Training in progress, step 50

hydra_config_1737997372.95926.yaml

@@ -0,0 +1,43 @@
+experiment:
+  seed: 42
+  name: debug
+  group: debug
+  dry_run: true
+  offline_run: false
+  evaluate_segmentation: false
+  evaluate_babyslm: false
+  blimp_tasks: null
+  resume_checkpoint_path: null
+  resume_run_id: null
+dataset:
+  name: phonemetransformers/CHILDES
+  subconfig: English
+  text_column: phonemized_utterance
+  is_phonemes: true
+  max_age: null
+  remove_child_utterances: true
+  valid_size: 10000
+tokenizer:
+  name: phonemetransformers/CHILDES-English-phoneme-tokenizer
+data_preprocessing:
+  max_input_length: 64
+  join_utts: dynamic
+  remove_word_boundaries: true
+  subsample: null
+  subsample_type: examples
+model:
+  name: gpt2_lm
+  model_kwargs:
+    n_layer: 2
+    n_head: 4
+    n_embd: 128
+    n_positions: 256
+    n_inner: 512
+trainer:
+  batch_size: 32
+  lr: 0.001
+  num_warmup_steps: 10
+  max_training_steps: 100
+  logging_steps: 1
+  save_steps: 50
+  eval_steps: 50

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1d9fc53ac7eb4187aa1a9e80c2a8d7d4fd61cafbaf6f1494cc6a41187660d879
+oid sha256:bcc2d127fa69dffa27ea42f968a43efd81c4817187ddb63daf6df4efe6af04e4
 size 1745032

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:cafec452906b73b08fa2554b45bc262cd2cb2f57e8a4b2a861ed87657b27d7f1
+oid sha256:47d68580ce5961d297755a206601e84fe8479b718eeee0d9ee11c150befb6423
 size 5432