Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- KOREAN_MODEL_DESIGN.md +44 -0
- configs/cpt_fixed_ko_3b_5k_5090.yaml +35 -0
- configs/cpt_v2_3b_5k_5090.yaml +34 -0
- configs/model_50m.yaml +20 -0
- configs/model_50m_5k.yaml +19 -0
- configs/pretrain_100m_5090.yaml +32 -0
- configs/pretrain_5b.yaml +32 -0
- configs/pretrain_5b_5090.yaml +31 -0
- configs/pretrain_fixed_recovery_100m_5090.yaml +32 -0
- configs/pretrain_fixed_recovery_1b_5090.yaml +35 -0
- configs/pretrain_fixed_recovery_plus5b_5090.yaml +35 -0
- configs/pretrain_fixed_scratch_100m_5090.yaml +32 -0
- configs/pretrain_smoke.yaml +31 -0
- configs/pretrain_v2_100m_5090.yaml +31 -0
- configs/pretrain_v2_20b_5090.yaml +33 -0
- configs/pretrain_v2b_100m_5090.yaml +31 -0
- configs/pretrain_v2c_100m_5090.yaml +31 -0
- configs/sft_ko_80m_5k_5090.yaml +35 -0
- kawk50m/__init__.py +4 -0
- kawk50m/binary_dataset.py +147 -0
- kawk50m/chat.py +119 -0
- kawk50m/checkpoint.py +111 -0
- kawk50m/cli.py +12 -0
- kawk50m/config.py +77 -0
- kawk50m/io.py +144 -0
- kawk50m/sft_dataset.py +139 -0
- kawk50m/text.py +337 -0
- pyproject.toml +26 -0
- reports/fixed-ko-cpt-3b-5k/byte-evaluation.json +17 -0
- reports/fixed-ko-cpt-3b-5k/evaluation.json +30 -0
- reports/fixed-ko-cpt-3b-5k/final_checkpoint.txt +1 -0
- reports/fixed-ko-cpt-3b-5k/gate.json +7 -0
- reports/fixed-ko-cpt-3b-5k/position-bucket-evaluation.json +46 -0
- reports/fixed-ko-cpt-3b-5k/train.log +532 -0
- reports/fixed-recovery-6b/byte-evaluation.json +17 -0
- reports/fixed-recovery-6b/final_checkpoint.txt +1 -0
- reports/fixed-recovery-6b/train.log +423 -0
- reports/kawk50m-ko-instruct-v1/baseline-evaluation.json +56 -0
- reports/kawk50m-ko-instruct-v1/train.log +112 -0
- requirements-data.txt +5 -0
- requirements-dev.txt +3 -0
- requirements.txt +8 -0
- runpod/README.md +98 -0
- runpod/bootstrap.sh +31 -0
- runpod/download_5b_parallel.sh +79 -0
- runpod/prepare_5b_after_download.sh +55 -0
- runpod/stream_v2_20b.sh +123 -0
- runpod/tokenize_5b_parallel.sh +79 -0
- runpod/train_5090.sh +37 -0
- scripts/__init__.py +2 -0
KOREAN_MODEL_DESIGN.md
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# KAWK-50M 한국어 전용 설계
|
| 2 |
+
|
| 3 |
+
KAWK-50M은 Supra 계열의 규모와 BASE→장문맥 CPT 개발 순서만 참고한다.
|
| 4 |
+
영어 모델의 vocabulary, 층 배분, 데이터 비율 및 평가 방식을 복제하지 않는다.
|
| 5 |
+
|
| 6 |
+
## 모델과 토크나이저
|
| 7 |
+
|
| 8 |
+
- Llama 계열 decoder-only Transformer
|
| 9 |
+
- 51,542,528 parameters
|
| 10 |
+
- 한국어 SentencePiece Unigram vocabulary 20,000
|
| 11 |
+
- 14 layers, hidden size 512, SwiGLU intermediate size 1,408
|
| 12 |
+
- attention heads 8, KV heads 4
|
| 13 |
+
- tied input/output embeddings
|
| 14 |
+
- BASE context 1,024, long-context CPT 5,120
|
| 15 |
+
- NFC 완성형 한글과 byte fallback 사용
|
| 16 |
+
|
| 17 |
+
32K 영어 vocabulary 대신 한국어 20K vocabulary를 사용해 절약한 embedding
|
| 18 |
+
parameter를 14개 Transformer layer에 배분한다. 한국어 문서 안에 자연스럽게
|
| 19 |
+
등장하는 영문·숫자는 유지하지만 영어·코드·수학 전용 문서는 학습하지 않는다.
