Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,13 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7.jsonl filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/latest-run/run-6hhs8vrz.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_100515-lvbcavkp/run-lvbcavkp.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_111805-th1rdka5/run-th1rdka5.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/run-6hhs8vrz.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7.jsonl filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/run-6hhs8vrz.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100515-lvbcavkp/run-lvbcavkp.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_111805-th1rdka5/run-th1rdka5.wandb filter=lfs diff=lfs merge=lfs -text
+prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250502_114824-6hhs8vrz/run-6hhs8vrz.wandb filter=lfs diff=lfs merge=lfs -text

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/checkpoints/latest-checkpoint.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:521937aca5962573d06f63e6fd4cfff59566869578a4eb8dade3e5c7cfcb19ae
+size 7247200761

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/checkpoints/step-055444-epoch-01-loss=0.6558.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:521937aca5962573d06f63e6fd4cfff59566869578a4eb8dade3e5c7cfcb19ae
+size 7247200761

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/config.json

@@ -0,0 +1,59 @@
+{
+  "dataset": {
+    "align_stage_components": [
+      "download/llava-laion-cc-sbu-558k/chat.json",
+      "download/llava-laion-cc-sbu-558k"
+    ],
+    "dataset_id": "llava-v15",
+    "dataset_root_dir": "data",
+    "finetune_stage_components": [
+      "download/llava-v1.5-instruct/llava_v1_5_mix665k.json",
+      "download/llava-v1.5-instruct"
+    ],
+    "type": "llava-v15"
+  },
+  "hf_token": "hf_token.txt",
+  "model": {
+    "align_epochs": 1,
+    "align_global_batch_size": 8,
+    "align_learning_rate": 0.001,
+    "align_lr_scheduler_type": "linear-warmup+cosine-decay",
+    "align_max_grad_norm": 1.0,
+    "align_max_steps": null,
+    "align_per_device_batch_size": 8,
+    "align_train_strategy": "fsdp-shard-grad-op",
+    "align_warmup_ratio": 0.03,
+    "align_weight_decay": 0.0,
+    "arch_specifier": "no-align+fused-gelu-mlp",
+    "enable_gradient_checkpointing": true,
+    "enable_mixed_precision_training": true,
+    "finetune_epochs": 2,
+    "finetune_global_batch_size": 24,
+    "finetune_learning_rate": 2e-05,
+    "finetune_lr_scheduler_type": "linear-warmup+cosine-decay",
+    "finetune_max_grad_norm": 1.0,
+    "finetune_max_steps": null,
+    "finetune_per_device_batch_size": 6,
+    "finetune_train_strategy": "fsdp-full-shard",
+    "finetune_warmup_ratio": 0.03,
+    "finetune_weight_decay": 0.1,
+    "image_resize_strategy": "resize-naive",
+    "llm_backbone_id": "qwen25-1_5b-extra",
+    "llm_max_length": 32768,
+    "model_id": "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "reduce_in_full_precision": false,
+    "type": "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "vision_backbone_id": "dinosiglip-vit-so-224px"
+  },
+  "pretrained_checkpoint": null,
+  "run_id": "prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7",
+  "run_root_dir": "runs",
+  "seed": 7,
+  "stage": "finetune",
+  "trackers": [
+    "jsonl",
+    "wandb"
+  ],
+  "wandb_entity": "arash-akbari-stu-northeastern-university",
+  "wandb_project": "Moxin-VLM"
+}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/config.yaml

@@ -0,0 +1,52 @@
+dataset:
+  align_stage_components:
+  - download/llava-laion-cc-sbu-558k/chat.json
+  - download/llava-laion-cc-sbu-558k
+  dataset_id: llava-v15
+  dataset_root_dir: data
+  finetune_stage_components:
+  - download/llava-v1.5-instruct/llava_v1_5_mix665k.json
+  - download/llava-v1.5-instruct
+  type: llava-v15
+hf_token: hf_token.txt
+model:
+  align_epochs: 1
+  align_global_batch_size: 8
+  align_learning_rate: 0.001
+  align_lr_scheduler_type: linear-warmup+cosine-decay
+  align_max_grad_norm: 1.0
+  align_max_steps: null
+  align_per_device_batch_size: 8
+  align_train_strategy: fsdp-shard-grad-op
+  align_warmup_ratio: 0.03
+  align_weight_decay: 0.0
+  arch_specifier: no-align+fused-gelu-mlp
+  enable_gradient_checkpointing: true
+  enable_mixed_precision_training: true
+  finetune_epochs: 2
+  finetune_global_batch_size: 24
+  finetune_learning_rate: 2.0e-05
+  finetune_lr_scheduler_type: linear-warmup+cosine-decay
+  finetune_max_grad_norm: 1.0
+  finetune_max_steps: null
+  finetune_per_device_batch_size: 6
+  finetune_train_strategy: fsdp-full-shard
+  finetune_warmup_ratio: 0.03
+  finetune_weight_decay: 0.1
+  image_resize_strategy: resize-naive
+  llm_backbone_id: qwen25-1_5b-extra
+  llm_max_length: 32768
+  model_id: prism-qwen25-extra-dinosiglip-224px+1_5b
+  reduce_in_full_precision: false
+  type: prism-qwen25-extra-dinosiglip-224px+1_5b
+  vision_backbone_id: dinosiglip-vit-so-224px
+pretrained_checkpoint: null
+run_id: prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7
+run_root_dir: runs
+seed: 7
+stage: finetune
+trackers:
+- jsonl
+- wandb
+wandb_entity: arash-akbari-stu-northeastern-university
+wandb_project: Moxin-VLM

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0960f58bd0a805c19ffbfe2ec514c34218a1bdf2636b4444f0aa326ec2e87333
+size 11339854

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/run-metrics.jsonl

@@ -0,0 +1 @@
+{"hparams": {"dataset": {"align_stage_components": ["download/llava-laion-cc-sbu-558k/chat.json", "download/llava-laion-cc-sbu-558k"], "dataset_id": "llava-v15", "dataset_root_dir": "data", "finetune_stage_components": ["download/llava-v1.5-instruct/llava_v1_5_mix665k.json", "download/llava-v1.5-instruct"], "type": "llava-v15"}, "hf_token": "hf_token.txt", "model": {"align_epochs": 1, "align_global_batch_size": 8, "align_learning_rate": 0.001, "align_lr_scheduler_type": "linear-warmup+cosine-decay", "align_max_grad_norm": 1.0, "align_max_steps": null, "align_per_device_batch_size": 8, "align_train_strategy": "fsdp-shard-grad-op", "align_warmup_ratio": 0.03, "align_weight_decay": 0.0, "arch_specifier": "no-align+fused-gelu-mlp", "enable_gradient_checkpointing": true, "enable_mixed_precision_training": true, "finetune_epochs": 2, "finetune_global_batch_size": 24, "finetune_learning_rate": 2e-05, "finetune_lr_scheduler_type": "linear-warmup+cosine-decay", "finetune_max_grad_norm": 1.0, "finetune_max_steps": null, "finetune_per_device_batch_size": 6, "finetune_train_strategy": "fsdp-full-shard", "finetune_warmup_ratio": 0.03, "finetune_weight_decay": 0.1, "image_resize_strategy": "resize-naive", "llm_backbone_id": "qwen25-1_5b-extra", "llm_max_length": 32768, "model_id": "prism-qwen25-extra-dinosiglip-224px+1_5b", "reduce_in_full_precision": false, "type": "prism-qwen25-extra-dinosiglip-224px+1_5b", "vision_backbone_id": "dinosiglip-vit-so-224px"}, "pretrained_checkpoint": null, "run_id": "prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7", "run_root_dir": "runs", "seed": 7, "stage": "finetune", "trackers": ["jsonl", "wandb"], "wandb_entity": "arash-akbari-stu-northeastern-university", "wandb_project": "Moxin-VLM"}, "run_id": "prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2025-05-02T11:48:24.512248306-07:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/logs/debug-core.log"}
+{"time":"2025-05-02T11:48:25.628691797-07:00","level":"INFO","msg":"created new stream","id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.62873432-07:00","level":"INFO","msg":"stream: started","id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.62879471-07:00","level":"INFO","msg":"handler: started","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.628819016-07:00","level":"INFO","msg":"writer: Do: started","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.628835531-07:00","level":"INFO","msg":"sender: started","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.959126298-07:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2025-05-02T11:50:45.633111697-07:00","level":"INFO","msg":"stream: closing","id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:45.633155452-07:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2025-05-02T11:50:45.633187911-07:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2025-05-02T11:50:46.081032653-07:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-05-02T11:50:46.287259059-07:00","level":"INFO","msg":"handler: closed","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:46.287318307-07:00","level":"INFO","msg":"writer: Close: closed","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:46.287337876-07:00","level":"INFO","msg":"sender: closed","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:46.287531815-07:00","level":"INFO","msg":"stream: closed","id":"6hhs8vrz"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/debug.log

