/step_125/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b0e52e9c4c57c89cb0f7d8d3541b5048064b79461a9f21167e98d9053eb70fc0 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_125/meta.json b/vs1/ckpt/Gclean/step_125/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..91ac7a24315a97bbb9f7c67ae707c170c3119169 --- /dev/null +++ b/vs1/ckpt/Gclean/step_125/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 125, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 02:15:21 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_125/new_rows.pt b/vs1/ckpt/Gclean/step_125/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..27abc5a150407b4920bbde641122a6fbe8b7922c --- /dev/null +++ b/vs1/ckpt/Gclean/step_125/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e28146c625a2ecad051a92632b05db8a1940c50e2a95f80496108e4429eaca82 +size 17072 diff --git a/vs1/ckpt/Gclean/step_125/optim_state.pt b/vs1/ckpt/Gclean/step_125/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..d68f8121f1295ecddefcd36ce2846e1abf990378 --- /dev/null +++ b/vs1/ckpt/Gclean/step_125/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5312f55a0fda24608d1bb9ed9d9bb5aacd25fa07365fdf7c2097cf672b4e4603 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_125/trainable_params.pt b/vs1/ckpt/Gclean/step_125/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..9a3a7c07c2f252dd7a2ecb84df3bb3b7b281ee62 --- /dev/null +++ b/vs1/ckpt/Gclean/step_125/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22351a5a351b986212820f60930ea5f355f28643393b32d0b4cb15fd4fe08c97 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_150/heads.pt b/vs1/ckpt/Gclean/step_150/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..2543a4b9a129956335b6aeaab7daaf4e80539535 --- /dev/null +++ b/vs1/ckpt/Gclean/step_150/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9c05455427ff0104c32b20786a30814185c57913aca50716e9eb2a330f5297ac +size 46299566 diff --git a/vs1/ckpt/Gclean/step_150/meta.json b/vs1/ckpt/Gclean/step_150/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..23f257f41f48d0c09a521847200953a20f626790 --- /dev/null +++ b/vs1/ckpt/Gclean/step_150/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 150, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 02:26:13 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_150/new_rows.pt b/vs1/ckpt/Gclean/step_150/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..1e501234faadb184451fc6a1d0080a0b151e436c --- /dev/null +++ b/vs1/ckpt/Gclean/step_150/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b1779386416c91f34e70e65d49077f9aacd438535ac66ac61dcd355acac6b2b5 +size 17072 diff --git a/vs1/ckpt/Gclean/step_150/optim_state.pt b/vs1/ckpt/Gclean/step_150/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..5696a23023c32e5385c2131697e7152f3420e382 --- /dev/null +++ b/vs1/ckpt/Gclean/step_150/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d29f6da0ed568ba22484dd0dfa05588fb12cc2e5c72b45c4cb4f5587ffd6e99c +size 357349235 diff --git a/vs1/ckpt/Gclean/step_150/trainable_params.pt b/vs1/ckpt/Gclean/step_150/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..dd1c8b602ff8723fe87a5fb8a571c6a4aec7a249 --- /dev/null +++ b/vs1/ckpt/Gclean/step_150/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b160b00a909f545a0364c83211935420420c0088ad2334b44a053118796ebc08 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_175/heads.pt b/vs1/ckpt/Gclean/step_175/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..4287f4197ca48e63f842754bf70a8e5918247d62 --- /dev/null +++ b/vs1/ckpt/Gclean/step_175/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:54377bfe5cdb67bcd9ae0ba39aea4084b1a234dce163c210de843809ee8b2ab2 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_175/meta.json b/vs1/ckpt/Gclean/step_175/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..98b79e52df039f39164f7dd999aa65f4ddb31bcb --- /dev/null +++ b/vs1/ckpt/Gclean/step_175/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 175, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 02:37:19 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_175/new_rows.pt b/vs1/ckpt/Gclean/step_175/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..469617df9d06f4d8b18ac09772eb8d1317daf1ae --- /dev/null +++ b/vs1/ckpt/Gclean/step_175/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b152146c526f18074949419cd17b98dd7c62214d70c4bf4dad3514bfed8b5181 +size 17072 diff --git a/vs1/ckpt/Gclean/step_175/optim_state.pt