|
| 20 |
+
|
| 21 |
+
## 복구 BASE 학습
|
| 22 |
+
|
| 23 |
+
이전 23B 학습은 label 이중 시프트 오류가 있었으므로 총 학습량으로 계산하지
|
| 24 |
+
않는다. 올바른 `labels = input_ids` objective로 복구한 100M 체크포인트에서:
|
| 25 |
+
|
| 26 |
+
1. 한국어 일반 문서 1B 추가 학습
|
| 27 |
+
2. 겹치지 않는 sample 구간에서 한국어 일반 문서 5B 추가 학습
|
| 28 |
+
3. 각 단계에서 loss, UTF-8 byte loss, 한국어 생성 샘플 평가
|
| 29 |
+
|
| 30 |
+
각 단계가 이전 단계보다 개선되지 않으면 자동으로 다음 학습을 중단한다.
|
| 31 |
+
|
| 32 |
+
## 한국어 장문맥 CPT
|
| 33 |
+
|
| 34 |
+
- 총 3B tokens, sequence length 5,120
|
| 35 |
+
- 한국어 일반 BASE 문서 2.9B
|
| 36 |
+
- 한국어 RAG 문서 50M
|
| 37 |
+
- 한국어 대화 문서 50M
|
| 38 |
+
- ChatML 원문은 `지침`, `사용자`, `도우미` 역할명으로 평문화
|
| 39 |
+
- 영어·코드·수학·reasoning 전용 source 제외
|
| 40 |
+
|
| 41 |
+
RoPE는 언어별 장치가 아니다. KAWK에서는 1,024 이하 위치의 분포를 압축하지
|
| 42 |
+
않고 `max_position_embeddings=5120`, `rope_theta=10000`으로 전체 가중치 CPT를
|
| 43 |
+
수행한다. 최종 평가는 한국어 validation 문서에서 1,024-token 위치 구간별
|
| 44 |
+
loss를 측정해 뒤쪽 문맥의 성능 붕괴를 별도로 확인한다.
|
configs/cpt_fixed_ko_3b_5k_5090.yaml
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m_5k.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bins:
|
| 6 |
+
- data/tokenized-v2-cpt-3b/base/*.bin
|
| 7 |
+
- data/tokenized-v2-cpt-3b/structured/*.bin
|
| 8 |
+
validation_bin: data/tokenized-v2-cpt-3b/validation.bin
|
| 9 |
+
sequence_length: 5120
|
| 10 |
+
sample_offset: 0
|
| 11 |
+
|
| 12 |
+
training:
|
| 13 |
+
output_dir: runs/kawk50m-fixed-ko-cpt-3b-5k-5090
|
| 14 |
+
seed: 4242
|
| 15 |
+
precision: bf16
|
| 16 |
+
per_device_batch_size: 4
|
| 17 |
+
gradient_accumulation_steps: 6
|
| 18 |
+
max_tokens: 3000000000
|
| 19 |
+
max_steps: null
|
| 20 |
+
learning_rate: 1.5e-4
|
| 21 |
+
min_learning_rate: 1.5e-5
|
| 22 |
+
warmup_ratio: 0.02
|
| 23 |
+
weight_decay: 0.1
|
| 24 |
+
adam_beta1: 0.9
|
| 25 |
+
adam_beta2: 0.95
|
| 26 |
+
adam_eps: 1.0e-8
|
| 27 |
+
max_grad_norm: 1.0
|
| 28 |
+
log_every: 50
|
| 29 |
+
eval_every: 1000
|
| 30 |
+
eval_batches: 20
|
| 31 |
+
save_every: 5000
|
| 32 |
+
save_final: true
|
| 33 |
+
torch_compile: true
|
| 34 |
+
compile_mode: default
|
| 35 |
+
|
configs/cpt_v2_3b_5k_5090.yaml
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m_5k.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bins:
|
| 6 |
+
- data/tokenized-v2-cpt-3b/base/*.bin
|
| 7 |
+
- data/tokenized-v2-cpt-3b/structured/*.bin
|
| 8 |
+
validation_bin: data/tokenized-v2-cpt-3b/validation.bin
|
| 9 |
+
sequence_length: 5120
|
| 10 |
+
|
| 11 |
+
training:
|
| 12 |
+
output_dir: runs/kawk-1.5-50m-v2-cpt-3b-5090
|
| 13 |
+
seed: 4242
|
| 14 |
+
precision: bf16
|
| 15 |
+
per_device_batch_size: 4
|
| 16 |
+
gradient_accumulation_steps: 6
|
| 17 |
+
max_tokens: 3000000000
|
| 18 |
+
max_steps: null
|
| 19 |
+
learning_rate: 2.0e-4
|
| 20 |
+
min_learning_rate: 2.0e-5
|
| 21 |
+
warmup_ratio: 0.02
|
| 22 |
+
weight_decay: 0.1
|
| 23 |
+
adam_beta1: 0.9
|
| 24 |
+
adam_beta2: 0.95
|
| 25 |
+
adam_eps: 1.0e-8
