Spaces:

Dovakiins
/

qwerrwe

Build error

App Files Files Community

qwerrwe / src /axolotl /prompt_strategies

72.5 kB

100 contributors

History: 65 commits

winglian's picture

ORPO Trainer replacement (#1551)

7d1d22f unverified over 1 year ago

dpo
fixes for dpo and orpo template loading (#1424) over 1 year ago
orpo
ORPO Trainer replacement (#1551) over 1 year ago
__init__.py

926 Bytes

support custom field for completion from yml (#580) about 2 years ago
alpaca_chat.py

3.67 kB

fixes for alpaca w chatml, and don't include attention_mask w mistral for flash attention (#728) almost 2 years ago
alpaca_instruct.py

638 Bytes

skip the system prompt over 2 years ago
alpaca_w_system.py

5.49 kB

support user defined prompters, pretokenized datasets in config, local parquet, local arrow files (#348) about 2 years ago
base.py

564 Bytes

ORPO (#1419) over 1 year ago
chat_template.py

2.43 kB

add gemma instruct chat template (#1341) over 1 year ago
completion.py

2.54 kB

split completion text to sequence_len (#616) about 2 years ago
context_qa.py

3.16 kB

add support for defined train split (#654) almost 2 years ago
creative_acr.py

6.53 kB

Apply isort then black over 2 years ago
input_output.py

1.7 kB

plain input/output prompt strategy w/o chat templates (#1346) over 1 year ago
instruct.py

1.05 kB

relora: magnitude pruning of the optimizer (#1245) over 1 year ago
llama2_chat.py

8.17 kB

Respect sequence_len in config for `type: llama2_chat` (#926) almost 2 years ago
metharme.py

2.41 kB

feat: add Metharme prompt strategy (#446) about 2 years ago
orcamini.py

1.49 kB

Added Orca Mini prompt strategy (#263) about 2 years ago
pretrain.py

1.74 kB

fix pretraining_ on odd datasets (#1463) over 1 year ago
pygmalion.py

3.57 kB

Fixed pre-commit problems, fixed small bug in logging_config to handle LOG_LEVEL env var about 2 years ago
sharegpt.py

6.6 kB

Feat: Add sharegpt multirole (#1137) over 1 year ago
sharegpt_jokes.py

981 Bytes

use fastchat conversations template (#578) almost 2 years ago
user_defined.py

2.63 kB

support user defined prompters, pretokenized datasets in config, local parquet, local arrow files (#348) about 2 years ago