@@ -0,0 +1,23 @@
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Configure stats pid to 1932762
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Loading settings from /home/user1/.config/wandb/settings
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Loading settings from /home/user1/arashwork/prismatic-vlms/wandb/settings
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:setup_run_log_directory():662] Logging user logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/logs/debug.log
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/logs/debug-internal.log
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():781] calling init triggers
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'type': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'model_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'arch_specifier': 'no-align+fused-gelu-mlp', 'vision_backbone_id': 'dinosiglip-vit-so-224px', 'llm_backbone_id': 'qwen25-1_5b-extra', 'image_resize_strategy': 'resize-naive', 'llm_max_length': 32768, 'align_epochs': 1, 'align_max_steps': None, 'align_global_batch_size': 8, 'align_per_device_batch_size': 8, 'align_learning_rate': 0.001, 'align_weight_decay': 0.0, 'align_max_grad_norm': 1.0, 'align_lr_scheduler_type': 'linear-warmup+cosine-decay', 'align_warmup_ratio': 0.03, 'align_train_strategy': 'fsdp-shard-grad-op', 'finetune_epochs': 2, 'finetune_max_steps': None, 'finetune_global_batch_size': 24, 'finetune_per_device_batch_size': 6, 'finetune_learning_rate': 2e-05, 'finetune_weight_decay': 0.1, 'finetune_max_grad_norm': 1.0, 'finetune_lr_scheduler_type': 'linear-warmup+cosine-decay', 'finetune_warmup_ratio': 0.03, 'finetune_train_strategy': 'fsdp-full-shard', 'enable_gradient_checkpointing': True, 'enable_mixed_precision_training': True, 'reduce_in_full_precision': False}, 'dataset': {'type': 'llava-v15', 'dataset_id': 'llava-v15', 'align_stage_components': ['download/llava-laion-cc-sbu-558k/chat.json', 'download/llava-laion-cc-sbu-558k'], 'finetune_stage_components': ['download/llava-v1.5-instruct/llava_v1_5_mix665k.json', 'download/llava-v1.5-instruct'], 'dataset_root_dir': 'data'}, 'stage': 'finetune', 'pretrained_checkpoint': None, 'run_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7', 'run_root_dir': 'runs', 'seed': 7, 'hf_token': 'hf_token.txt', 'trackers': ['jsonl', 'wandb'], 'wandb_project': 'Moxin-VLM', 'wandb_entity': 'arash-akbari-stu-northeastern-university', '_wandb': {}}
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():809] starting backend
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():813] sending inform_init request
+2025-05-02 11:48:24,510 INFO    MainThread:1932762 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-05-02 11:48:24,510 INFO    MainThread:1932762 [wandb_init.py:init():823] backend started and connected
+2025-05-02 11:48:24,512 INFO    MainThread:1932762 [wandb_init.py:init():915] updated telemetry
+2025-05-02 11:48:24,517 INFO    MainThread:1932762 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-05-02 11:48:25,957 INFO    MainThread:1932762 [wandb_init.py:init():1014] starting run threads in backend
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_console_start():2454] atexit reg
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-05-02 11:48:26,002 INFO    MainThread:1932762 [wandb_init.py:init():1056] run started, returning control to user process
+2025-05-02 11:50:45,632 INFO    MsgRouterThr:1932762 [mailbox.py:close():129] [no run ID] Closing mailbox, abandoning 1 handles.

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/files/config.yaml

@@ -0,0 +1,100 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m: []
+        python_version: 3.10.16
+        t:
+            "1":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "2":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "3":
+                - 13
+                - 16
+                - 23
+                - 55
+                - 61
+            "4": 3.10.16
+            "5": 0.19.9
+            "6": 4.38.1
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+dataset:
+    value:
+        align_stage_components:
+            - download/llava-laion-cc-sbu-558k/chat.json
+            - download/llava-laion-cc-sbu-558k
+        dataset_id: llava-v15
+        dataset_root_dir: data
+        finetune_stage_components:
+            - download/llava-v1.5-instruct/llava_v1_5_mix665k.json
+            - download/llava-v1.5-instruct
+        type: llava-v15
+hf_token:
+    value: hf_token.txt
+model:
+    value:
+        align_epochs: 1
+        align_global_batch_size: 8
+        align_learning_rate: 0.001
+        align_lr_scheduler_type: linear-warmup+cosine-decay
+        align_max_grad_norm: 1
+        align_max_steps: null
+        align_per_device_batch_size: 8
+        align_train_strategy: fsdp-shard-grad-op
+        align_warmup_ratio: 0.03
+        align_weight_decay: 0
+        arch_specifier: no-align+fused-gelu-mlp
+        enable_gradient_checkpointing: true
+        enable_mixed_precision_training: true
+        finetune_epochs: 2
+        finetune_global_batch_size: 24
+        finetune_learning_rate: 2e-05
+        finetune_lr_scheduler_type: linear-warmup+cosine-decay
+        finetune_max_grad_norm: 1
+        finetune_max_steps: null
+        finetune_per_device_batch_size: 6
+        finetune_train_strategy: fsdp-full-shard
+        finetune_warmup_ratio: 0.03
+        finetune_weight_decay: 0.1
+        image_resize_strategy: resize-naive
+        llm_backbone_id: qwen25-1_5b-extra
+        llm_max_length: 32768
+        model_id: prism-qwen25-extra-dinosiglip-224px+1_5b
+        reduce_in_full_precision: false
+        type: prism-qwen25-extra-dinosiglip-224px+1_5b
+        vision_backbone_id: dinosiglip-vit-so-224px
+pretrained_checkpoint:
+    value: null
+run_id:
+    value: prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7
+run_root_dir:
+    value: runs
+seed:
+    value: 7
+stage:
+    value: finetune
+trackers:
+    value:
+        - jsonl
+        - wandb
+wandb_entity:
+    value: arash-akbari-stu-northeastern-university
+wandb_project:
+    value: Moxin-VLM

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/files/output.log

@@ -0,0 +1,2 @@
+05/02 [11:48:26] INFO     | >> [*] Starting Training Loop                                                                                                                                                                                          pretrain.py:238
+=>> [Global Step] 000112 =>> LR :: 0.000001 -- Loss :: 2.4465:   0%|          | 112/55442 [02:15<16:32:24,  1.08s/it]

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/files/requirements.txt

@@ -0,0 +1,78 @@
+pyyaml-include==1.4.1
+torchaudio==2.2.0
+psutil==7.0.0
+nvidia-cufft-cu12==11.0.2.54
+nvidia-nvtx-cu12==12.1.105
+wheel==0.45.1
+nvidia-cuda-cupti-cu12==12.1.105
+pillow==11.1.0
+draccus==0.10.0
+Jinja2==3.1.6
+ninja==1.11.1.4
+MarkupSafe==3.0.2
+nvidia-cublas-cu12==12.1.3.1
+tqdm==4.67.1
+fsspec==2025.3.2
+wandb==0.19.9
+GitPython==3.1.44
+click==8.1.8
+timm==0.9.10
+gitdb==4.0.12
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-cudnn-cu12==8.9.2.26
+mergedeep==1.3.4
+annotated-types==0.7.0
+protobuf==5.29.4
+huggingface-hub==0.30.2
+mdurl==0.1.2
+urllib3==2.3.0
+typing_extensions==4.13.1
+numpy==1.26.4
+torchvision==0.17.0
+nvidia-cusparse-cu12==12.1.0.106
+networkx==3.4.2
+regex==2024.11.6
+mypy-extensions==1.0.0
+pip==25.0
+sentencepiece==0.2.0
+Pygments==2.19.1
+smmap==5.0.2
+sympy==1.13.3
+setuptools==75.8.0
+nvidia-nccl-cu12==2.19.3
+charset-normalizer==3.4.1
+typing-inspection==0.4.0
+platformdirs==4.3.7
+packaging==24.2
+setproctitle==1.3.5
+idna==3.10
+markdown-it-py==3.0.0
+safetensors==0.5.3
+rich==14.0.0
+requests==2.32.3
+sentry-sdk==2.25.1
+jsonlines==4.0.0
+transformers==4.38.1
+PyYAML==6.0.2
+pydantic_core==2.33.1
+flash-attn==2.5.5
+mpmath==1.3.0
+attrs==25.3.0
+einops==0.8.1
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-curand-cu12==10.3.2.106
+filelock==3.18.0
+prismatic==0.0.2
+certifi==2025.1.31
+peft==0.5.0
+typing-inspect==0.9.0
+nvidia-cuda-nvrtc-cu12==12.1.105
+accelerate==0.25.0
+pydantic==2.11.3
+six==1.17.0
+nvidia-cusolver-cu12==11.4.5.107
+torch==2.2.0
+docker-pycreds==0.4.0
+tokenizers==0.15.2
+toml==0.10.2
+triton==2.2.0

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/files/wandb-metadata.json

@@ -0,0 +1,94 @@
+{
+  "os": "Linux-6.8.0-52-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.16",
+  "startedAt": "2025-05-02T18:48:24.510790Z",
+  "args": [
+    "--model.type",
+    "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "--wandb_project",
+    "Moxin-VLM",
+    "--wandb_entity",
+    "arash-akbari-stu-northeastern-university",
+    "--model.enable_mixed_precision_training",
+    "True"
+  ],
+  "program": "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py",
+  "codePath": "scripts/pretrain.py",
+  "git": {
+    "remote": "git@github.com:arashakb/prismatic-vlms.git",
+    "commit": "f94a585b2bb45d34a6947f2c80f2378f2af6ca66"
+  },
+  "email": "arash.akbari.stu@gmail.com",
+  "root": "runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7",
+  "host": "nnmc72",
+  "executable": "/home/user1/anaconda3/envs/prism/bin/python",
+  "codePathLocal": "scripts/pretrain.py",
+  "cpu_count": 48,
+  "cpu_count_logical": 96,
+  "gpu": "NVIDIA H100 NVL",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "30476149334016",
+      "used": "19451255332864"
+    }
+  },
+  "memory": {
+    "total": "811294752768"
+  },
+  "cpu": {
+    "count": 48,
+    "countLogical": 96
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    }
+  ],
+  "cudaVersion": "12.4"
+}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/files/wandb-summary.json

@@ -0,0 +1 @@
+{"Finetune/Learning Rate":1.3469633193024655e-06,"Finetune/Step":112,"Finetune/Loss (Raw)":2.335305690765381,"Finetune/Step Time":1.2350007040160043,"_step":112,"Finetune/Loss":2.4464573860168457,"_runtime":141.122340856,"_wandb":{"runtime":141},"_timestamp":1.7462118443747003e+09}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/logs/debug-core.log