|
| 26 |
+
max_grad_norm: 1.0
|
| 27 |
+
log_every: 10
|
| 28 |
+
eval_every: 250
|
| 29 |
+
eval_batches: 20
|
| 30 |
+
save_every: 1000
|
| 31 |
+
save_final: true
|
| 32 |
+
torch_compile: true
|
| 33 |
+
compile_mode: default
|
| 34 |
+
gradient_checkpointing: false
|
configs/model_50m.yaml
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model:
|
| 2 |
+
vocab_size: 20000
|
| 3 |
+
hidden_size: 512
|
| 4 |
+
intermediate_size: 1408
|
| 5 |
+
num_hidden_layers: 14
|
| 6 |
+
num_attention_heads: 8
|
| 7 |
+
num_key_value_heads: 4
|
| 8 |
+
max_position_embeddings: 1024
|
| 9 |
+
hidden_act: silu
|
| 10 |
+
rms_norm_eps: 1.0e-6
|
| 11 |
+
rope_theta: 10000.0
|
| 12 |
+
attention_bias: false
|
| 13 |
+
mlp_bias: false
|
| 14 |
+
tie_word_embeddings: true
|
| 15 |
+
bos_token_id: 1
|
| 16 |
+
eos_token_id: 2
|
| 17 |
+
pad_token_id: 3
|
| 18 |
+
initializer_range: 0.02
|
| 19 |
+
expected_parameters: 51542528
|
| 20 |
+
|
configs/model_50m_5k.yaml
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model:
|
| 2 |
+
vocab_size: 20000
|
| 3 |
+
hidden_size: 512
|
| 4 |
+
intermediate_size: 1408
|
| 5 |
+
num_hidden_layers: 14
|
| 6 |
+
num_attention_heads: 8
|
| 7 |
+
num_key_value_heads: 4
|
| 8 |
+
max_position_embeddings: 5120
|
| 9 |
+
hidden_act: silu
|
| 10 |
+
rms_norm_eps: 1.0e-6
|
| 11 |
+
rope_theta: 10000.0
|
| 12 |
+
attention_bias: false
|
| 13 |
+
mlp_bias: false
|
| 14 |
+
tie_word_embeddings: true
|
| 15 |
+
bos_token_id: 1
|
| 16 |
+
eos_token_id: 2
|
| 17 |
+
pad_token_id: 3
|
| 18 |
+
initializer_range: 0.02
|
| 19 |
+
expected_parameters: 51542528
|
configs/pretrain_100m_5090.yaml
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-pilot
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-pilot/train.bin
|
| 6 |
+
validation_bin: data/tokenized-pilot/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-100m-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 100000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 100
|
| 27 |
+
eval_batches: 20
|
| 28 |
+
save_every: 250
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
| 32 |
+
|
configs/pretrain_5b.yaml
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized/train.bin
|
| 6 |
+
validation_bin: data/tokenized/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-5b
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 8
|
| 14 |
+
gradient_accumulation_steps: 16
|
| 15 |
+
max_tokens: 5000000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 500
|
| 27 |
+
eval_batches: 50
|
| 28 |
+
save_every: 1000
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
| 32 |
+
|
configs/pretrain_5b_5090.yaml
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-pilot
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-5b/train.bin
|
| 6 |
+
validation_bin: data/tokenized-5b/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-5b-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 5000000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 500
