generated from allenai/python-package-template
-
Notifications
You must be signed in to change notification settings - Fork 397
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #27 from allenai/molmo
Molmo
- Loading branch information
Showing
10 changed files
with
3,274 additions
and
29 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,87 @@ | ||
| model: | ||
| name_or_path: allenai/Molmo-7B-O-0924 | ||
| arch: causal | ||
| use_flash_attn: true | ||
|
|
||
| wandb: | ||
| project: pdelfin | ||
| entity: ai2-llm | ||
|
|
||
| generate: | ||
| max_length: 8192 | ||
|
|
||
| train_data: | ||
| seed: 1337 | ||
| cache_location: /data/jakep/pdfdata/pdelfin_cache | ||
| sources: | ||
| - name: openai_batch_data_v5_1_train | ||
| response_glob_path: /data/jakep/pdfdata/openai_batch_data_v5_1_train_done/*.json | ||
| target_longest_image_dim: [1024] | ||
| target_anchor_text_len: [6000] | ||
| - name: openai_batch_data_v5_1_iabooks_train | ||
| response_glob_path: /data/jakep/pdfdata/openai_batch_data_v5_1_iabooks_train_done/*.json | ||
| target_longest_image_dim: [1024] | ||
| target_anchor_text_len: [6000] | ||
|
|
||
| valid_data: | ||
| cache_location: /data/jakep/pdfdata/pdelfin_cache | ||
| metric_for_best_model: openai_batch_data_v5_1_eval_loss | ||
| sources: | ||
| # These tend to be small, so you can load from s3 it's no big deal | ||
| - name: openai_batch_data_v5_1_eval | ||
| response_glob_path: s3://ai2-oe-data/jakep/pdfdata/openai_batch_done_v5_1_eval/*.json | ||
| target_longest_image_dim: [1024] | ||
| target_anchor_text_len: [6000] | ||
| - name: openai_batch_data_v5_1_iabooks_eval | ||
| response_glob_path: s3://ai2-oe-data/jakep/pdfdata/openai_batch_done_v5_1_iabooks_eval/*.json | ||
| target_longest_image_dim: [1024] | ||
| target_anchor_text_len: [6000] | ||
|
|
||
|
|
||
| # Mostly pulled from https://github.com/QwenLM/Qwen2/blob/main/examples/sft/finetune.sh | ||
| hparams: | ||
| batch_size: 1 | ||
| eval_batch_size: 1 | ||
| gradient_accumulation_steps: 4 | ||
| gradient_checkpointing: true | ||
| find_unused_parameters: true | ||
| clip_grad_norm: 1.0 | ||
| learning_rate: 3e-4 | ||
| max_steps: 10000 | ||
| pad_multiple_of: 16 | ||
| log_every_steps: 10 | ||
| eval_every_steps: 100 | ||
| optim: adamw_torch | ||
| lr_scheduler: cosine | ||
| weight_decay: 0.01 | ||
| warmup_ratio: 0.03 | ||
|
|
||
| # From https://github.com/QwenLM/Qwen2/blob/main/examples/sft/finetune.py | ||
| lora: | ||
| rank: 32 | ||
| alpha: 32 | ||
| dropout: 0.05 | ||
| task_type: CAUSAL_LM | ||
| target_modules: | ||
| # attention layers in main transformer | ||
| - att_proj | ||
| - ff_proj | ||
| - attn_out | ||
| - ff_out | ||
| # vision transformer attention and FF | ||
| - attention.wq | ||
| - attention.wk | ||
| - attention.wv | ||
| - attention.wo | ||
| - feed_forward.w1 | ||
| - feed_forward.w2 | ||
| # vision image projector | ||
| - vision_backbone.image_projector.w1 | ||
| - vision_backbone.image_projector.w2 | ||
| - vision_backbone.image_projector.w3 | ||
|
|
||
| save: | ||
| path: s3://ai2-oe-data/jakep/experiments/molmo-o-0924/v1/models/ | ||
| save_every_steps: 1000 | ||
|
|
||
| max_workers: 10 |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Empty file.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,60 @@ | ||
| from typing import List | ||
|
|
||
| from transformers import PretrainedConfig, AutoTokenizer | ||
|
|
||
|
|
||
| class MolmoConfig(PretrainedConfig): | ||
| model_type = "molmo" | ||
| keys_to_ignore_at_inference = ["past_key_values"] | ||
|
|
||
| def __init__( | ||
| self, | ||
| vocab_size=50304, | ||
| embedding_size=50304, | ||
| hidden_size=4096, | ||
| intermediate_size=11008, | ||
| num_hidden_layers=32, | ||
| num_attention_heads=32, | ||
| num_key_value_heads=None, | ||
| max_position_embeddings=2048, | ||
| initializer_range=0.02, | ||
| use_cache=True, | ||
| layer_norm_eps: float = 1e-5, | ||
| rope_theta=10000.0, | ||
| clip_qkv=None, | ||
| qkv_bias: bool = False, | ||
| weight_tying: bool = False, | ||
| use_position_ids: bool=True, | ||
| tie_word_embeddings: bool=True, | ||
| attention_layer_norm: bool=False, | ||
| norm_after: bool = False, | ||
| layer_norm_type: str="rms", | ||
| **kwargs, | ||
| ): | ||
| self.vocab_size = vocab_size | ||
| self.embedding_size = embedding_size | ||
| self.max_position_embeddings = max_position_embeddings | ||
| self.hidden_size = hidden_size | ||
| self.intermediate_size = intermediate_size | ||
| self.num_hidden_layers = num_hidden_layers | ||
| self.num_attention_heads = num_attention_heads | ||
| self.layer_norm_eps = layer_norm_eps | ||
| self.weight_tying = weight_tying | ||
| self.use_position_ids = use_position_ids | ||
| self.attention_layer_norm = attention_layer_norm | ||
| self.num_key_value_heads = num_key_value_heads | ||
| self.initializer_range = initializer_range | ||
| self.use_cache = use_cache | ||
| self.rope_theta = rope_theta | ||
| self.clip_qkv = clip_qkv | ||
| self.qkv_bias = qkv_bias | ||
| self.norm_after = norm_after | ||
| self.tie_word_embeddings = tie_word_embeddings | ||
| self.layer_norm_type = layer_norm_type | ||
|
|
||
| super().__init__( | ||
| tie_word_embeddings=tie_word_embeddings, | ||
| **kwargs, | ||
| ) | ||
|
|
||
| MolmoConfig.register_for_auto_class() |
Oops, something went wrong.