b/vs1/ckpt/Gclean/step_175/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..253607e2d9ff073a90555c6757722dae7b3148c0 --- /dev/null +++ b/vs1/ckpt/Gclean/step_175/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:92e60474de047aeea4615864cd7bf8d6c3ce11c9b846f87e02f397dacecd3c96 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_175/trainable_params.pt b/vs1/ckpt/Gclean/step_175/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..81ea4659be8c29d62ba5f8280463f47608b6818b --- /dev/null +++ b/vs1/ckpt/Gclean/step_175/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:26d83e3cfa5b00c73768267e67b74089a1ea61cd86ce61b18e650e5fc2704b7f +size 132339780 diff --git a/vs1/ckpt/Gclean/step_200/heads.pt b/vs1/ckpt/Gclean/step_200/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..256000d01c63d8d89e82c38a7cf504ad68de7beb --- /dev/null +++ b/vs1/ckpt/Gclean/step_200/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5618f8cd21af5e75275fff8cb792f504a2a483b366ec177413762f32a91a6876 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_200/meta.json b/vs1/ckpt/Gclean/step_200/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..d70518d90fb3b057c6b008fbafed3b44fd1b682e --- /dev/null +++ b/vs1/ckpt/Gclean/step_200/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 200, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 02:48:24 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_200/new_rows.pt b/vs1/ckpt/Gclean/step_200/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..8440c033cb28ddbc0e50a6a12cc0e914bf68a722 --- /dev/null +++ b/vs1/ckpt/Gclean/step_200/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2b398c253e9f66c34443543dc4402e504341cfd368973761c9124944c6307ba +size 17072 diff --git a/vs1/ckpt/Gclean/step_200/optim_state.pt b/vs1/ckpt/Gclean/step_200/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..a49e77530a6db03d40b7241309135b28a701354a --- /dev/null +++ b/vs1/ckpt/Gclean/step_200/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:016d856e8369058d60358287baa95377ad179c68db6b600f0ecaae299783e845 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_200/trainable_params.pt b/vs1/ckpt/Gclean/step_200/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..625b824a5ae18c9afccb94de575924bf13dc6995 --- /dev/null +++ b/vs1/ckpt/Gclean/step_200/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ff9ed42533b7dbf57be186633fb9fa0dabc686d78bbf7685fbf4e39e41f75e8c +size 132339780 diff --git a/vs1/ckpt/Gclean/step_225/heads.pt b/vs1/ckpt/Gclean/step_225/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..6b827b3bda3fa8c84b49bd8d37847ac26b3303b7 --- /dev/null +++ b/vs1/ckpt/Gclean/step_225/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:de2b59ac89abef7bb0e5a2f1e47eb0dfce0080f94a97e5f3fd1f7dfc834e0462 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_225/meta.json b/vs1/ckpt/Gclean/step_225/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..6dd3591f37731295fb08d41e092aef1fc01c28b4 --- /dev/null +++ b/vs1/ckpt/Gclean/step_225/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 225, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 02:59:28 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_225/new_rows.pt b/vs1/ckpt/Gclean/step_225/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..69396b6ae1d62af903e4e4c9fc455aaf68309f53 --- /dev/null +++ b/vs1/ckpt/Gclean/step_225/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65bd084dbf72f7ea4ae73cae1120aef53689b2bdaf7a0bf13940d41bcd315d24 +size 17072 diff --git a/vs1/ckpt/Gclean/step_225/optim_state.pt b/vs1/ckpt/Gclean/step_225/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..59b739989a667244cdd3348748ff98fff33b76f0 --- /dev/null +++ b/vs1/ckpt/Gclean/step_225/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e5706e57c5903000b09e33e9c1cf603cd23253c95d865504684ee1644781c812 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_225/trainable_params.pt b/vs1/ckpt/Gclean/step_225/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..fbc815b1fd77d8efe2a777fe8b072b7f2fea450b --- /dev/null +++ b/vs1/ckpt/Gclean/step_225/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b18a3f3d407ada41e88ba972cb41b6347a6b63bd8c548058703e3a82b9b3b632 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_25/heads.pt b/vs1/ckpt/Gclean/step_25/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..a88900bdbca418962cf413e6217e5a9b9b7b7fbf --- /dev/null +++ b/vs1/ckpt/Gclean/step_25/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:105bae569398534a00f1ca330705fc0738288f0060b7ea755787e61842c4a5d3 