|
| 27 |
+
eval_batches: 50
|
| 28 |
+
save_every: 1000
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
configs/pretrain_fixed_recovery_100m_5090.yaml
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-v2b-100m/train.bin
|
| 6 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-fixed-recovery-100m-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 100000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 100
|
| 27 |
+
eval_batches: 20
|
| 28 |
+
save_every: 250
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
| 32 |
+
|
configs/pretrain_fixed_recovery_1b_5090.yaml
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bins:
|
| 6 |
+
- data/tokenized-v2-20b/initial/*.bin
|
| 7 |
+
- data/tokenized-v2-20b/stream/part-*/*.bin
|
| 8 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 9 |
+
sequence_length: 1024
|
| 10 |
+
sample_offset: 0
|
| 11 |
+
|
| 12 |
+
training:
|
| 13 |
+
output_dir: runs/kawk50m-fixed-recovery-1b-5090
|
| 14 |
+
seed: 142
|
| 15 |
+
precision: bf16
|
| 16 |
+
per_device_batch_size: 32
|
| 17 |
+
gradient_accumulation_steps: 4
|
| 18 |
+
max_tokens: 1000000000
|
| 19 |
+
max_steps: null
|
| 20 |
+
learning_rate: 3.0e-4
|
| 21 |
+
min_learning_rate: 3.0e-5
|
| 22 |
+
warmup_ratio: 0.02
|
| 23 |
+
weight_decay: 0.1
|
| 24 |
+
adam_beta1: 0.9
|
| 25 |
+
adam_beta2: 0.95
|
| 26 |
+
adam_eps: 1.0e-8
|
| 27 |
+
max_grad_norm: 1.0
|
| 28 |
+
log_every: 50
|
| 29 |
+
eval_every: 500
|
| 30 |
+
eval_batches: 20
|
| 31 |
+
save_every: 2000
|
| 32 |
+
save_final: true
|
| 33 |
+
torch_compile: true
|
| 34 |
+
compile_mode: default
|
| 35 |
+
|
configs/pretrain_fixed_recovery_plus5b_5090.yaml
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bins:
|
| 6 |
+
- data/tokenized-v2-20b/initial/*.bin
|
| 7 |
+
- data/tokenized-v2-20b/stream/part-*/*.bin
|
| 8 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 9 |
+
sequence_length: 1024
|
| 10 |
+
sample_offset: 976640
|
| 11 |
+
|
| 12 |
+
training:
|
| 13 |
+
output_dir: runs/kawk50m-fixed-recovery-6b-5090
|
| 14 |
+
seed: 142
|
| 15 |
+
precision: bf16
|
| 16 |
+
per_device_batch_size: 32
|
| 17 |
+
gradient_accumulation_steps: 4
|
| 18 |
+
max_tokens: 5000000000
|
| 19 |
+
max_steps: null
|
| 20 |
+
learning_rate: 2.0e-4
|
| 21 |
+
min_learning_rate: 2.0e-5
|
| 22 |
+
warmup_ratio: 0.02
|
| 23 |
+
weight_decay: 0.1
|
| 24 |
+
adam_beta1: 0.9
|
| 25 |
+
adam_beta2: 0.95
|
| 26 |
+
adam_eps: 1.0e-8
|
| 27 |
+
max_grad_norm: 1.0
|
| 28 |
+
log_every: 100
|
| 29 |
+
eval_every: 2000
|
| 30 |
+
eval_batches: 20
|
| 31 |
+
save_every: 5000
|
| 32 |
+
save_final: true
|
| 33 |
+
torch_compile: true
|
| 34 |
+
compile_mode: default
|
| 35 |
+
|
configs/pretrain_fixed_scratch_100m_5090.yaml
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-v2b-100m/train.bin