@@ -0,0 +1,10 @@
+{"time":"2025-05-02T11:48:23.936394432-07:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpsxhf49n8/port-1932762.txt","pid":1932762,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false}
+{"time":"2025-05-02T11:48:23.938033658-07:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":1932762}
+{"time":"2025-05-02T11:48:23.938396597-07:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":36947,"Zone":""}}
+{"time":"2025-05-02T11:48:24.08241367-07:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:39034"}
+{"time":"2025-05-02T11:48:24.511951866-07:00","level":"INFO","msg":"handleInformInit: received","streamId":"6hhs8vrz","id":"127.0.0.1:39034"}
+{"time":"2025-05-02T11:48:25.628737745-07:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"6hhs8vrz","id":"127.0.0.1:39034"}
+{"time":"2025-05-02T11:50:45.632975375-07:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:39034"}
+{"time":"2025-05-02T11:50:45.633108663-07:00","level":"INFO","msg":"connection: closing","id":"127.0.0.1:39034"}
+{"time":"2025-05-02T11:50:45.63317393-07:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-05-02T11:50:45.633311554-07:00","level":"INFO","msg":"connection: closed successfully","id":"127.0.0.1:39034"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/logs/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2025-05-02T11:48:24.512248306-07:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/logs/debug-core.log"}
+{"time":"2025-05-02T11:48:25.628691797-07:00","level":"INFO","msg":"created new stream","id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.62873432-07:00","level":"INFO","msg":"stream: started","id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.62879471-07:00","level":"INFO","msg":"handler: started","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.628819016-07:00","level":"INFO","msg":"writer: Do: started","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.628835531-07:00","level":"INFO","msg":"sender: started","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:48:25.959126298-07:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2025-05-02T11:50:45.633111697-07:00","level":"INFO","msg":"stream: closing","id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:45.633155452-07:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2025-05-02T11:50:45.633187911-07:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2025-05-02T11:50:46.081032653-07:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-05-02T11:50:46.287259059-07:00","level":"INFO","msg":"handler: closed","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:46.287318307-07:00","level":"INFO","msg":"writer: Close: closed","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:46.287337876-07:00","level":"INFO","msg":"sender: closed","stream_id":"6hhs8vrz"}
+{"time":"2025-05-02T11:50:46.287531815-07:00","level":"INFO","msg":"stream: closed","id":"6hhs8vrz"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/logs/debug.log

@@ -0,0 +1,23 @@
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Configure stats pid to 1932762
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Loading settings from /home/user1/.config/wandb/settings
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Loading settings from /home/user1/arashwork/prismatic-vlms/wandb/settings
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:setup_run_log_directory():662] Logging user logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/logs/debug.log
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250502_114824-6hhs8vrz/logs/debug-internal.log
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():781] calling init triggers
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'type': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'model_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'arch_specifier': 'no-align+fused-gelu-mlp', 'vision_backbone_id': 'dinosiglip-vit-so-224px', 'llm_backbone_id': 'qwen25-1_5b-extra', 'image_resize_strategy': 'resize-naive', 'llm_max_length': 32768, 'align_epochs': 1, 'align_max_steps': None, 'align_global_batch_size': 8, 'align_per_device_batch_size': 8, 'align_learning_rate': 0.001, 'align_weight_decay': 0.0, 'align_max_grad_norm': 1.0, 'align_lr_scheduler_type': 'linear-warmup+cosine-decay', 'align_warmup_ratio': 0.03, 'align_train_strategy': 'fsdp-shard-grad-op', 'finetune_epochs': 2, 'finetune_max_steps': None, 'finetune_global_batch_size': 24, 'finetune_per_device_batch_size': 6, 'finetune_learning_rate': 2e-05, 'finetune_weight_decay': 0.1, 'finetune_max_grad_norm': 1.0, 'finetune_lr_scheduler_type': 'linear-warmup+cosine-decay', 'finetune_warmup_ratio': 0.03, 'finetune_train_strategy': 'fsdp-full-shard', 'enable_gradient_checkpointing': True, 'enable_mixed_precision_training': True, 'reduce_in_full_precision': False}, 'dataset': {'type': 'llava-v15', 'dataset_id': 'llava-v15', 'align_stage_components': ['download/llava-laion-cc-sbu-558k/chat.json', 'download/llava-laion-cc-sbu-558k'], 'finetune_stage_components': ['download/llava-v1.5-instruct/llava_v1_5_mix665k.json', 'download/llava-v1.5-instruct'], 'dataset_root_dir': 'data'}, 'stage': 'finetune', 'pretrained_checkpoint': None, 'run_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7', 'run_root_dir': 'runs', 'seed': 7, 'hf_token': 'hf_token.txt', 'trackers': ['jsonl', 'wandb'], 'wandb_project': 'Moxin-VLM', 'wandb_entity': 'arash-akbari-stu-northeastern-university', '_wandb': {}}
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():809] starting backend
+2025-05-02 11:48:24,503 INFO    MainThread:1932762 [wandb_init.py:init():813] sending inform_init request
+2025-05-02 11:48:24,510 INFO    MainThread:1932762 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-05-02 11:48:24,510 INFO    MainThread:1932762 [wandb_init.py:init():823] backend started and connected
+2025-05-02 11:48:24,512 INFO    MainThread:1932762 [wandb_init.py:init():915] updated telemetry
+2025-05-02 11:48:24,517 INFO    MainThread:1932762 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-05-02 11:48:25,957 INFO    MainThread:1932762 [wandb_init.py:init():1014] starting run threads in backend
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_console_start():2454] atexit reg
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-05-02 11:48:26,000 INFO    MainThread:1932762 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-05-02 11:48:26,002 INFO    MainThread:1932762 [wandb_init.py:init():1056] run started, returning control to user process
+2025-05-02 11:50:45,632 INFO    MsgRouterThr:1932762 [mailbox.py:close():129] [no run ID] Closing mailbox, abandoning 1 handles.

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/latest-run/run-6hhs8vrz.wandb

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:514635c836cd32f2106e2050edf14ff70f51a51daf2b1e5aa29387effe9be97f
+size 143258

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/files/config.yaml

@@ -0,0 +1,100 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m: []
+        python_version: 3.10.16
+        t:
+            "1":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "2":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "3":
+                - 13
+                - 16
+                - 23
+                - 55
+                - 61
+            "4": 3.10.16
+            "5": 0.19.9
+            "6": 4.51.1
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+dataset:
+    value:
+        align_stage_components:
+            - download/llava-laion-cc-sbu-558k/chat.json
+            - download/llava-laion-cc-sbu-558k
+        dataset_id: llava-v15
+        dataset_root_dir: data
+        finetune_stage_components:
+            - download/llava-v1.5-instruct/llava_v1_5_mix665k.json
+            - download/llava-v1.5-instruct
+        type: llava-v15
+hf_token:
+    value: hf_token.txt
+model:
+    value:
+        align_epochs: 1
+        align_global_batch_size: 8
+        align_learning_rate: 0.001
+        align_lr_scheduler_type: linear-warmup+cosine-decay
+        align_max_grad_norm: 1
+        align_max_steps: null
+        align_per_device_batch_size: 8
+        align_train_strategy: fsdp-shard-grad-op
+        align_warmup_ratio: 0.03
+        align_weight_decay: 0
+        arch_specifier: no-align+fused-gelu-mlp
+        enable_gradient_checkpointing: true
+        enable_mixed_precision_training: true
+        finetune_epochs: 2
+        finetune_global_batch_size: 2
+        finetune_learning_rate: 2e-05
+        finetune_lr_scheduler_type: linear-warmup+cosine-decay
+        finetune_max_grad_norm: 1
+        finetune_max_steps: null
+        finetune_per_device_batch_size: 2
+        finetune_train_strategy: fsdp-full-shard
+        finetune_warmup_ratio: 0.03
+        finetune_weight_decay: 0.1
+        image_resize_strategy: resize-naive
+        llm_backbone_id: qwen25-1_5b-extra
+        llm_max_length: 32768
+        model_id: prism-qwen25-extra-dinosiglip-224px+1_5b
+        reduce_in_full_precision: false
+        type: prism-qwen25-extra-dinosiglip-224px+1_5b
+        vision_backbone_id: dinosiglip-vit-so-224px
+pretrained_checkpoint:
+    value: null
+run_id:
+    value: prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7
+run_root_dir:
+    value: runs
+seed:
+    value: 7
+stage:
+    value: finetune
+trackers:
+    value:
+        - jsonl
+        - wandb
+wandb_entity:
+    value: arash-akbari-stu-northeastern-university
+wandb_project:
+    value: Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/files/output.log

@@ -0,0 +1,2 @@
+04/22 [09:52:04] INFO     | >> [*] Starting Training Loop                                                                                                                                    pretrain.py:238
+=>> [Global Step] 000015 =>> LR :: 0.000000 -- Loss :: 2.7283:   0%|          | 15/665298 [00:11<46:11:45,  4.00it/s] 

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/files/requirements.txt

@@ -0,0 +1,94 @@
+pyyaml-include==1.4.1
+torchaudio==2.2.0
+psutil==7.0.0
+nvidia-cufft-cu12==11.0.2.54
+nvidia-nvtx-cu12==12.1.105
+wheel==0.45.1
+nvidia-cuda-cupti-cu12==12.1.105
+pillow==11.1.0
+draccus==0.10.0
+Jinja2==3.1.6
+ninja==1.11.1.4
+MarkupSafe==3.0.2
+nvidia-cublas-cu12==12.1.3.1
+tqdm==4.67.1
+fsspec==2025.3.2
+wandb==0.19.9
+GitPython==3.1.44
+click==8.1.8
+timm==0.9.10
+gitdb==4.0.12
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-cudnn-cu12==8.9.2.26
+mergedeep==1.3.4
+annotated-types==0.7.0
+protobuf==5.29.4
+huggingface-hub==0.30.2
+mdurl==0.1.2
+urllib3==2.3.0
+typing_extensions==4.13.1
+numpy==1.26.4
+torchvision==0.17.0
+nvidia-cusparse-cu12==12.1.0.106
+networkx==3.4.2
+regex==2024.11.6
+mypy-extensions==1.0.0
+pip==25.0
+peft==0.15.1
+sentencepiece==0.2.0
+Pygments==2.19.1
+smmap==5.0.2
+sympy==1.13.3
+setuptools==75.8.0
+nvidia-nccl-cu12==2.19.3
+tokenizers==0.21.1
+charset-normalizer==3.4.1
+typing-inspection==0.4.0
+platformdirs==4.3.7
+packaging==24.2
+setproctitle==1.3.5
+idna==3.10
+markdown-it-py==3.0.0
+safetensors==0.5.3
+rich==14.0.0
+transformers==4.51.1
+requests==2.32.3
+sentry-sdk==2.25.1
+jsonlines==4.0.0
+PyYAML==6.0.2
+pydantic_core==2.33.1
+flash-attn==2.5.5
+mpmath==1.3.0
+attrs==25.3.0
+einops==0.8.1
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-curand-cu12==10.3.2.106
+filelock==3.18.0
+prismatic==0.0.2
+certifi==2025.1.31
+accelerate==1.6.0
+typing-inspect==0.9.0
+nvidia-cuda-nvrtc-cu12==12.1.105
+pydantic==2.11.3
+six==1.17.0
+nvidia-cusolver-cu12==11.4.5.107
+torch==2.2.0
+docker-pycreds==0.4.0
+toml==0.10.2
+triton==2.2.0
+importlib_metadata==8.0.0
+tomli==2.0.1
+zipp==3.19.2
+jaraco.context==5.3.0
+inflect==7.3.1
+autocommand==2.2.2
+typing_extensions==4.12.2
+jaraco.collections==5.1.0
+jaraco.functools==4.0.1
+packaging==24.2
+wheel==0.43.0
+backports.tarfile==1.2.0
+platformdirs==4.2.2
+more-itertools==10.3.0
+jaraco.text==3.12.1
+typeguard==4.3.0