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_25/meta.json b/vs1/ckpt/Gclean/step_25/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..61e198f89b368c6b694e254e74ff735f3ad107ce --- /dev/null +++ b/vs1/ckpt/Gclean/step_25/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 25, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 01:41:31 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_25/new_rows.pt b/vs1/ckpt/Gclean/step_25/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..8790b020c770d7d4ab5668cfed14cd3a26f939e3 --- /dev/null +++ b/vs1/ckpt/Gclean/step_25/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4b78e560f72031c03fb3fd769f9bafeef416ddb22407ef34620c4bbc103efb56 +size 17072 diff --git a/vs1/ckpt/Gclean/step_25/optim_state.pt b/vs1/ckpt/Gclean/step_25/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..d09aec4cab2dd1f3bdfc139ce3c761254a34b4f1 --- /dev/null +++ b/vs1/ckpt/Gclean/step_25/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:99d4e622f2480ebf85e69e362c88e198c1f24fb06430a2b1dab1cfc63c6c3506 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_25/trainable_params.pt b/vs1/ckpt/Gclean/step_25/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..731d5dd5541e400e6f63303a07b1709c08bb9586 --- /dev/null +++ b/vs1/ckpt/Gclean/step_25/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2c7f43f8d957cdc6321119af8c10e4b1139f9c5a26a91b82a49beec4d0e5c8e6 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_250/heads.pt b/vs1/ckpt/Gclean/step_250/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..0c58939ef5b5e78e72d0d047b15c7eaf2d914e21 --- /dev/null +++ b/vs1/ckpt/Gclean/step_250/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6ba4e01747650abfb96e8509beaecb7fed2e59eb391700e4ea88d256d9353f00 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_250/meta.json b/vs1/ckpt/Gclean/step_250/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..230bbcb170cd5de6dbc66c14954baec9db210da5 --- /dev/null +++ b/vs1/ckpt/Gclean/step_250/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 250, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 03:11:08 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_250/new_rows.pt b/vs1/ckpt/Gclean/step_250/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..6a518c3d0ab8453b1a7ff4ed310908d9eafb5c17 --- /dev/null +++ b/vs1/ckpt/Gclean/step_250/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:241087e9bb1fcca189e7dd12079922d6eb2d59e46b8384a328c70048431c27d2 +size 17072 diff --git a/vs1/ckpt/Gclean/step_250/optim_state.pt b/vs1/ckpt/Gclean/step_250/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..f80db35a0ee10f5cde7152f501b87d170bf458de --- /dev/null +++ b/vs1/ckpt/Gclean/step_250/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa1312f14dc636fadef8e9d1e3d1e1aa9abce2b25c85a010e117680c04541f1c +size 357349235 diff --git a/vs1/ckpt/Gclean/step_250/trainable_params.pt b/vs1/ckpt/Gclean/step_250/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..b8f23116f832c1616b62cab0e25aa68f04d4d71c --- /dev/null +++ b/vs1/ckpt/Gclean/step_250/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b608561f23770bdc8f8dee399a390de34ba9b760cce04fce8160f96bcffecec4 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_275/heads.pt b/vs1/ckpt/Gclean/step_275/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..6b8f8c92385f4b6021fa4fd2e9149f65bb1c654a --- /dev/null +++ b/vs1/ckpt/Gclean/step_275/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:95ddab7c6a01970eebf886b5f00a6f27a8ead44e75059d2cea7da78a0cb1224e +size 46299566 diff --git a/vs1/ckpt/Gclean/step_275/meta.json b/vs1/ckpt/Gclean/step_275/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..2da9c24cadf90be165574d59bd5ae7361832b0e9 --- /dev/null +++ b/vs1/ckpt/Gclean/step_275/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 275, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 03:22:35 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_275/new_rows.pt b/vs1/ckpt/Gclean/step_275/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..596ae7fba1a5a1a2574a34375fbac7178a6cc7c3 --- /dev/null +++ b/vs1/ckpt/Gclean/step_275/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b8676f11d16b307f66a56eeb6bb0df213ab17c1085881f49a72b6bbbe2e75a04 +size 17072 diff --git a/vs1/ckpt/Gclean/step_275/optim_state.pt b/vs1/ckpt/Gclean/step_275/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..2fe4db353b288f7a45283cfdcbc0339394e43940 --- /dev/null +++ b/vs1/ckpt/Gclean/step_275/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e77347ab526d2d92971595835cc3dbdd1db420aace0381648effcec7f07dda84 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_275/trainable_params.pt b/vs1/ckpt/Gclean/step_275/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..00c3927e7b246f4f56ea7e134e2ea2e9a1537e7a --- /dev/null +++ b/vs1/ckpt/Gclean/step_275/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2806e88328d5aed126f14a2cc45e4fa393f79f0ce1b766f53eb039fc0e18b3e +size 132339780 diff --git a/vs1/ckpt/Gclean/step_300/heads.pt b/vs1/ckpt/Gclean/step_300/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..777ef367a8872782ee907b9234e940a0ac25d09f --- /dev/null +++ b/vs1/ckpt/Gclean/step_300/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e094264358f6eeb737c0f8cae8bd2bd659f0411fc37f45604f161f50f240f5ad +size 46299566 diff --git a/vs1/ckpt/Gclean/step_300/meta.json b/vs1/ckpt/Gclean/step_300/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..bf8c7279526b88e58f3efc4c1c6790f3a1c21e59 --- /dev/null +++ b/vs1/ckpt/Gclean/step_300/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 300, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 03:33:53 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_300/new_rows.pt b/vs1/ckpt/Gclean/step_300/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..72f2ffd17ff56d5da31f91732efbe42bcfec59de --- /dev/null +++ b/vs1/ckpt/Gclean/step_300/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:888c55711a0c0d511939813e14976bc47e22c892e57f23795bd0162a17bf64f2 +size 17072 diff --git a/vs1/ckpt/Gclean/step_300/optim_state.pt b/vs1/ckpt/Gclean/step_300/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..08dafc0b6aa776443820ae558e76f29c0e220b28 --- /dev/null +++ b/vs1/ckpt/Gclean/step_300/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:149f3f8182752d8b7d4c8944e45f7b445cab98a13b4c5ba6203afb4adbebe737 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_300/trainable_params.pt b/vs1/ckpt/Gclean/step_300/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..236dc0f957e6685cf87db357ffb37bed44ad0a37 --- /dev/null +++ b/vs1/ckpt/Gclean/step_300/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:85f9a5b455eaed216f1c550aa2b713d2b7df59933d2d43da47ce4c40cefbaf69 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_325/heads.pt b/vs1/ckpt/Gclean/step_325/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..daf84aa93d1380424d5698e0a75cf7869c5ff9bb --- /dev/null +++ b/vs1/ckpt/Gclean/step_325/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:46326d1437767c85302348b9a08bc9e164e1e70a85d0e591903a496ca95cacce +size 46299566 diff --git a/vs1/ckpt/Gclean/step_325/meta.json b/vs1/ckpt/Gclean/step_325/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..99139b60b5001cddcb45dd9505f0a4bfff466e15 --- /dev/null +++ b/vs1/ckpt/Gclean/step_325/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 325, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 03:44:35 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_325/new_rows.pt b/vs1/ckpt/Gclean/step_325/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..aaa2525c3cc3f093f2f698dfb5c59d9d2fe09c25 --- /dev/null +++ b/vs1/ckpt/Gclean/step_325/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:536300761839a1afd485b3596239db83d5618273b70f6f056fec9354f51850bb +size 17072 diff --git a/vs1/ckpt/Gclean/step_325/optim_state.pt b/vs1/ckpt/Gclean/step_325/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..d8a6b43e6343ae69b92d6d6a4bcc35ea34f053a0 --- /dev/null +++ b/vs1/ckpt/Gclean/step_325/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:001b98f06329818596a6a0123e88f0245ad8cf34acd04598c3faa3b1c474f524 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_325/trainable_params.pt b/vs1/ckpt/Gclean/step_325/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..44a57cb1645605364ef6b16c784862c6ac11bfc5 --- /dev/null +++ b/vs1/ckpt/Gclean/step_325/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fa7798af89d8a0557548e9fdf4241f897f343dbf8101fdb817f1232c4309f056 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_350/heads.pt b/vs1/ckpt/Gclean/step_350/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..894f2532bc2766fcef4b2ff77d3bf6767be46302 --- /dev/null +++ b/vs1/ckpt/Gclean/step_350/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d2d71eb9976810db800ea5a636cf4ba9bddffe31bc4e7825980790a038be3559 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_350/meta.json b/vs1/ckpt/Gclean/step_350/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..b67e705ea136168973ee6d84cb8a424f9d8b09e1 --- /dev/null +++ b/vs1/ckpt/Gclean/step_350/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 350, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 03:54:50 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_350/new_rows.pt b/vs1/ckpt/Gclean/step_350/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..863a2b78706c3a6cd8f952002bb362663c77081b --- /dev/null +++ b/vs1/ckpt/Gclean/step_350/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77932ae51e40b4c0769d8b79bc0ed1f3a871dd3ad6a9ca5d6e2ee23ce3045a43 +size 17072 diff --git a/vs1/ckpt/Gclean/step_350/optim_state.pt b/vs1/ckpt/Gclean/step_350/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..751ac2847382d50cb49a09e6ad7fb4b204a383f8 --- /dev/null +++ b/vs1/ckpt/Gclean/step_350/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3fc4d2dfe6ec7ad0e04d4c3693e99134b59a0baa0d2f770e03d985463aee705a +size 357349235 diff --git a/vs1/ckpt/Gclean/step_350/trainable_params.pt b/vs1/ckpt/Gclean/step_350/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..f86d008470a82e3bda610495114e748f4a676256 --- /dev/null +++ b/vs1/ckpt/Gclean/step_350/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c9bb7e9b0ce318da2aee8855010984c8ca0e79cbf02c7ee39a7f2c7f13afd710 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_375/heads.pt b/vs1/ckpt/Gclean/step_375/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..b13a277b50435ebd9338bffbeb7e2db30bde116b --- /dev/null +++ b/vs1/ckpt/Gclean/step_375/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e89c77d316845407117bf2d19d423cff8f27b60a2f1d005bdf4ed48436b8131a +size 46299566 diff --git a/vs1/ckpt/Gclean/step_375/meta.json b/vs1/ckpt/Gclean/step_375/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..e1ae338570174ca40fd0f1eac2404020ceee2558 --- /dev/null +++ b/vs1/ckpt/Gclean/step_375/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 375, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 04:05:10 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_375/new_rows.pt b/vs1/ckpt/Gclean/step_375/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..9a0024ab10197ff898537e4d5f2850d6e13b0106 --- /dev/null +++ b/vs1/ckpt/Gclean/step_375/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4abfdd49da79d009f6be356a4a536c8b80f813dd7bbf77de2205348486a22e1d +size 17072 diff --git a/vs1/ckpt/Gclean/step_375/optim_state.pt b/vs1/ckpt/Gclean/step_375/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..f4dd2f2815b70fe3834756137e32e6cb33aeb5e9 --- /dev/null +++ b/vs1/ckpt/Gclean/step_375/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ddd533c6ac2b3f711c710c31e12bdb0495d56ea24930c7c5b52c8ede1132f9c9 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_375/trainable_params.pt b/vs1/ckpt/Gclean/step_375/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..02ffe495572a420c63e43e4a8ef1b655124c49c2 --- /dev/null +++ b/vs1/ckpt/Gclean/step_375/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2f8ed0b6ad9dcb3839bf48f5757b087fd9612952c7b7d2be3ea0f29c89f0c59 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_400/heads.pt b/vs1/ckpt/Gclean/step_400/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..3441707adf18e90fa2efc6ad64d0f772207b45ad --- /dev/null +++ b/vs1/ckpt/Gclean/step_400/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:48aa4b54787fec79cfd9b9044b1a1cc544a7271d6c4ca4bf89822a9a3a26d0a8 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_400/meta.json b/vs1/ckpt/Gclean/step_400/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..1926da42190ff51d0164ef937293900888ba8f3e --- /dev/null +++ b/vs1/ckpt/Gclean/step_400/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 400, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 04:15:40 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_400/new_rows.pt b/vs1/ckpt/Gclean/step_400/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..e98de590d1b40db103ee341d15094d6d65256365 --- /dev/null +++ b/vs1/ckpt/Gclean/step_400/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:65e2ae88ec3ba381c85f847bc63674c3bf5188ff76f54458aa18ab2b05919220 +size 17072 diff --git a/vs1/ckpt/Gclean/step_400/optim_state.pt b/vs1/ckpt/Gclean/step_400/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..05993b8ee30ae89c8b0d721d78d6e0af71d27785 --- /dev/null +++ b/vs1/ckpt/Gclean/step_400/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c468b73116500355011101d19c4f7445ea22aefbc8d3f219e4eda6c22457208a +size 357349235 diff --git a/vs1/ckpt/Gclean/step_400/trainable_params.pt