|
| 6 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-fixed-scratch-100m-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 100000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 100
|
| 27 |
+
eval_batches: 20
|
| 28 |
+
save_every: 250
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
| 32 |
+
|
configs/pretrain_smoke.yaml
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized/train.bin
|
| 6 |
+
validation_bin: data/tokenized/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-smoke
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 2
|
| 14 |
+
gradient_accumulation_steps: 2
|
| 15 |
+
max_tokens: null
|
| 16 |
+
max_steps: 5
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 1
|
| 26 |
+
eval_every: 5
|
| 27 |
+
eval_batches: 2
|
| 28 |
+
save_every: 5
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: false
|
| 31 |
+
|
configs/pretrain_v2_100m_5090.yaml
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v2
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-v2-100m/train.bin
|
| 6 |
+
validation_bin: data/tokenized-v2-100m/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-v2-100m-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 100000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 100
|
| 27 |
+
eval_batches: 20
|
| 28 |
+
save_every: 250
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
configs/pretrain_v2_20b_5090.yaml
ADDED
|
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bins:
|
| 6 |
+
- data/tokenized-v2-20b/initial/*.bin
|
| 7 |
+
- data/tokenized-v2-20b/stream/part-*/*.bin
|
| 8 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 9 |
+
sequence_length: 1024
|
| 10 |
+
|
| 11 |
+
training:
|
| 12 |
+
output_dir: runs/kawk50m-v2-20b-5090
|
| 13 |
+
seed: 42
|
| 14 |
+
precision: bf16
|
| 15 |
+
per_device_batch_size: 32
|
| 16 |
+
gradient_accumulation_steps: 4
|
| 17 |
+
max_tokens: 20000000000
|
| 18 |
+
max_steps: null
|
| 19 |
+
learning_rate: 6.0e-4
|
| 20 |
+
min_learning_rate: 6.0e-5
|
| 21 |
+
warmup_ratio: 0.02
|
| 22 |
+
weight_decay: 0.1
|
| 23 |
+
adam_beta1: 0.9
|
| 24 |
+
adam_beta2: 0.95
|
| 25 |
+
adam_eps: 1.0e-8
|
| 26 |
+
max_grad_norm: 1.0
|
| 27 |
+
log_every: 20
|
| 28 |
+
eval_every: 1000
|
| 29 |
+
eval_batches: 20
|
| 30 |
+
save_every: 20000
|
| 31 |
+
save_final: true
|
| 32 |
+
torch_compile: true
|
| 33 |
+
compile_mode: default
|
configs/pretrain_v2b_100m_5090.yaml
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-v2b-100m/train.bin
|
| 6 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-v2b-100m-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 100000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 100
|
| 27 |
+
eval_batches: 20
|
| 28 |
+
save_every: 250
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
configs/pretrain_v2c_100m_5090.yaml
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_bin: data/tokenized-5b/train.bin
