-
Update config.json
-
Using max_position_embeddings instead of max_sequence_length to standardise with HF
-
Update config.json
-
Update config.json
-
Fix causal mode
-
fixed output format
-
fix NaN issues
-
Update config.json
-
Update modeling_gptbert.py
typo fix -
Update modeling_gptbert.py
None bias and weight for the LayerNorms? -
Update config.json
is_decoder needs to be set always
-
Fixing small errors
Returning of attention probabilities is currently not handled.
Fixing error in naming of loss
-
make FlashAttention logic more robust
-
Upload model.safetensors with huggingface_hub
-
Update config.json
-
removed SDPA
-
Update modeling_gptbert.py
-
fixed SDPA for older PyTorch versions
-
Update config.json
-
Update README.md
-
Update config.json
-
Upload model_performance.png
-
FlashAttention support
-
FlashAttention support
-
FlashAttention support
-
Update modeling_gptbert.py
-
Update modeling_gptbert.py
-
Update modeling_gptbert.py
-
Update modeling_gptbert.py
-
Update modeling_gptbert.py
-
Update special_tokens_map.json
-
Update modeling_gptbert.py
-
Delete state_dict.bin
-
Upload folder using huggingface_hub
-
initial commit
