Initial evaluation loss on 1k subset of HuggingFaceTB/cosmopedia-100k dataset was 1.182. 100 steps of LISA training reduced this to 1.117.
Comparison to control: cosmo-1b started out with 1.003 loss on (a different subset of) dataset, increasing to 1.024 at 100 steps.
See axolotl config
axolotl version:
0.4.0
base_model:HuggingFaceTB/cosmo-1bmodel_type:LlamaForCausalLMtokenizer_type:LlamaTokenizerload_in_8bit:falseload_in_4bit:falsestrict:falsedatasets:-path:Vezora/Tested-22k-Python-Alpacatype:alpacadataset_prepared_path:prepared-qloraval_set_size:0.05output_dir:./galore-outsequence_len:2048sample_packing:truepad_to_sequence_len:trueadapter:lora_model_dir:lora_r:lora_alpha:lora_dropout:lora_target_linear:lora_fan_in_fan_out:wandb_project:cosmo-python-galorewandb_entity:wandb_watch:wandb_name:wandb_log_model:gradient_accumulation_steps:4micro_batch_size:2num_epochs:1optimizer:galore_adamw_8bitlr_scheduler:cosinelearning_rate:0.0005optim_target_modules:-self_attn# for llama-mlpoptim_args:rank:256scale:1train_on_inputs:falsegroup_by_length:falsebf16:autofp16:tf32:falsegradient_checkpointing:trueearly_stopping_patience:resume_from_checkpoint:local_rank:logging_steps:1xformers_attention:flash_attention:truewarmup_steps:10evals_per_epoch:4saves_per_epoch:1debug:deepspeed:weight_decay:0.0fsdp:fsdp_config:special_tokens:
galore-out
This model is a fine-tuned version of
HuggingFaceTB/cosmo-1b
on the None dataset.
It achieves the following results on the evaluation set:
Loss: 0.2426
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
learning_rate: 0.0005
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 10
num_epochs: 1
Training results
Training Loss
Epoch
Step
Validation Loss
0.6299
0.0
1
0.6469
0.4353
0.25
215
0.4139
0.3721
0.5
430
0.2957
0.3514
0.75
645
0.2426
Framework versions
Transformers 4.40.0.dev0
Pytorch 2.1.2+cu118
Datasets 2.18.0
Tokenizers 0.15.0
Runs of Lambent cosmo-1b-galore-pythontest on huggingface.co
27
Total runs
0
24-hour runs
3
3-day runs
4
7-day runs
14
30-day runs
More Information About cosmo-1b-galore-pythontest huggingface.co Model
More cosmo-1b-galore-pythontest license Visit here:
cosmo-1b-galore-pythontest huggingface.co is an AI model on huggingface.co that provides cosmo-1b-galore-pythontest's model effect (), which can be used instantly with this Lambent cosmo-1b-galore-pythontest model. huggingface.co supports a free trial of the cosmo-1b-galore-pythontest model, and also provides paid use of the cosmo-1b-galore-pythontest. Support call cosmo-1b-galore-pythontest model through api, including Node.js, Python, http.
cosmo-1b-galore-pythontest huggingface.co is an online trial and call api platform, which integrates cosmo-1b-galore-pythontest's modeling effects, including api services, and provides a free online trial of cosmo-1b-galore-pythontest, you can try cosmo-1b-galore-pythontest online for free by clicking the link below.
Lambent cosmo-1b-galore-pythontest online free url in huggingface.co:
cosmo-1b-galore-pythontest is an open source model from GitHub that offers a free installation service, and any user can find cosmo-1b-galore-pythontest on GitHub to install. At the same time, huggingface.co provides the effect of cosmo-1b-galore-pythontest install, users can directly use cosmo-1b-galore-pythontest installed effect in huggingface.co for debugging and trial. It also supports api for free installation.
cosmo-1b-galore-pythontest install url in huggingface.co: