Upload folder using huggingface_hub

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/checkpoints/latest-checkpoint.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:92cd948b279a905508cb483313e8b868afea238da49812dafc3094bf34283d59
+size 15336322587

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/checkpoints/latest-optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:324015c252de171b474d90a0b3892c50e9c90b49315a81b59018a647c292877c
+size 30672706842

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/checkpoints/latest-scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0ba4d4ded2fe863df5701c02896d2a60bb9e25903168be9a20a3a5a9fe9e6ba5
+size 1192

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/config.json

@@ -0,0 +1,74 @@
+{
+  "dataset": {
+    "align_stage_components": [
+      "download/llava-laion-cc-sbu-558k/chat.json",
+      "download/llava-laion-cc-sbu-558k"
+    ],
+    "dataset_id": "obelics",
+    "dataset_resampled": true,
+    "dataset_root_dir": "/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/datacomp/datacomp_hq_single_pkl_pil:/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/OBELICS/obelics_hq_35_single_pkl_pil:/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/synthdog-en/synthdog_single_pkl_pil/",
+    "finetune_stage_components": [
+      "/fsx-training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/llava/llava_v1_5_mix665k.json",
+      "/fsx-training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/llava/data"
+    ],
+    "max_num_images": 6,
+    "min_num_images": 1,
+    "train_num_samples": 2441408,
+    "type": "obelics",
+    "workers": 4
+  },
+  "model": {
+    "align_epochs": 1,
+    "align_global_batch_size": 256,
+    "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": 16,
+    "align_train_strategy": "fsdp-shard-grad-op",
+    "align_warmup_ratio": 0.03,
+    "align_weight_decay": 0.0,
+    "arch_specifier": "no-align+avgpool",
+    "enable_gradient_checkpointing": true,
+    "enable_mixed_precision_training": true,
+    "finetune_epochs": 1,
+    "finetune_global_batch_size": 128,
+    "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": 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": "phi3-3b",
+    "llm_max_length": 4096,
+    "model_id": "phi3-3b-continue-training-hq_35",
+    "pretrain_epochs": 1,
+    "pretrain_global_batch_size": 256,
+    "pretrain_learning_rate": 5e-05,
+    "pretrain_lr_scheduler_type": "linear-warmup+cosine-decay",
+    "pretrain_max_grad_norm": 1.0,
+    "pretrain_max_steps": 9536,
+    "pretrain_per_device_batch_size": 8,
+    "pretrain_train_strategy": "fsdp-full-shard",
+    "pretrain_warmup_ratio": 0.03,
+    "pretrain_weight_decay": 0.01,
+    "reduce_in_full_precision": false,
+    "type": "one-stage+7b",
+    "vision_backbone_id": "siglip-vit-so400m-384px"
+  },
+  "mount_path": "/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/models",
+  "pretrained_checkpoint": null,
+  "run_id": "obelics+phi3-3b-continue-training-hq_35+stage-pretrain+x7",
+  "run_root_dir": "/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/checkpoints/mllm_pretrain_checkpoints",
+  "seed": 7,
+  "stage": "pretrain",
+  "trackers": [
+    "jsonl",
+    "wandb"
+  ],
+  "wandb_entity": null,
+  "wandb_project": "mmpretrain"
+}