|
| 6 |
+
validation_bin: data/tokenized-v2b-100m/validation.bin
|
| 7 |
+
sequence_length: 1024
|
| 8 |
+
|
| 9 |
+
training:
|
| 10 |
+
output_dir: runs/kawk50m-v2c-100m-5090
|
| 11 |
+
seed: 42
|
| 12 |
+
precision: bf16
|
| 13 |
+
per_device_batch_size: 32
|
| 14 |
+
gradient_accumulation_steps: 4
|
| 15 |
+
max_tokens: 100000000
|
| 16 |
+
max_steps: null
|
| 17 |
+
learning_rate: 6.0e-4
|
| 18 |
+
min_learning_rate: 6.0e-5
|
| 19 |
+
warmup_ratio: 0.02
|
| 20 |
+
weight_decay: 0.1
|
| 21 |
+
adam_beta1: 0.9
|
| 22 |
+
adam_beta2: 0.95
|
| 23 |
+
adam_eps: 1.0e-8
|
| 24 |
+
max_grad_norm: 1.0
|
| 25 |
+
log_every: 10
|
| 26 |
+
eval_every: 100
|
| 27 |
+
eval_batches: 20
|
| 28 |
+
save_every: 250
|
| 29 |
+
save_final: true
|
| 30 |
+
torch_compile: true
|
| 31 |
+
compile_mode: default
|
configs/sft_ko_80m_5k_5090.yaml
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_config: configs/model_50m_5k.yaml
|
| 2 |
+
tokenizer_dir: artifacts/tokenizer-20k-v3
|
| 3 |
+
|
| 4 |
+
data:
|
| 5 |
+
train_prefixes:
|
| 6 |
+
- data/sft-ko-v1/chat/train
|
| 7 |
+
- data/sft-ko-v1/rag/train
|
| 8 |
+
validation_prefixes:
|
| 9 |
+
- data/sft-ko-v1/chat/validation
|
| 10 |
+
- data/sft-ko-v1/rag/validation
|
| 11 |
+
sequence_length: 5120
|
| 12 |
+
|
| 13 |
+
training:
|
| 14 |
+
output_dir: runs/kawk50m-ko-instruct-v1-5090
|
| 15 |
+
seed: 5150
|
| 16 |
+
precision: bf16
|
| 17 |
+
per_device_batch_size: 4
|
| 18 |
+
gradient_accumulation_steps: 6
|
| 19 |
+
epochs: 2.0
|
| 20 |
+
max_steps: null
|
| 21 |
+
learning_rate: 5.0e-5
|
| 22 |
+
min_learning_rate: 5.0e-6
|
| 23 |
+
warmup_ratio: 0.03
|
| 24 |
+
weight_decay: 0.01
|
| 25 |
+
adam_beta1: 0.9
|
| 26 |
+
adam_beta2: 0.95
|
| 27 |
+
adam_eps: 1.0e-8
|
| 28 |
+
max_grad_norm: 1.0
|
| 29 |
+
log_every: 20
|
| 30 |
+
eval_every: 200
|
| 31 |
+
eval_batches: 20
|
| 32 |
+
save_every: 500
|
| 33 |
+
save_final: true
|
| 34 |
+
torch_compile: true
|
| 35 |
+
compile_mode: default
|
kawk50m/__init__.py
ADDED
|
@@ -0,0 +1,4 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Utilities for training the KAWK50M Korean language model."""
|
| 2 |
+
|
| 3 |
+
__version__ = "0.1.0"
|
| 4 |
+
|
kawk50m/binary_dataset.py
ADDED
|
@@ -0,0 +1,147 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import math
|
| 4 |
+
import glob
|
| 5 |
+
from bisect import bisect_right
|
| 6 |
+
from collections.abc import Sequence
|
| 7 |
+
from pathlib import Path
|
| 8 |
+
|
| 9 |
+
import numpy as np
|
| 10 |
+
import torch
|
| 11 |
+
|
| 12 |
+
|
| 13 |
+
class PackedTokenDataset:
|
| 14 |
+
"""Read fixed-length Hugging Face causal-LM samples from uint16 streams.
|
| 15 |
+
|
| 16 |
+
Labels intentionally mirror ``input_ids`` because causal language-model
|
| 17 |
+
heads such as ``LlamaForCausalLM`` perform the one-token label shift inside
|
| 18 |
+
their loss function.