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/files/wandb-metadata.json

@@ -0,0 +1,94 @@
+{
+  "os": "Linux-6.8.0-52-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.16",
+  "startedAt": "2025-04-22T16:52:03.056575Z",
+  "args": [
+    "--model.type",
+    "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "--wandb_project",
+    "Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune",
+    "--wandb_entity",
+    "arash-akbari-stu-northeastern-university",
+    "--model.enable_mixed_precision_training",
+    "True"
+  ],
+  "program": "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py",
+  "codePath": "scripts/pretrain.py",
+  "git": {
+    "remote": "git@github.com:arashakb/prismatic-vlms.git",
+    "commit": "8082f651fe3ba730b13899691a222fdef82d2c66"
+  },
+  "email": "arash.akbari.stu@gmail.com",
+  "root": "runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7",
+  "host": "nnmc72",
+  "executable": "/home/user1/anaconda3/envs/prism/bin/python",
+  "codePathLocal": "scripts/pretrain.py",
+  "cpu_count": 48,
+  "cpu_count_logical": 96,
+  "gpu": "NVIDIA H100 NVL",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "30476149334016",
+      "used": "18270204149760"
+    }
+  },
+  "memory": {
+    "total": "811294752768"
+  },
+  "cpu": {
+    "count": 48,
+    "countLogical": 96
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    }
+  ],
+  "cudaVersion": "12.4"
+}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/files/wandb-summary.json

@@ -0,0 +1 @@
+{"Finetune/Step Time":0.97449951171875,"Finetune/Loss":2.72833514213562,"_wandb":{"runtime":17},"Finetune/Learning Rate":1.5031566289207335e-08,"Finetune/Loss (Raw)":2.510840654373169,"_step":15,"_timestamp":1.7453407389854724e+09,"Finetune/Step":15,"_runtime":17.423786677}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/logs/debug-core.log

@@ -0,0 +1,10 @@
+{"time":"2025-04-22T09:52:02.705725804-07:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpmvl60rms/port-491736.txt","pid":491736,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false}
+{"time":"2025-04-22T09:52:02.706961119-07:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":491736}
+{"time":"2025-04-22T09:52:02.706929822-07:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":39925,"Zone":""}}
+{"time":"2025-04-22T09:52:02.895174529-07:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:55192"}
+{"time":"2025-04-22T09:52:03.059360916-07:00","level":"INFO","msg":"handleInformInit: received","streamId":"e4gyde5j","id":"127.0.0.1:55192"}
+{"time":"2025-04-22T09:52:03.956493803-07:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"e4gyde5j","id":"127.0.0.1:55192"}
+{"time":"2025-04-22T09:52:20.480226436-07:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:55192"}
+{"time":"2025-04-22T09:52:20.480361056-07:00","level":"INFO","msg":"connection: closing","id":"127.0.0.1:55192"}
+{"time":"2025-04-22T09:52:20.480421636-07:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-04-22T09:52:20.480583807-07:00","level":"INFO","msg":"connection: closed successfully","id":"127.0.0.1:55192"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/logs/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2025-04-22T09:52:03.059719599-07:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_095203-e4gyde5j/logs/debug-core.log"}
+{"time":"2025-04-22T09:52:03.956429127-07:00","level":"INFO","msg":"created new stream","id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:03.956484028-07:00","level":"INFO","msg":"stream: started","id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:03.956526962-07:00","level":"INFO","msg":"handler: started","stream_id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:03.956515976-07:00","level":"INFO","msg":"writer: Do: started","stream_id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:03.956572249-07:00","level":"INFO","msg":"sender: started","stream_id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:04.277054892-07:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2025-04-22T09:52:20.480344461-07:00","level":"INFO","msg":"stream: closing","id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:20.480372753-07:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2025-04-22T09:52:20.480413764-07:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2025-04-22T09:52:20.820531468-07:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-04-22T09:52:21.041731773-07:00","level":"INFO","msg":"handler: closed","stream_id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:21.041794947-07:00","level":"INFO","msg":"sender: closed","stream_id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:21.041797381-07:00","level":"INFO","msg":"writer: Close: closed","stream_id":"e4gyde5j"}
+{"time":"2025-04-22T09:52:21.042007493-07:00","level":"INFO","msg":"stream: closed","id":"e4gyde5j"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095203-e4gyde5j/logs/debug.log

@@ -0,0 +1,23 @@
+2025-04-22 09:52:03,050 INFO    MainThread:491736 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-04-22 09:52:03,050 INFO    MainThread:491736 [wandb_setup.py:_flush():67] Configure stats pid to 491736
+2025-04-22 09:52:03,050 INFO    MainThread:491736 [wandb_setup.py:_flush():67] Loading settings from /home/user1/.config/wandb/settings
+2025-04-22 09:52:03,050 INFO    MainThread:491736 [wandb_setup.py:_flush():67] Loading settings from /home/user1/arashwork/prismatic-vlms/wandb/settings
+2025-04-22 09:52:03,050 INFO    MainThread:491736 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-04-22 09:52:03,050 INFO    MainThread:491736 [wandb_init.py:setup_run_log_directory():662] Logging user logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_095203-e4gyde5j/logs/debug.log
+2025-04-22 09:52:03,051 INFO    MainThread:491736 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_095203-e4gyde5j/logs/debug-internal.log
+2025-04-22 09:52:03,051 INFO    MainThread:491736 [wandb_init.py:init():781] calling init triggers
+2025-04-22 09:52:03,051 INFO    MainThread:491736 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'type': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'model_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'arch_specifier': 'no-align+fused-gelu-mlp', 'vision_backbone_id': 'dinosiglip-vit-so-224px', 'llm_backbone_id': 'qwen25-1_5b-extra', 'image_resize_strategy': 'resize-naive', 'llm_max_length': 32768, 'align_epochs': 1, 'align_max_steps': None, 'align_global_batch_size': 8, 'align_per_device_batch_size': 8, 'align_learning_rate': 0.001, 'align_weight_decay': 0.0, 'align_max_grad_norm': 1.0, 'align_lr_scheduler_type': 'linear-warmup+cosine-decay', 'align_warmup_ratio': 0.03, 'align_train_strategy': 'fsdp-shard-grad-op', 'finetune_epochs': 2, 'finetune_max_steps': None, 'finetune_global_batch_size': 2, 'finetune_per_device_batch_size': 2, 'finetune_learning_rate': 2e-05, 'finetune_weight_decay': 0.1, 'finetune_max_grad_norm': 1.0, 'finetune_lr_scheduler_type': 'linear-warmup+cosine-decay', 'finetune_warmup_ratio': 0.03, 'finetune_train_strategy': 'fsdp-full-shard', 'enable_gradient_checkpointing': True, 'enable_mixed_precision_training': True, 'reduce_in_full_precision': False}, 'dataset': {'type': 'llava-v15', 'dataset_id': 'llava-v15', 'align_stage_components': ['download/llava-laion-cc-sbu-558k/chat.json', 'download/llava-laion-cc-sbu-558k'], 'finetune_stage_components': ['download/llava-v1.5-instruct/llava_v1_5_mix665k.json', 'download/llava-v1.5-instruct'], 'dataset_root_dir': 'data'}, 'stage': 'finetune', 'pretrained_checkpoint': None, 'run_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7', 'run_root_dir': 'runs', 'seed': 7, 'hf_token': 'hf_token.txt', 'trackers': ['jsonl', 'wandb'], 'wandb_project': 'Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune', 'wandb_entity': 'arash-akbari-stu-northeastern-university', '_wandb': {}}
+2025-04-22 09:52:03,051 INFO    MainThread:491736 [wandb_init.py:init():809] starting backend
+2025-04-22 09:52:03,051 INFO    MainThread:491736 [wandb_init.py:init():813] sending inform_init request
+2025-04-22 09:52:03,056 INFO    MainThread:491736 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-04-22 09:52:03,056 INFO    MainThread:491736 [wandb_init.py:init():823] backend started and connected
+2025-04-22 09:52:03,058 INFO    MainThread:491736 [wandb_init.py:init():915] updated telemetry
+2025-04-22 09:52:03,064 INFO    MainThread:491736 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-04-22 09:52:04,234 INFO    MainThread:491736 [wandb_init.py:init():1014] starting run threads in backend
+2025-04-22 09:52:04,313 INFO    MainThread:491736 [wandb_run.py:_console_start():2454] atexit reg
+2025-04-22 09:52:04,314 INFO    MainThread:491736 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-04-22 09:52:04,314 INFO    MainThread:491736 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-04-22 09:52:04,314 INFO    MainThread:491736 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-04-22 09:52:04,315 INFO    MainThread:491736 [wandb_init.py:init():1056] run started, returning control to user process
+2025-04-22 09:52:20,479 INFO    MsgRouterThr:491736 [mailbox.py:close():129] [no run ID] Closing mailbox, abandoning 1 handles.