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/config.yaml

@@ -0,0 +1,67 @@
+dataset:
+  align_stage_components:
+  - download/llava-laion-cc-sbu-558k/chat.json
+  - download/llava-laion-cc-sbu-558k
+  dataset_id: obelics
+  dataset_resampled: true
+  dataset_root_dir: /fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/datacomp/datacomp_hq_single_pkl_pil:/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/OBELICS/obelics_hq_35_single_pkl_pil:/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/synthdog-en/synthdog_single_pkl_pil/
+  finetune_stage_components:
+  - /fsx-training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/llava/llava_v1_5_mix665k.json
+  - /fsx-training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/llava/data
+  max_num_images: 6
+  min_num_images: 1
+  train_num_samples: 2441408
+  type: obelics
+  workers: 4
+model:
+  align_epochs: 1
+  align_global_batch_size: 256
+  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: 16
+  align_train_strategy: fsdp-shard-grad-op
+  align_warmup_ratio: 0.03
+  align_weight_decay: 0.0
+  arch_specifier: no-align+avgpool
+  enable_gradient_checkpointing: true
+  enable_mixed_precision_training: true
+  finetune_epochs: 1
+  finetune_global_batch_size: 128
+  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: 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: phi3-3b
+  llm_max_length: 4096
+  model_id: phi3-3b-continue-training-hq_35
+  pretrain_epochs: 1
+  pretrain_global_batch_size: 256
+  pretrain_learning_rate: 5.0e-05
+  pretrain_lr_scheduler_type: linear-warmup+cosine-decay
+  pretrain_max_grad_norm: 1.0
+  pretrain_max_steps: 9536
+  pretrain_per_device_batch_size: 8
+  pretrain_train_strategy: fsdp-full-shard
+  pretrain_warmup_ratio: 0.03
+  pretrain_weight_decay: 0.01
+  reduce_in_full_precision: false
+  type: one-stage+7b
+  vision_backbone_id: siglip-vit-so400m-384px
+mount_path: /fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/models
+pretrained_checkpoint: null
+run_id: obelics+phi3-3b-continue-training-hq_35+stage-pretrain+x7
+run_root_dir: /fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/checkpoints/mllm_pretrain_checkpoints
+seed: 7
+stage: pretrain
+trackers:
+- jsonl
+- wandb
+wandb_entity: null
+wandb_project: mmpretrain

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/notes.txt

@@ -0,0 +1,3 @@
+we use the joint 280k caption and interleaved data of Claude-3.5 generation to train this classifier
+then we select top30%
+the endofchunk token is addded to the corpus

obelics+phi3-3b-continue-training-hq-35-run1+stage-pretrain+x7/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": "obelics", "dataset_resampled": true, "dataset_root_dir": "/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/datacomp/datacomp_hq_single_pkl_pil:/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/OBELICS/obelics_hq_35_single_pkl_pil:/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/synthdog-en/synthdog_single_pkl_pil/", "finetune_stage_components": ["/fsx-training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/llava/llava_v1_5_mix665k.json", "/fsx-training/shopqa-training-fsx-prod-us-east-1/home/wzwang/data/llava/data"], "max_num_images": 6, "min_num_images": 1, "train_num_samples": 2441408, "type": "obelics", "workers": 4}, "model": {"align_epochs": 1, "align_global_batch_size": 256, "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": 16, "align_train_strategy": "fsdp-shard-grad-op", "align_warmup_ratio": 0.03, "align_weight_decay": 0.0, "arch_specifier": "no-align+avgpool", "enable_gradient_checkpointing": true, "enable_mixed_precision_training": true, "finetune_epochs": 1, "finetune_global_batch_size": 128, "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": 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": "phi3-3b", "llm_max_length": 4096, "model_id": "phi3-3b-continue-training-hq_35", "pretrain_epochs": 1, "pretrain_global_batch_size": 256, "pretrain_learning_rate": 5e-05, "pretrain_lr_scheduler_type": "linear-warmup+cosine-decay", "pretrain_max_grad_norm": 1.0, "pretrain_max_steps": 9536, "pretrain_per_device_batch_size": 8, "pretrain_train_strategy": "fsdp-full-shard", "pretrain_warmup_ratio": 0.03, "pretrain_weight_decay": 0.01, "reduce_in_full_precision": false, "type": "one-stage+7b", "vision_backbone_id": "siglip-vit-so400m-384px"}, "mount_path": "/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/models", "pretrained_checkpoint": null, "run_id": "obelics+phi3-3b-continue-training-hq_35+stage-pretrain+x7", "run_root_dir": "/fsx-Training/shopqa-training-fsx-prod-us-east-1/home/wzwang/checkpoints/mllm_pretrain_checkpoints", "seed": 7, "stage": "pretrain", "trackers": ["jsonl", "wandb"], "wandb_entity": null, "wandb_project": "mmpretrain"}, "run_id": "obelics+phi3-3b-continue-training-hq_35+stage-pretrain+x7"}