diff --git a/configs/cerebras_1_3B_alpaca.yml b/configs/cerebras_1_3B_alpaca.yml index d749ca72b..2c5534552 100644 --- a/configs/cerebras_1_3B_alpaca.yml +++ b/configs/cerebras_1_3B_alpaca.yml @@ -24,7 +24,7 @@ lora_fan_in_fan_out: false wandb_project: pythia-1.4b-lora wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-alpaca batch_size: 32 micro_batch_size: 4 diff --git a/configs/galactica_1_3B.yml b/configs/galactica_1_3B.yml index 1682849cf..cc1aa7ced 100644 --- a/configs/galactica_1_3B.yml +++ b/configs/galactica_1_3B.yml @@ -21,7 +21,7 @@ lora_fan_in_fan_out: false wandb_project: wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-llama-alpaca batch_size: 32 micro_batch_size: 16 diff --git a/configs/gpt_neox_20b.yml b/configs/gpt_neox_20b.yml index 91698ffaa..260b648b5 100644 --- a/configs/gpt_neox_20b.yml +++ b/configs/gpt_neox_20b.yml @@ -23,7 +23,7 @@ lora_fan_in_fan_out: true # pythia/GPTNeoX lora specific wandb_project: gpt4all-neox-20b wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./gpt4all-neox-20b batch_size: 48 micro_batch_size: 4 diff --git a/configs/llama_13B_alpaca.yml b/configs/llama_13B_alpaca.yml index 420a62209..3386f65c7 100644 --- a/configs/llama_13B_alpaca.yml +++ b/configs/llama_13B_alpaca.yml @@ -21,7 +21,7 @@ lora_fan_in_fan_out: false wandb_project: wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./llama-13b-sharegpt batch_size: 64 micro_batch_size: 2 diff --git a/configs/llama_65B_alpaca.yml b/configs/llama_65B_alpaca.yml index ab809defd..1c865626d 100644 --- a/configs/llama_65B_alpaca.yml +++ b/configs/llama_65B_alpaca.yml @@ -27,7 +27,7 @@ lora_fan_in_fan_out: false wandb_project: llama-65b-lora wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-llama-alpaca batch_size: 128 micro_batch_size: 16 diff --git a/configs/llama_7B_4bit.yml b/configs/llama_7B_4bit.yml index 0033a6e3a..feb4e21a1 100644 --- a/configs/llama_7B_4bit.yml +++ b/configs/llama_7B_4bit.yml @@ -24,7 +24,7 @@ lora_fan_in_fan_out: false wandb_project: wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-test batch_size: 8 micro_batch_size: 2 diff --git a/configs/llama_7B_alpaca.yml b/configs/llama_7B_alpaca.yml index 20efd58d3..66af807dd 100644 --- a/configs/llama_7B_alpaca.yml +++ b/configs/llama_7B_alpaca.yml @@ -26,7 +26,7 @@ lora_fan_in_fan_out: false wandb_project: llama-7b-lora wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-llama-alpaca batch_size: 128 micro_batch_size: 16 diff --git a/configs/llama_7B_jeopardy.yml b/configs/llama_7B_jeopardy.yml index f73bec348..4a20ddec6 100644 --- a/configs/llama_7B_jeopardy.yml +++ b/configs/llama_7B_jeopardy.yml @@ -22,7 +22,7 @@ lora_fan_in_fan_out: false wandb_project: jeopardy-bot-7b wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./jeopardy-bot-7b batch_size: 4 micro_batch_size: 1 diff --git a/configs/pythia_1_2B_alpaca.yml b/configs/pythia_1_2B_alpaca.yml index 5047be6bf..aa0587b90 100644 --- a/configs/pythia_1_2B_alpaca.yml +++ b/configs/pythia_1_2B_alpaca.yml @@ -26,7 +26,7 @@ lora_fan_in_fan_out: true # pythia/GPTNeoX lora specific wandb_project: pythia-1.4b-lora wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-alpaca batch_size: 48 micro_batch_size: 4 diff --git a/configs/quickstart.yml b/configs/quickstart.yml index f29df1478..a671c6e7f 100644 --- a/configs/quickstart.yml +++ b/configs/quickstart.yml @@ -24,7 +24,7 @@ lora_fan_in_fan_out: false wandb_project: wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-test batch_size: 4 micro_batch_size: 1 diff --git a/configs/sample.yml b/configs/sample.yml index 7d076c573..d8b7afa71 100644 --- a/configs/sample.yml +++ b/configs/sample.yml @@ -49,7 +49,7 @@ lora_fan_in_fan_out: false wandb_project: wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: # where to save the finsihed model to output_dir: ./completed-model # training hyperparameters diff --git a/configs/stability_3b.yml b/configs/stability_3b.yml index c5f2198d8..ecbd37980 100644 --- a/configs/stability_3b.yml +++ b/configs/stability_3b.yml @@ -20,7 +20,7 @@ lora_fan_in_fan_out: false wandb_project: stable-alpaca-3b wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./stable-alpaca-3b batch_size: 2 micro_batch_size: 1 diff --git a/configs/vicuna_13B_4bit_reflect.yml b/configs/vicuna_13B_4bit_reflect.yml index 872ce59ca..7ad409f26 100644 --- a/configs/vicuna_13B_4bit_reflect.yml +++ b/configs/vicuna_13B_4bit_reflect.yml @@ -28,7 +28,7 @@ lora_fan_in_fan_out: false wandb_project: wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./lora-reflect batch_size: 8 micro_batch_size: 2 diff --git a/examples/gptq-lora-7b/config.yml b/examples/gptq-lora-7b/config.yml index 0d57a6d8d..e676d967b 100644 --- a/examples/gptq-lora-7b/config.yml +++ b/examples/gptq-lora-7b/config.yml @@ -24,7 +24,7 @@ lora_fan_in_fan_out: false wandb_project: llama-7b-lora-int4 wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./llama-7b-lora-int4 batch_size: 1 micro_batch_size: 1 diff --git a/examples/mpt-7b/config.yml b/examples/mpt-7b/config.yml index f33452266..ca6a0d5c3 100644 --- a/examples/mpt-7b/config.yml +++ b/examples/mpt-7b/config.yml @@ -22,7 +22,7 @@ lora_fan_in_fan_out: false wandb_project: mpt-alpaca-7b wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./mpt-alpaca-7b batch_size: 1 micro_batch_size: 1 diff --git a/examples/redpajama/config-3b.yml b/examples/redpajama/config-3b.yml index 229d6615c..e7342b2f7 100644 --- a/examples/redpajama/config-3b.yml +++ b/examples/redpajama/config-3b.yml @@ -23,7 +23,7 @@ lora_fan_in_fan_out: false wandb_project: redpajama-alpaca-3b wandb_watch: wandb_run_id: -wandb_log_model: checkpoint +wandb_log_model: output_dir: ./redpajama-alpaca-3b batch_size: 4 micro_batch_size: 1