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/files/config.yaml

@@ -0,0 +1,99 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m: []
+        python_version: 3.10.16
+        t:
+            "1":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "2":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "3":
+                - 13
+                - 16
+                - 23
+                - 55
+            "4": 3.10.16
+            "5": 0.19.9
+            "6": 4.51.1
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+dataset:
+    value:
+        align_stage_components:
+            - download/llava-laion-cc-sbu-558k/chat.json
+            - download/llava-laion-cc-sbu-558k
+        dataset_id: llava-v15
+        dataset_root_dir: data
+        finetune_stage_components:
+            - download/llava-v1.5-instruct/llava_v1_5_mix665k.json
+            - download/llava-v1.5-instruct
+        type: llava-v15
+hf_token:
+    value: hf_token.txt
+model:
+    value:
+        align_epochs: 1
+        align_global_batch_size: 8
+        align_learning_rate: 0.001
+        align_lr_scheduler_type: linear-warmup+cosine-decay
+        align_max_grad_norm: 1
+        align_max_steps: null
+        align_per_device_batch_size: 8
+        align_train_strategy: fsdp-shard-grad-op
+        align_warmup_ratio: 0.03
+        align_weight_decay: 0
+        arch_specifier: no-align+fused-gelu-mlp
+        enable_gradient_checkpointing: true
+        enable_mixed_precision_training: true
+        finetune_epochs: 2
+        finetune_global_batch_size: 128
+        finetune_learning_rate: 2e-05
+        finetune_lr_scheduler_type: linear-warmup+cosine-decay
+        finetune_max_grad_norm: 1
+        finetune_max_steps: null
+        finetune_per_device_batch_size: 16
+        finetune_train_strategy: fsdp-full-shard
+        finetune_warmup_ratio: 0.03
+        finetune_weight_decay: 0.1
+        image_resize_strategy: resize-naive
+        llm_backbone_id: qwen25-1_5b-extra
+        llm_max_length: 32768
+        model_id: prism-qwen25-extra-dinosiglip-224px+1_5b
+        reduce_in_full_precision: false
+        type: prism-qwen25-extra-dinosiglip-224px+1_5b
+        vision_backbone_id: dinosiglip-vit-so-224px
+pretrained_checkpoint:
+    value: null
+run_id:
+    value: prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7
+run_root_dir:
+    value: runs
+seed:
+    value: 7
+stage:
+    value: finetune
+trackers:
+    value:
+        - jsonl
+        - wandb
+wandb_entity:
+    value: arash-akbari-stu-northeastern-university
+wandb_project:
+    value: Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/files/output.log

@@ -0,0 +1,45 @@
+04/22 [09:58:06] INFO     | >> [*] Starting Training Loop                                                                                                                                    pretrain.py:238
+Traceback (most recent call last):                                                    
+  File "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py", line 252, in <module>
+    pretrain()
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/draccus/argparsing.py", line 225, in wrapper_inner
+    response = fn(cfg, *args, **kwargs)
+  File "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py", line 239, in pretrain
+    train_strategy.run_training(train_dataset, collator, metrics, stage=cfg.stage, seed=cfg.seed)
+  File "/home/user1/arashwork/prismatic-vlms/prismatic/training/strategies/base_strategy.py", line 183, in run_training
+    output: CausalLMOutputWithPast = self.vlm(
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/distributed/fsdp/fully_sharded_data_parallel.py", line 849, in forward
+    output = self._fsdp_wrapped_module(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/home/user1/arashwork/prismatic-vlms/prismatic/models/vlms/prismatic.py", line 410, in forward
+    return self.llm_backbone(
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/home/user1/arashwork/prismatic-vlms/prismatic/models/backbones/llm/base_llm.py", line 229, in forward
+    output: CausalLMOutputWithPast = self.llm(
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1511, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1520, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/transformers/utils/generic.py", line 965, in wrapper
+    output = func(self, *args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/transformers/utils/deprecation.py", line 172, in wrapped_func
+    return func(*args, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/transformers/models/qwen2/modeling_qwen2.py", line 843, in forward
+    loss = self.loss_function(logits=logits, labels=labels, vocab_size=self.config.vocab_size, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/transformers/loss/loss_utils.py", line 63, in ForCausalLMLoss
+    loss = fixed_cross_entropy(logits, shift_labels, num_items_in_batch, ignore_index, **kwargs)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/transformers/loss/loss_utils.py", line 35, in fixed_cross_entropy
+    loss = nn.functional.cross_entropy(source, target, ignore_index=ignore_index, reduction=reduction)
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/nn/functional.py", line 3059, in cross_entropy
+    return torch._C._nn.cross_entropy_loss(input, target, weight, _Reduction.get_enum(reduction), ignore_index, label_smoothing)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 49.14 GiB. GPU 0 has a total capacity of 93.00 GiB of which 6.57 GiB is free. Including non-PyTorch memory, this process has 86.41 GiB memory in use. Of the allocated memory 83.47 GiB is allocated by PyTorch, and 1.97 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/files/requirements.txt

@@ -0,0 +1,94 @@
+pyyaml-include==1.4.1
+torchaudio==2.2.0
+psutil==7.0.0
+nvidia-cufft-cu12==11.0.2.54
+nvidia-nvtx-cu12==12.1.105
+wheel==0.45.1
+nvidia-cuda-cupti-cu12==12.1.105
+pillow==11.1.0
+draccus==0.10.0
+Jinja2==3.1.6
+ninja==1.11.1.4
+MarkupSafe==3.0.2
+nvidia-cublas-cu12==12.1.3.1
+tqdm==4.67.1
+fsspec==2025.3.2
+wandb==0.19.9
+GitPython==3.1.44
+click==8.1.8
+timm==0.9.10
+gitdb==4.0.12
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-cudnn-cu12==8.9.2.26
+mergedeep==1.3.4
+annotated-types==0.7.0
+protobuf==5.29.4
+huggingface-hub==0.30.2
+mdurl==0.1.2
+urllib3==2.3.0
+typing_extensions==4.13.1
+numpy==1.26.4
+torchvision==0.17.0
+nvidia-cusparse-cu12==12.1.0.106
+networkx==3.4.2
+regex==2024.11.6
+mypy-extensions==1.0.0
+pip==25.0
+peft==0.15.1
+sentencepiece==0.2.0
+Pygments==2.19.1
+smmap==5.0.2
+sympy==1.13.3
+setuptools==75.8.0
+nvidia-nccl-cu12==2.19.3
+tokenizers==0.21.1
+charset-normalizer==3.4.1
+typing-inspection==0.4.0
+platformdirs==4.3.7
+packaging==24.2
+setproctitle==1.3.5
+idna==3.10
+markdown-it-py==3.0.0
+safetensors==0.5.3
+rich==14.0.0
+transformers==4.51.1
+requests==2.32.3
+sentry-sdk==2.25.1
+jsonlines==4.0.0
+PyYAML==6.0.2
+pydantic_core==2.33.1
+flash-attn==2.5.5
+mpmath==1.3.0
+attrs==25.3.0
+einops==0.8.1
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-curand-cu12==10.3.2.106
+filelock==3.18.0
+prismatic==0.0.2
+certifi==2025.1.31
+accelerate==1.6.0
+typing-inspect==0.9.0
+nvidia-cuda-nvrtc-cu12==12.1.105
+pydantic==2.11.3
+six==1.17.0
+nvidia-cusolver-cu12==11.4.5.107
+torch==2.2.0
+docker-pycreds==0.4.0
+toml==0.10.2
+triton==2.2.0
+importlib_metadata==8.0.0
+tomli==2.0.1
+zipp==3.19.2
+jaraco.context==5.3.0
+inflect==7.3.1
+autocommand==2.2.2
+typing_extensions==4.12.2
+jaraco.collections==5.1.0
+jaraco.functools==4.0.1
+packaging==24.2
+wheel==0.43.0
+backports.tarfile==1.2.0
+platformdirs==4.2.2
+more-itertools==10.3.0
+jaraco.text==3.12.1
+typeguard==4.3.0

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/files/wandb-metadata.json

@@ -0,0 +1,94 @@
+{
+  "os": "Linux-6.8.0-52-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.16",
+  "startedAt": "2025-04-22T16:58:05.167030Z",
+  "args": [
+    "--model.type",
+    "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "--wandb_project",
+    "Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune",
+    "--wandb_entity",
+    "arash-akbari-stu-northeastern-university",
+    "--model.enable_mixed_precision_training",
+    "True"
+  ],
+  "program": "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py",
+  "codePath": "scripts/pretrain.py",
+  "git": {
+    "remote": "git@github.com:arashakb/prismatic-vlms.git",
+    "commit": "8082f651fe3ba730b13899691a222fdef82d2c66"
+  },
+  "email": "arash.akbari.stu@gmail.com",
+  "root": "runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7",
+  "host": "nnmc72",
+  "executable": "/home/user1/anaconda3/envs/prism/bin/python",
+  "codePathLocal": "scripts/pretrain.py",
+  "cpu_count": 48,
+  "cpu_count_logical": 96,
+  "gpu": "NVIDIA H100 NVL",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "30476149334016",
+      "used": "18270204944384"
+    }
+  },
+  "memory": {
+    "total": "811294752768"
+  },
+  "cpu": {
+    "count": 48,
+    "countLogical": 96
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    }
+  ],
+  "cudaVersion": "12.4"
+}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_wandb":{"runtime":15}}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/logs/debug-core.log

@@ -0,0 +1,14 @@
+{"time":"2025-04-22T09:58:04.848042565-07:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmp7evbevo7/port-494110.txt","pid":494110,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false}
+{"time":"2025-04-22T09:58:04.849131413-07:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":494110}
+{"time":"2025-04-22T09:58:04.849105444-07:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":34993,"Zone":""}}
+{"time":"2025-04-22T09:58:05.037650239-07:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:05.168079456-07:00","level":"INFO","msg":"handleInformInit: received","streamId":"cpx6iuc8","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:05.747564085-07:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"cpx6iuc8","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.093776359-07:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.093854295-07:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-04-22T09:58:21.093848236-07:00","level":"INFO","msg":"connection: closing","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.094006501-07:00","level":"INFO","msg":"connection: closed successfully","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.1669656-07:00","level":"ERROR","msg":"processOutgoingData: flush error","error":"write tcp 127.0.0.1:34993->127.0.0.1:42808: use of closed network connection","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.728742764-07:00","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.728765407-07:00","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"127.0.0.1:42808"}
+{"time":"2025-04-22T09:58:21.728779829-07:00","level":"INFO","msg":"server is closed"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/logs/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2025-04-22T09:58:05.168284731-07:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_095805-cpx6iuc8/logs/debug-core.log"}
+{"time":"2025-04-22T09:58:05.74750679-07:00","level":"INFO","msg":"created new stream","id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:05.747559949-07:00","level":"INFO","msg":"stream: started","id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:05.74760112-07:00","level":"INFO","msg":"sender: started","stream_id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:05.747609823-07:00","level":"INFO","msg":"writer: Do: started","stream_id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:05.747646928-07:00","level":"INFO","msg":"handler: started","stream_id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:06.048552657-07:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2025-04-22T09:58:21.093879462-07:00","level":"INFO","msg":"stream: closing","id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:21.093897119-07:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2025-04-22T09:58:21.093946622-07:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2025-04-22T09:58:21.630469097-07:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-04-22T09:58:21.728396559-07:00","level":"INFO","msg":"handler: closed","stream_id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:21.728458532-07:00","level":"INFO","msg":"writer: Close: closed","stream_id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:21.728467014-07:00","level":"INFO","msg":"sender: closed","stream_id":"cpx6iuc8"}
+{"time":"2025-04-22T09:58:21.728627693-07:00","level":"INFO","msg":"stream: closed","id":"cpx6iuc8"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_095805-cpx6iuc8/logs/debug.log