b/vs1/ckpt/Gclean/step_400/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..1fb17967ddc77a3dd89f070a264480d5c39bac1c --- /dev/null +++ b/vs1/ckpt/Gclean/step_400/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9e097520a9cdfa7dff0be5da28956b45e475bbcf7cd36b0cbb1e1cf079d68f0d +size 132339780 diff --git a/vs1/ckpt/Gclean/step_425/heads.pt b/vs1/ckpt/Gclean/step_425/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..911f471239392133f6e68016512fdb13e180e174 --- /dev/null +++ b/vs1/ckpt/Gclean/step_425/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8a16fd549054534ecc0d14bcf02b5ca76ad0127d343cb0758423c3a5e610d427 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_425/meta.json b/vs1/ckpt/Gclean/step_425/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..6cead130060006efc431dc06cd44039ea0b8f948 --- /dev/null +++ b/vs1/ckpt/Gclean/step_425/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 425, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 04:26:00 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_425/new_rows.pt b/vs1/ckpt/Gclean/step_425/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..f067e1080bd6e20e0cf89dd5b87f1274787cd373 --- /dev/null +++ b/vs1/ckpt/Gclean/step_425/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:77ebc70b4d5702b12b2cc9894700f9c80ea1703d4034a1fa3608521d6491984f +size 17072 diff --git a/vs1/ckpt/Gclean/step_425/optim_state.pt b/vs1/ckpt/Gclean/step_425/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..fc548a52c61a8b1a1a28b06bbdcab58d9f2240c8 --- /dev/null +++ b/vs1/ckpt/Gclean/step_425/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:91bfe1b5e4743cb39aab89978d9ab40dce5637a7e9afbfb4fb7148cb6cd63b96 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_425/trainable_params.pt b/vs1/ckpt/Gclean/step_425/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..cf70335375390e93e3a87a91ec3d79b6e7df8f1e --- /dev/null +++ b/vs1/ckpt/Gclean/step_425/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f4b4806db081d80bf986e0f511024ad2cbbd533f709e88ffbace4b0390f31950 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_450/heads.pt b/vs1/ckpt/Gclean/step_450/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..3e3bb2d71c1aa03e1894c3dd74b8fa0fa9c57319 --- /dev/null +++ b/vs1/ckpt/Gclean/step_450/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:353f2119d4b838c5912dcc79b23e27dd63e7e63b420b89112ecb26365ee14ba7 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_450/meta.json b/vs1/ckpt/Gclean/step_450/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..f0e7865a2c0ab74a35127e734e88c82b31283591 --- /dev/null +++ b/vs1/ckpt/Gclean/step_450/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 450, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 04:36:34 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_450/new_rows.pt b/vs1/ckpt/Gclean/step_450/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..e6aa8db30a3dc8aa51af4148c3203bd334ee9367 --- /dev/null +++ b/vs1/ckpt/Gclean/step_450/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:996ca09b7c44a7b3964d5f1667c86ed7d20c47a143128c94a71cfd6338fb3d73 +size 17072 diff --git a/vs1/ckpt/Gclean/step_450/optim_state.pt b/vs1/ckpt/Gclean/step_450/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..467f062b6c956f8e681a4ee5c8d9a9657551293a --- /dev/null +++ b/vs1/ckpt/Gclean/step_450/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f394563d326b6621e818f89138fa9e2dd20ca123003f47b91bbef76d8ecca0d7 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_450/trainable_params.pt b/vs1/ckpt/Gclean/step_450/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..df602e081129a03987ff3e5384f4bb69881b5eae --- /dev/null +++ b/vs1/ckpt/Gclean/step_450/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f77d8d5368f933db207cdfe06dc832be543b3fda7b0e1aa3976c5780b1b50c9 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_475/heads.pt b/vs1/ckpt/Gclean/step_475/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..ff97e490738f0ea555b21d2e6a235acb5812481d --- /dev/null +++ b/vs1/ckpt/Gclean/step_475/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2161ad3d9c33a6df086027474fe61ad5d439b63cee84a7b30ab5cad50ced9fc0 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_475/meta.json b/vs1/ckpt/Gclean/step_475/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..2aa2ee62aa80c3492b884e18f210eece1dad467a --- /dev/null +++ b/vs1/ckpt/Gclean/step_475/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 475, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 04:47:27 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_475/new_rows.pt b/vs1/ckpt/Gclean/step_475/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..5c9aab96423dd481d7a452fb10dc504384d61e92 --- /dev/null +++ b/vs1/ckpt/Gclean/step_475/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6cfd4a2b32ab35ec4166058a6f8a8fd35827c106535794d14a6625c234cf2c39 +size 17072 diff --git a/vs1/ckpt/Gclean/step_475/optim_state.pt b/vs1/ckpt/Gclean/step_475/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..fedb0a3b534d57598659b055613275a95e576d5b --- /dev/null +++ b/vs1/ckpt/Gclean/step_475/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:27a10f2e0b7b062719f8a817772fde6e34d36156c97a7641f0f045dd57adc988 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_475/trainable_params.pt b/vs1/ckpt/Gclean/step_475/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..b349f47a670a7dbc0958e514a8f3efae7f973b2b --- /dev/null +++ b/vs1/ckpt/Gclean/step_475/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f684ceaed6308329e6deee3f917bb59927da71d5d3542efb2911f0c7585dfabe +size 132339780 diff --git a/vs1/ckpt/Gclean/step_50/heads.pt b/vs1/ckpt/Gclean/step_50/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..d14e847520e6366c1d5dd43f164d775951b536ab --- /dev/null +++ b/vs1/ckpt/Gclean/step_50/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:86baaacab1a96d0eb20d49a1dac24bc26116140aa9d3d6909b80a608dc667c9e +size 46299566 diff --git a/vs1/ckpt/Gclean/step_50/meta.json b/vs1/ckpt/Gclean/step_50/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..49af44f21b5bff8ec383a1f51584c3129efcff51 --- /dev/null +++ b/vs1/ckpt/Gclean/step_50/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 50, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 01:47:10 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_50/new_rows.pt b/vs1/ckpt/Gclean/step_50/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..20181b0a43cee17e3e2e69171f01bbf1d688dd09 --- /dev/null +++ b/vs1/ckpt/Gclean/step_50/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:60da611c35622f6322ba3facd5dd75bce37e59efdb555cc5f4927ea0b631f32e +size 17072 diff --git a/vs1/ckpt/Gclean/step_50/optim_state.pt b/vs1/ckpt/Gclean/step_50/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..901c015537168f61a00e367f93eef38e3ecba4e2 --- /dev/null +++ b/vs1/ckpt/Gclean/step_50/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:bc7b36363b4c20ce891c8e169eff947468ca43168504618e5a5777b9fe4c219d +size 357349235 diff --git a/vs1/ckpt/Gclean/step_50/trainable_params.pt b/vs1/ckpt/Gclean/step_50/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..238b280ac2700b6227338cee479c43cac58f0f41 --- /dev/null +++ b/vs1/ckpt/Gclean/step_50/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7faf522b0d01b77e5f69ac0a3e83dd835e73f42182cbcd3a47013517d3c90c16 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_500/heads.pt b/vs1/ckpt/Gclean/step_500/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..880d7a5f1aed4b7c59b6e4a37d584f6969dc85dd --- /dev/null +++ b/vs1/ckpt/Gclean/step_500/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:429484cba726757cb29fdcca3ad64853af323e4b3ecdcdc27f285b77d97147de +size 46299566 diff --git a/vs1/ckpt/Gclean/step_500/meta.json b/vs1/ckpt/Gclean/step_500/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..bd0b600299f581c126b23ec47e59b600496d646d --- /dev/null +++ b/vs1/ckpt/Gclean/step_500/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 500, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 04:58:10 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_500/new_rows.pt b/vs1/ckpt/Gclean/step_500/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..a22cb8607904948fbe52d4c7e7081ce515efedf1 --- /dev/null +++ b/vs1/ckpt/Gclean/step_500/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:002601528b7447b113c4a7ad375c1b3062e25e4fc2d94a4539e49bca35db932f +size 17072 diff --git a/vs1/ckpt/Gclean/step_500/optim_state.pt b/vs1/ckpt/Gclean/step_500/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..aa648f76efa6092e4d9a34a5298ca3edae943f57 --- /dev/null +++ b/vs1/ckpt/Gclean/step_500/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21b8b4d6ce7099efd7e3da800a6b3788bcc776dad338886f1c96009eea5399f7 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_500/trainable_params.pt b/vs1/ckpt/Gclean/step_500/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..e2e388c408d15db81be7d1e8c129b00c9375eab8 --- /dev/null +++ b/vs1/ckpt/Gclean/step_500/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab43adcf9cb5e3f255e7693c4e38cebb3d12a9f973661ffac1ffad57c39d0b66 +size 132339780 diff --git a/vs1/ckpt/Gclean/step_75/heads.pt b/vs1/ckpt/Gclean/step_75/heads.pt new file mode 100644 index 0000000000000000000000000000000000000000..572e0db292a6a0ccb19f535d08f66711b7c12033 --- /dev/null +++ b/vs1/ckpt/Gclean/step_75/heads.