@@ -0,0 +1,23 @@
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_setup.py:_flush():67] Configure stats pid to 494110
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_setup.py:_flush():67] Loading settings from /home/user1/.config/wandb/settings
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_setup.py:_flush():67] Loading settings from /home/user1/arashwork/prismatic-vlms/wandb/settings
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_init.py:setup_run_log_directory():662] Logging user logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_095805-cpx6iuc8/logs/debug.log
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_095805-cpx6iuc8/logs/debug-internal.log
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_init.py:init():781] calling init triggers
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'type': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'model_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'arch_specifier': 'no-align+fused-gelu-mlp', 'vision_backbone_id': 'dinosiglip-vit-so-224px', 'llm_backbone_id': 'qwen25-1_5b-extra', 'image_resize_strategy': 'resize-naive', 'llm_max_length': 32768, 'align_epochs': 1, 'align_max_steps': None, 'align_global_batch_size': 8, 'align_per_device_batch_size': 8, 'align_learning_rate': 0.001, 'align_weight_decay': 0.0, 'align_max_grad_norm': 1.0, 'align_lr_scheduler_type': 'linear-warmup+cosine-decay', 'align_warmup_ratio': 0.03, 'align_train_strategy': 'fsdp-shard-grad-op', 'finetune_epochs': 2, 'finetune_max_steps': None, 'finetune_global_batch_size': 128, 'finetune_per_device_batch_size': 16, 'finetune_learning_rate': 2e-05, 'finetune_weight_decay': 0.1, 'finetune_max_grad_norm': 1.0, 'finetune_lr_scheduler_type': 'linear-warmup+cosine-decay', 'finetune_warmup_ratio': 0.03, 'finetune_train_strategy': 'fsdp-full-shard', 'enable_gradient_checkpointing': True, 'enable_mixed_precision_training': True, 'reduce_in_full_precision': False}, 'dataset': {'type': 'llava-v15', 'dataset_id': 'llava-v15', 'align_stage_components': ['download/llava-laion-cc-sbu-558k/chat.json', 'download/llava-laion-cc-sbu-558k'], 'finetune_stage_components': ['download/llava-v1.5-instruct/llava_v1_5_mix665k.json', 'download/llava-v1.5-instruct'], 'dataset_root_dir': 'data'}, 'stage': 'finetune', 'pretrained_checkpoint': None, 'run_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7', 'run_root_dir': 'runs', 'seed': 7, 'hf_token': 'hf_token.txt', 'trackers': ['jsonl', 'wandb'], 'wandb_project': 'Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune', 'wandb_entity': 'arash-akbari-stu-northeastern-university', '_wandb': {}}
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_init.py:init():809] starting backend
+2025-04-22 09:58:05,162 INFO    MainThread:494110 [wandb_init.py:init():813] sending inform_init request
+2025-04-22 09:58:05,166 INFO    MainThread:494110 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-04-22 09:58:05,166 INFO    MainThread:494110 [wandb_init.py:init():823] backend started and connected
+2025-04-22 09:58:05,169 INFO    MainThread:494110 [wandb_init.py:init():915] updated telemetry
+2025-04-22 09:58:05,175 INFO    MainThread:494110 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-04-22 09:58:06,046 INFO    MainThread:494110 [wandb_init.py:init():1014] starting run threads in backend
+2025-04-22 09:58:06,086 INFO    MainThread:494110 [wandb_run.py:_console_start():2454] atexit reg
+2025-04-22 09:58:06,086 INFO    MainThread:494110 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-04-22 09:58:06,087 INFO    MainThread:494110 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-04-22 09:58:06,087 INFO    MainThread:494110 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-04-22 09:58:06,088 INFO    MainThread:494110 [wandb_init.py:init():1056] run started, returning control to user process
+2025-04-22 09:58:21,093 INFO    MsgRouterThr:494110 [mailbox.py:close():129] [no run ID] Closing mailbox, abandoning 2 handles.

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/files/config.yaml

@@ -0,0 +1,99 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m: []
+        python_version: 3.10.16
+        t:
+            "1":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "2":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "3":
+                - 13
+                - 16
+                - 23
+                - 55
+            "4": 3.10.16
+            "5": 0.19.9
+            "6": 4.51.1
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+dataset:
+    value:
+        align_stage_components:
+            - download/llava-laion-cc-sbu-558k/chat.json
+            - download/llava-laion-cc-sbu-558k
+        dataset_id: llava-v15
+        dataset_root_dir: data
+        finetune_stage_components:
+            - download/llava-v1.5-instruct/llava_v1_5_mix665k.json
+            - download/llava-v1.5-instruct
+        type: llava-v15
+hf_token:
+    value: hf_token.txt
+model:
+    value:
+        align_epochs: 1
+        align_global_batch_size: 8
+        align_learning_rate: 0.001
+        align_lr_scheduler_type: linear-warmup+cosine-decay
+        align_max_grad_norm: 1
+        align_max_steps: null
+        align_per_device_batch_size: 8
+        align_train_strategy: fsdp-shard-grad-op
+        align_warmup_ratio: 0.03
+        align_weight_decay: 0
+        arch_specifier: no-align+fused-gelu-mlp
+        enable_gradient_checkpointing: true
+        enable_mixed_precision_training: true
+        finetune_epochs: 2
+        finetune_global_batch_size: 64
+        finetune_learning_rate: 2e-05
+        finetune_lr_scheduler_type: linear-warmup+cosine-decay
+        finetune_max_grad_norm: 1
+        finetune_max_steps: null
+        finetune_per_device_batch_size: 8
+        finetune_train_strategy: fsdp-full-shard
+        finetune_warmup_ratio: 0.03
+        finetune_weight_decay: 0.1
+        image_resize_strategy: resize-naive
+        llm_backbone_id: qwen25-1_5b-extra
+        llm_max_length: 32768
+        model_id: prism-qwen25-extra-dinosiglip-224px+1_5b
+        reduce_in_full_precision: false
+        type: prism-qwen25-extra-dinosiglip-224px+1_5b
+        vision_backbone_id: dinosiglip-vit-so-224px
+pretrained_checkpoint:
+    value: null
+run_id:
+    value: prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7
+run_root_dir:
+    value: runs
+seed:
+    value: 7
+stage:
+    value: finetune
+trackers:
+    value:
+        - jsonl
+        - wandb
+wandb_entity:
+    value: arash-akbari-stu-northeastern-university
+wandb_project:
+    value: Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/files/output.log

@@ -0,0 +1,15 @@
+04/22 [10:00:10] INFO     | >> [*] Starting Training Loop                                                                                                                                    pretrain.py:238
+Traceback (most recent call last):                                                    
+  File "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py", line 252, in <module>
+    pretrain()
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/draccus/argparsing.py", line 225, in wrapper_inner
+    response = fn(cfg, *args, **kwargs)
+  File "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py", line 239, in pretrain
+    train_strategy.run_training(train_dataset, collator, metrics, stage=cfg.stage, seed=cfg.seed)
+  File "/home/user1/arashwork/prismatic-vlms/prismatic/training/strategies/base_strategy.py", line 208, in run_training
+    normalized_loss.backward()
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/_tensor.py", line 522, in backward
+    torch.autograd.backward(
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/autograd/__init__.py", line 266, in backward
+    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.57 GiB. GPU 0 has a total capacity of 93.00 GiB of which 23.65 GiB is free. Including non-PyTorch memory, this process has 69.33 GiB memory in use. Of the allocated memory 66.95 GiB is allocated by PyTorch, and 1.40 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/files/requirements.txt

@@ -0,0 +1,94 @@
+pyyaml-include==1.4.1
+torchaudio==2.2.0
+psutil==7.0.0
+nvidia-cufft-cu12==11.0.2.54
+nvidia-nvtx-cu12==12.1.105
+wheel==0.45.1
+nvidia-cuda-cupti-cu12==12.1.105
+pillow==11.1.0
+draccus==0.10.0
+Jinja2==3.1.6
+ninja==1.11.1.4
+MarkupSafe==3.0.2
+nvidia-cublas-cu12==12.1.3.1
+tqdm==4.67.1
+fsspec==2025.3.2
+wandb==0.19.9
+GitPython==3.1.44
+click==8.1.8
+timm==0.9.10
+gitdb==4.0.12
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-cudnn-cu12==8.9.2.26
+mergedeep==1.3.4
+annotated-types==0.7.0
+protobuf==5.29.4
+huggingface-hub==0.30.2
+mdurl==0.1.2
+urllib3==2.3.0
+typing_extensions==4.13.1
+numpy==1.26.4
+torchvision==0.17.0
+nvidia-cusparse-cu12==12.1.0.106
+networkx==3.4.2
+regex==2024.11.6
+mypy-extensions==1.0.0
+pip==25.0
+peft==0.15.1
+sentencepiece==0.2.0
+Pygments==2.19.1
+smmap==5.0.2
+sympy==1.13.3
+setuptools==75.8.0
+nvidia-nccl-cu12==2.19.3
+tokenizers==0.21.1
+charset-normalizer==3.4.1
+typing-inspection==0.4.0
+platformdirs==4.3.7
+packaging==24.2
+setproctitle==1.3.5
+idna==3.10
+markdown-it-py==3.0.0
+safetensors==0.5.3
+rich==14.0.0
+transformers==4.51.1
+requests==2.32.3
+sentry-sdk==2.25.1
+jsonlines==4.0.0
+PyYAML==6.0.2
+pydantic_core==2.33.1
+flash-attn==2.5.5
+mpmath==1.3.0
+attrs==25.3.0
+einops==0.8.1
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-curand-cu12==10.3.2.106
+filelock==3.18.0
+prismatic==0.0.2
+certifi==2025.1.31
+accelerate==1.6.0
+typing-inspect==0.9.0
+nvidia-cuda-nvrtc-cu12==12.1.105
+pydantic==2.11.3
+six==1.17.0
+nvidia-cusolver-cu12==11.4.5.107
+torch==2.2.0
+docker-pycreds==0.4.0
+toml==0.10.2
+triton==2.2.0
+importlib_metadata==8.0.0
+tomli==2.0.1
+zipp==3.19.2
+jaraco.context==5.3.0
+inflect==7.3.1
+autocommand==2.2.2
+typing_extensions==4.12.2
+jaraco.collections==5.1.0
+jaraco.functools==4.0.1
+packaging==24.2
+wheel==0.43.0
+backports.tarfile==1.2.0
+platformdirs==4.2.2
+more-itertools==10.3.0
+jaraco.text==3.12.1
+typeguard==4.3.0