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9bfca10ae06205424a5db4a7cf069142074b51c6ebeca46ab2010005af1ab232 +size 46299566 diff --git a/vs1/ckpt/Gclean/step_75/meta.json b/vs1/ckpt/Gclean/step_75/meta.json new file mode 100644 index 0000000000000000000000000000000000000000..f1c771b913b37a47d5fe7deb483d1e245f3dad0f --- /dev/null +++ b/vs1/ckpt/Gclean/step_75/meta.json @@ -0,0 +1,97 @@ +{ + "variant": "Gclean", + "step": 75, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "lam": 1.0, + "p1": false, + "run_name": "Gclean", + "time": "2026-10-23 01:52:48 KST", + "cfg": { + "variant": "Gclean", + "base_variant": "G", + "fmt": "clean (지시서 53 §3, 49 §4.3)", + "p1": false, + "p1_ids": null, + "lora_r": 16, + "lora_alpha": 32, + "lora_dropout": 0.05, + "lora_targets": [ + "q_proj", + "k_proj", + "v_proj", + "o_proj", + "gate_proj", + "up_proj", + "down_proj" + ], + "lora_regex": "model\\.language_model\\.layers\\.\\d+\\.(self_attn\\.(q_proj|k_proj|v_proj|o_proj)|mlp\\.(gate_proj|up_proj|down_proj))", + "micro_batch": 1, + "grad_accum": 8, + "lr": 0.0001, + "head_lr": 0.0001, + "new_token_lr": 0.0001, + "scheduler": "constant, linear warmup 10 steps (lr_k = 1e-4·min(1,k/10))", + "warmup_steps": 10, + "weight_decay": 0.0, + "max_grad_norm": 1.0, + "optimizer": "AdamW", + "seed": 0, + "bf16": true, + "grad_ckpt": true, + "frame_rule": "D16", + "n_input_frames": 32, + "grid_t": 16, + "frame_hw": [ + 480, + 640 + ], + "base_model_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B-qwen3vl", + "processor_dir": "/workspace/code/spatial_reasoning/repro_baseline22/weights/GeoVR-VGGT-Qwen3-VL-4B", + "lam": 1.0, + "n_samples": 5000, + "n_steps_full": 625, + "steps_this_run": 625, + "order_sha1": "d6a78f37e8dad34cfe01e38cc4458a84c1bb8985", + "data": "/workspace/code/spatial_reasoning/VS1/data/train_v2.jsonl", + "data_sha256": "5f03730d4cfd2310c740ccf15d25779d9fabdf68b90f182b8f770e656b916438", + "ckpt_name": "Gclean", + "git_commit": "205619c04e5ad473afc7df31a8e3bea59efb9858", + "workers": 4, + "new_tokens": [ + "", + "", + "" + ], + "new_token_ids": [ + 151669, + 151670, + 151671 + ], + "heads": [ + "box", + "cam(GeoVR CameraDec)", + "rel" + ], + "cam_init": "GeoVR model.camera_token.weight[1]", + "cam_head": "GeoVR camera_head (CameraDec 2560, fp32, ckpt)", + "cam_target": "gconv.cam_g: first-frame w2c [s·t, quat xyzw w>=0], s=VGGT scale", + "cam_loss": "|dt|.clamp(100).mean + min(|q̂-q|,|q̂+q|).mean, no FoV, no 2.5x", + "scale_s1": false, + "scale_file_sha256": "9e2f0a5abe5f1dbc4f4f6a074fd5584f668c209302f997b778e373bfea42bb95", + "cam_mask": { + "n_pair": 4000, + "n_valid": 4000, + "frame0_invalid": 0, + "scale_none": 0, + "scale_missing_scene": 0, + "frames_mismatch": 0, + "s1_used": false, + "n_masked": 0 + }, + "cam_mode": "untied", + "ckpt_every": 25, + "curve_every": 25, + "curve_ids": "/workspace/code/spatial_reasoning/VS1/K/data/g_curve_ids.json", + "attn": "flash_attention_2" + } +} \ No newline at end of file diff --git a/vs1/ckpt/Gclean/step_75/new_rows.pt b/vs1/ckpt/Gclean/step_75/new_rows.pt new file mode 100644 index 0000000000000000000000000000000000000000..d79eaefe445c47387545f3e7988b33c6533e69e5 --- /dev/null +++ b/vs1/ckpt/Gclean/step_75/new_rows.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:56307778186ed04ecdad444478033f8826d769ea7448762d03730342b93ea1ee +size 17072 diff --git a/vs1/ckpt/Gclean/step_75/optim_state.pt b/vs1/ckpt/Gclean/step_75/optim_state.pt new file mode 100644 index 0000000000000000000000000000000000000000..6660ea1e67ecbacb57d1bb5cdb158bdd670e9c42 --- /dev/null +++ b/vs1/ckpt/Gclean/step_75/optim_state.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f53826d80d36407ea77e9811b96d78bdca978bd5fc0adc978b28f9fdc7efea19 +size 357349235 diff --git a/vs1/ckpt/Gclean/step_75/trainable_params.pt b/vs1/ckpt/Gclean/step_75/trainable_params.pt new file mode 100644 index 0000000000000000000000000000000000000000..fb4a68954f7dd18e4f434dc9eee31456feb62cc3 --- /dev/null +++ b/vs1/ckpt/Gclean/step_75/trainable_params.pt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9b1e2b270a81f960d445e726b357e89bceebc8028cf421c7083ca9916d3bc8c4 +size 132339780