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/files/wandb-metadata.json

@@ -0,0 +1,94 @@
+{
+  "os": "Linux-6.8.0-52-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.16",
+  "startedAt": "2025-04-22T17:00:09.283012Z",
+  "args": [
+    "--model.type",
+    "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "--wandb_project",
+    "Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune",
+    "--wandb_entity",
+    "arash-akbari-stu-northeastern-university",
+    "--model.enable_mixed_precision_training",
+    "True"
+  ],
+  "program": "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py",
+  "codePath": "scripts/pretrain.py",
+  "git": {
+    "remote": "git@github.com:arashakb/prismatic-vlms.git",
+    "commit": "8082f651fe3ba730b13899691a222fdef82d2c66"
+  },
+  "email": "arash.akbari.stu@gmail.com",
+  "root": "runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7",
+  "host": "nnmc72",
+  "executable": "/home/user1/anaconda3/envs/prism/bin/python",
+  "codePathLocal": "scripts/pretrain.py",
+  "cpu_count": 48,
+  "cpu_count_logical": 96,
+  "gpu": "NVIDIA H100 NVL",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "30476149334016",
+      "used": "18270205763584"
+    }
+  },
+  "memory": {
+    "total": "811294752768"
+  },
+  "cpu": {
+    "count": 48,
+    "countLogical": 96
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    }
+  ],
+  "cudaVersion": "12.4"
+}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_wandb":{"runtime":13}}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/logs/debug-core.log

@@ -0,0 +1,13 @@
+{"time":"2025-04-22T10:00:08.948681376-07:00","level":"INFO","msg":"main: starting server","port-filename":"/tmp/tmpki64cq_w/port-497136.txt","pid":497136,"log-level":0,"disable-analytics":false,"shutdown-on-parent-exit":false}
+{"time":"2025-04-22T10:00:08.949621573-07:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":497136}
+{"time":"2025-04-22T10:00:08.949626079-07:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":44241,"Zone":""}}
+{"time":"2025-04-22T10:00:09.137709691-07:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:09.285230582-07:00","level":"INFO","msg":"handleInformInit: received","streamId":"722cxxmu","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:09.9820411-07:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"722cxxmu","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:22.283734161-07:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:22.283869291-07:00","level":"INFO","msg":"server is shutting down"}
+{"time":"2025-04-22T10:00:22.283862411-07:00","level":"INFO","msg":"connection: closing","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:22.28400978-07:00","level":"INFO","msg":"connection: closed successfully","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:22.820172517-07:00","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:22.820205436-07:00","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"127.0.0.1:56432"}
+{"time":"2025-04-22T10:00:22.820226758-07:00","level":"INFO","msg":"server is closed"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/logs/debug-internal.log

@@ -0,0 +1,15 @@
+{"time":"2025-04-22T10:00:09.285532831-07:00","level":"INFO","msg":"stream: starting","core version":"0.19.9","symlink path":"runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_100009-722cxxmu/logs/debug-core.log"}
+{"time":"2025-04-22T10:00:09.98198035-07:00","level":"INFO","msg":"created new stream","id":"722cxxmu"}
+{"time":"2025-04-22T10:00:09.98203457-07:00","level":"INFO","msg":"stream: started","id":"722cxxmu"}
+{"time":"2025-04-22T10:00:09.982109321-07:00","level":"INFO","msg":"writer: Do: started","stream_id":"722cxxmu"}
+{"time":"2025-04-22T10:00:09.982130132-07:00","level":"INFO","msg":"handler: started","stream_id":"722cxxmu"}
+{"time":"2025-04-22T10:00:09.982171173-07:00","level":"INFO","msg":"sender: started","stream_id":"722cxxmu"}
+{"time":"2025-04-22T10:00:10.212439726-07:00","level":"INFO","msg":"Starting system monitor"}
+{"time":"2025-04-22T10:00:22.283892446-07:00","level":"INFO","msg":"stream: closing","id":"722cxxmu"}
+{"time":"2025-04-22T10:00:22.283943562-07:00","level":"INFO","msg":"Stopping system monitor"}
+{"time":"2025-04-22T10:00:22.284020005-07:00","level":"INFO","msg":"Stopped system monitor"}
+{"time":"2025-04-22T10:00:22.67777009-07:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"}
+{"time":"2025-04-22T10:00:22.819743269-07:00","level":"INFO","msg":"handler: closed","stream_id":"722cxxmu"}
+{"time":"2025-04-22T10:00:22.819852882-07:00","level":"INFO","msg":"sender: closed","stream_id":"722cxxmu"}
+{"time":"2025-04-22T10:00:22.819847424-07:00","level":"INFO","msg":"writer: Close: closed","stream_id":"722cxxmu"}
+{"time":"2025-04-22T10:00:22.819995704-07:00","level":"INFO","msg":"stream: closed","id":"722cxxmu"}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100009-722cxxmu/logs/debug.log

@@ -0,0 +1,23 @@
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_setup.py:_flush():67] Current SDK version is 0.19.9
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_setup.py:_flush():67] Configure stats pid to 497136
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_setup.py:_flush():67] Loading settings from /home/user1/.config/wandb/settings
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_setup.py:_flush():67] Loading settings from /home/user1/arashwork/prismatic-vlms/wandb/settings
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_setup.py:_flush():67] Loading settings from environment variables
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_init.py:setup_run_log_directory():662] Logging user logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_100009-722cxxmu/logs/debug.log
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_init.py:setup_run_log_directory():663] Logging internal logs to runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7/wandb/run-20250422_100009-722cxxmu/logs/debug-internal.log
+2025-04-22 10:00:09,276 INFO    MainThread:497136 [wandb_init.py:init():781] calling init triggers
+2025-04-22 10:00:09,277 INFO    MainThread:497136 [wandb_init.py:init():786] wandb.init called with sweep_config: {}
+config: {'model': {'type': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'model_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b', 'arch_specifier': 'no-align+fused-gelu-mlp', 'vision_backbone_id': 'dinosiglip-vit-so-224px', 'llm_backbone_id': 'qwen25-1_5b-extra', 'image_resize_strategy': 'resize-naive', 'llm_max_length': 32768, 'align_epochs': 1, 'align_max_steps': None, 'align_global_batch_size': 8, 'align_per_device_batch_size': 8, 'align_learning_rate': 0.001, 'align_weight_decay': 0.0, 'align_max_grad_norm': 1.0, 'align_lr_scheduler_type': 'linear-warmup+cosine-decay', 'align_warmup_ratio': 0.03, 'align_train_strategy': 'fsdp-shard-grad-op', 'finetune_epochs': 2, 'finetune_max_steps': None, 'finetune_global_batch_size': 64, 'finetune_per_device_batch_size': 8, 'finetune_learning_rate': 2e-05, 'finetune_weight_decay': 0.1, 'finetune_max_grad_norm': 1.0, 'finetune_lr_scheduler_type': 'linear-warmup+cosine-decay', 'finetune_warmup_ratio': 0.03, 'finetune_train_strategy': 'fsdp-full-shard', 'enable_gradient_checkpointing': True, 'enable_mixed_precision_training': True, 'reduce_in_full_precision': False}, 'dataset': {'type': 'llava-v15', 'dataset_id': 'llava-v15', 'align_stage_components': ['download/llava-laion-cc-sbu-558k/chat.json', 'download/llava-laion-cc-sbu-558k'], 'finetune_stage_components': ['download/llava-v1.5-instruct/llava_v1_5_mix665k.json', 'download/llava-v1.5-instruct'], 'dataset_root_dir': 'data'}, 'stage': 'finetune', 'pretrained_checkpoint': None, 'run_id': 'prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7', 'run_root_dir': 'runs', 'seed': 7, 'hf_token': 'hf_token.txt', 'trackers': ['jsonl', 'wandb'], 'wandb_project': 'Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune', 'wandb_entity': 'arash-akbari-stu-northeastern-university', '_wandb': {}}
+2025-04-22 10:00:09,277 INFO    MainThread:497136 [wandb_init.py:init():809] starting backend
+2025-04-22 10:00:09,277 INFO    MainThread:497136 [wandb_init.py:init():813] sending inform_init request
+2025-04-22 10:00:09,282 INFO    MainThread:497136 [backend.py:_multiprocessing_setup():101] multiprocessing start_methods=fork,spawn,forkserver, using: spawn
+2025-04-22 10:00:09,282 INFO    MainThread:497136 [wandb_init.py:init():823] backend started and connected
+2025-04-22 10:00:09,285 INFO    MainThread:497136 [wandb_init.py:init():915] updated telemetry
+2025-04-22 10:00:09,290 INFO    MainThread:497136 [wandb_init.py:init():939] communicating run to backend with 90.0 second timeout
+2025-04-22 10:00:10,209 INFO    MainThread:497136 [wandb_init.py:init():1014] starting run threads in backend
+2025-04-22 10:00:10,279 INFO    MainThread:497136 [wandb_run.py:_console_start():2454] atexit reg
+2025-04-22 10:00:10,279 INFO    MainThread:497136 [wandb_run.py:_redirect():2306] redirect: wrap_raw
+2025-04-22 10:00:10,280 INFO    MainThread:497136 [wandb_run.py:_redirect():2371] Wrapping output streams.
+2025-04-22 10:00:10,280 INFO    MainThread:497136 [wandb_run.py:_redirect():2394] Redirects installed.
+2025-04-22 10:00:10,281 INFO    MainThread:497136 [wandb_init.py:init():1056] run started, returning control to user process
+2025-04-22 10:00:22,283 INFO    MsgRouterThr:497136 [mailbox.py:close():129] [no run ID] Closing mailbox, abandoning 1 handles.

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100320-p7tc08z7/files/config.yaml

@@ -0,0 +1,99 @@
+_wandb:
+    value:
+        cli_version: 0.19.9
+        m: []
+        python_version: 3.10.16
+        t:
+            "1":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "2":
+                - 1
+                - 11
+                - 41
+                - 49
+                - 55
+                - 63
+                - 71
+                - 98
+            "3":
+                - 13
+                - 16
+                - 23
+                - 55
+            "4": 3.10.16
+            "5": 0.19.9
+            "6": 4.51.1
+            "8":
+                - 5
+            "12": 0.19.9
+            "13": linux-x86_64
+dataset:
+    value:
+        align_stage_components:
+            - download/llava-laion-cc-sbu-558k/chat.json
+            - download/llava-laion-cc-sbu-558k
+        dataset_id: llava-v15
+        dataset_root_dir: data
+        finetune_stage_components:
+            - download/llava-v1.5-instruct/llava_v1_5_mix665k.json
+            - download/llava-v1.5-instruct
+        type: llava-v15
+hf_token:
+    value: hf_token.txt
+model:
+    value:
+        align_epochs: 1
+        align_global_batch_size: 8
+        align_learning_rate: 0.001
+        align_lr_scheduler_type: linear-warmup+cosine-decay
+        align_max_grad_norm: 1
+        align_max_steps: null
+        align_per_device_batch_size: 8
+        align_train_strategy: fsdp-shard-grad-op
+        align_warmup_ratio: 0.03
+        align_weight_decay: 0
+        arch_specifier: no-align+fused-gelu-mlp
+        enable_gradient_checkpointing: true
+        enable_mixed_precision_training: true
+        finetune_epochs: 2
+        finetune_global_batch_size: 56
+        finetune_learning_rate: 2e-05
+        finetune_lr_scheduler_type: linear-warmup+cosine-decay
+        finetune_max_grad_norm: 1
+        finetune_max_steps: null
+        finetune_per_device_batch_size: 8
+        finetune_train_strategy: fsdp-full-shard
+        finetune_warmup_ratio: 0.03
+        finetune_weight_decay: 0.1
+        image_resize_strategy: resize-naive
+        llm_backbone_id: qwen25-1_5b-extra
+        llm_max_length: 32768
+        model_id: prism-qwen25-extra-dinosiglip-224px+1_5b
+        reduce_in_full_precision: false
+        type: prism-qwen25-extra-dinosiglip-224px+1_5b
+        vision_backbone_id: dinosiglip-vit-so-224px
+pretrained_checkpoint:
+    value: null
+run_id:
+    value: prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7
+run_root_dir:
+    value: runs
+seed:
+    value: 7
+stage:
+    value: finetune
+trackers:
+    value:
+        - jsonl
+        - wandb
+wandb_entity:
+    value: arash-akbari-stu-northeastern-university
+wandb_project:
+    value: Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100320-p7tc08z7/files/output.log

@@ -0,0 +1,15 @@
+04/22 [10:03:21] INFO     | >> [*] Starting Training Loop                                                                                                                                    pretrain.py:238
+Traceback (most recent call last):                                                    
+  File "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py", line 252, in <module>
+    pretrain()
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/draccus/argparsing.py", line 225, in wrapper_inner
+    response = fn(cfg, *args, **kwargs)
+  File "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py", line 239, in pretrain
+    train_strategy.run_training(train_dataset, collator, metrics, stage=cfg.stage, seed=cfg.seed)
+  File "/home/user1/arashwork/prismatic-vlms/prismatic/training/strategies/base_strategy.py", line 208, in run_training
+    normalized_loss.backward()
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/_tensor.py", line 522, in backward
+    torch.autograd.backward(
+  File "/home/user1/anaconda3/envs/prism/lib/python3.10/site-packages/torch/autograd/__init__.py", line 266, in backward
+    Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.57 GiB. GPU 0 has a total capacity of 93.00 GiB of which 23.56 GiB is free. Including non-PyTorch memory, this process has 69.42 GiB memory in use. Of the allocated memory 67.09 GiB is allocated by PyTorch, and 1.35 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation.  See documentation for Memory Management  (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100320-p7tc08z7/files/requirements.txt

@@ -0,0 +1,94 @@
+pyyaml-include==1.4.1
+torchaudio==2.2.0
+psutil==7.0.0
+nvidia-cufft-cu12==11.0.2.54
+nvidia-nvtx-cu12==12.1.105
+wheel==0.45.1
+nvidia-cuda-cupti-cu12==12.1.105
+pillow==11.1.0
+draccus==0.10.0
+Jinja2==3.1.6
+ninja==1.11.1.4
+MarkupSafe==3.0.2
+nvidia-cublas-cu12==12.1.3.1
+tqdm==4.67.1
+fsspec==2025.3.2
+wandb==0.19.9
+GitPython==3.1.44
+click==8.1.8
+timm==0.9.10
+gitdb==4.0.12
+nvidia-nvjitlink-cu12==12.8.93
+nvidia-cudnn-cu12==8.9.2.26
+mergedeep==1.3.4
+annotated-types==0.7.0
+protobuf==5.29.4
+huggingface-hub==0.30.2
+mdurl==0.1.2
+urllib3==2.3.0
+typing_extensions==4.13.1
+numpy==1.26.4
+torchvision==0.17.0
+nvidia-cusparse-cu12==12.1.0.106
+networkx==3.4.2
+regex==2024.11.6
+mypy-extensions==1.0.0
+pip==25.0
+peft==0.15.1
+sentencepiece==0.2.0
+Pygments==2.19.1
+smmap==5.0.2
+sympy==1.13.3
+setuptools==75.8.0
+nvidia-nccl-cu12==2.19.3
+tokenizers==0.21.1
+charset-normalizer==3.4.1
+typing-inspection==0.4.0
+platformdirs==4.3.7
+packaging==24.2
+setproctitle==1.3.5
+idna==3.10
+markdown-it-py==3.0.0
+safetensors==0.5.3
+rich==14.0.0
+transformers==4.51.1
+requests==2.32.3
+sentry-sdk==2.25.1
+jsonlines==4.0.0
+PyYAML==6.0.2
+pydantic_core==2.33.1
+flash-attn==2.5.5
+mpmath==1.3.0
+attrs==25.3.0
+einops==0.8.1
+nvidia-cuda-runtime-cu12==12.1.105
+nvidia-curand-cu12==10.3.2.106
+filelock==3.18.0
+prismatic==0.0.2
+certifi==2025.1.31
+accelerate==1.6.0
+typing-inspect==0.9.0
+nvidia-cuda-nvrtc-cu12==12.1.105
+pydantic==2.11.3
+six==1.17.0
+nvidia-cusolver-cu12==11.4.5.107
+torch==2.2.0
+docker-pycreds==0.4.0
+toml==0.10.2
+triton==2.2.0
+importlib_metadata==8.0.0
+tomli==2.0.1
+zipp==3.19.2
+jaraco.context==5.3.0
+inflect==7.3.1
+autocommand==2.2.2
+typing_extensions==4.12.2
+jaraco.collections==5.1.0
+jaraco.functools==4.0.1
+packaging==24.2
+wheel==0.43.0
+backports.tarfile==1.2.0
+platformdirs==4.2.2
+more-itertools==10.3.0
+jaraco.text==3.12.1
+typeguard==4.3.0

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100320-p7tc08z7/files/wandb-metadata.json

@@ -0,0 +1,94 @@
+{
+  "os": "Linux-6.8.0-52-generic-x86_64-with-glibc2.35",
+  "python": "CPython 3.10.16",
+  "startedAt": "2025-04-22T17:03:20.553675Z",
+  "args": [
+    "--model.type",
+    "prism-qwen25-extra-dinosiglip-224px+1_5b",
+    "--wandb_project",
+    "Qwen25-Extra-DINOSigLIP-224px-1_5B-full-finetune",
+    "--wandb_entity",
+    "arash-akbari-stu-northeastern-university",
+    "--model.enable_mixed_precision_training",
+    "True"
+  ],
+  "program": "/home/user1/arashwork/prismatic-vlms/scripts/pretrain.py",
+  "codePath": "scripts/pretrain.py",
+  "git": {
+    "remote": "git@github.com:arashakb/prismatic-vlms.git",
+    "commit": "8082f651fe3ba730b13899691a222fdef82d2c66"
+  },
+  "email": "arash.akbari.stu@gmail.com",
+  "root": "runs/prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7",
+  "host": "nnmc72",
+  "executable": "/home/user1/anaconda3/envs/prism/bin/python",
+  "codePathLocal": "scripts/pretrain.py",
+  "cpu_count": 48,
+  "cpu_count_logical": 96,
+  "gpu": "NVIDIA H100 NVL",
+  "gpu_count": 8,
+  "disk": {
+    "/": {
+      "total": "30476149334016",
+      "used": "18270206685184"
+    }
+  },
+  "memory": {
+    "total": "811294752768"
+  },
+  "cpu": {
+    "count": 48,
+    "countLogical": 96
+  },
+  "gpu_nvidia": [
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    },
+    {
+      "name": "NVIDIA H100 NVL",
+      "memoryTotal": "100485038080",
+      "cudaCores": 16896,
+      "architecture": "Hopper"
+    }
+  ],
+  "cudaVersion": "12.4"
+}

prism-qwen25-extra-dinosiglip-224px+1_5b+stage-finetune+x7--original_hyperparameters/wandb/run-20250422_100320-p7tc08z7/files/wandb-summary.json

@@ -0,0 +1 @@
+{"_wandb":{"runtime":11}}