|
| 19 |
+
"""
|
| 20 |
+
|
| 21 |
+
def __init__(
|
| 22 |
+
self,
|
| 23 |
+
path: str | Path | Sequence[str | Path],
|
| 24 |
+
sequence_length: int,
|
| 25 |
+
) -> None:
|
| 26 |
+
if sequence_length <= 0:
|
| 27 |
+
raise ValueError("sequence_length must be positive")
|
| 28 |
+
raw_paths = [path] if isinstance(path, (str, Path)) else list(path)
|
| 29 |
+
resolved: set[Path] = set()
|
| 30 |
+
for raw_path in raw_paths:
|
| 31 |
+
value = str(raw_path)
|
| 32 |
+
matches = [Path(item) for item in glob.glob(value)]
|
| 33 |
+
if matches:
|
| 34 |
+
resolved.update(item.resolve() for item in matches if item.is_file())
|
| 35 |
+
else:
|
| 36 |
+
candidate = Path(raw_path)
|
| 37 |
+
if candidate.is_file():
|
| 38 |
+
resolved.add(candidate.resolve())
|
| 39 |
+
self.paths = sorted(resolved)
|
| 40 |
+
if not self.paths:
|
| 41 |
+
raise FileNotFoundError(f"No token binaries found for: {raw_paths}")
|
| 42 |
+
self.path = self.paths[0]
|
| 43 |
+
self.sequence_length = sequence_length
|
| 44 |
+
self.token_streams = [
|
| 45 |
+
np.memmap(item, mode="r", dtype=np.uint16) for item in self.paths
|
| 46 |
+
]
|
| 47 |
+
sequence_counts = [
|
| 48 |
+
max(0, (len(tokens) - 1) // self.sequence_length)
|
| 49 |
+
for tokens in self.token_streams
|
| 50 |
+
]
|
| 51 |
+
self.sequence_offsets: list[int] = []
|
| 52 |
+
running_total = 0
|
| 53 |
+
for count in sequence_counts:
|
| 54 |
+
running_total += count
|
| 55 |
+
self.sequence_offsets.append(running_total)
|
| 56 |
+
self.num_sequences = running_total
|
| 57 |
+
if self.num_sequences == 0:
|
| 58 |
+
raise ValueError(
|
| 59 |
+
f"{self.paths} need at least {self.sequence_length + 1} tokens"
|
| 60 |
+
)
|
| 61 |
+
self._affine_cache: dict[int, tuple[int, int]] = {}
|
| 62 |
+
|
| 63 |
+
def __len__(self) -> int:
|
| 64 |
+
return self.num_sequences
|
| 65 |
+
|
| 66 |
+
def sequence(self, index: int) -> np.ndarray:
|
| 67 |
+
if not 0 <= index < self.num_sequences:
|
| 68 |
+
raise IndexError(index)
|
| 69 |
+
stream_index = bisect_right(self.sequence_offsets, index)
|
| 70 |
+
previous_offset = (
|
| 71 |
+
self.sequence_offsets[stream_index - 1] if stream_index > 0 else 0
|
| 72 |
+
)
|
| 73 |
+
local_index = index - previous_offset
|
| 74 |
+
start = local_index * self.sequence_length
|
| 75 |
+
return np.asarray(
|
| 76 |
+
self.token_streams[stream_index][
|
| 77 |
+
start : start + self.sequence_length + 1
|
| 78 |
+
],
|
| 79 |
+
dtype=np.int64,
|
| 80 |
+
)
|
| 81 |
+
|
| 82 |
+
def _affine_parameters(self, epoch: int, seed: int) -> tuple[int, int]:
|
| 83 |
+
cache_key = (seed << 32) ^ epoch
|
| 84 |
+
if cache_key in self._affine_cache:
|
| 85 |
+
return self._affine_cache[cache_key]
|
| 86 |
+
rng = np.random.default_rng(seed + epoch * 1_000_003)
|
| 87 |
+
multiplier = int(rng.integers(1, max(self.num_sequences, 2)))
|
| 88 |
+
while math.gcd(multiplier, self.num_sequences) != 1:
|
| 89 |
+
multiplier += 1
|
| 90 |
+
if multiplier >= self.num_sequences:
|
| 91 |
+
multiplier = 1
|
| 92 |
+
offset = int(rng.integers(0, self.num_sequences))
|
| 93 |
+
self._affine_cache[cache_key] = (multiplier, offset)
|
| 94 |
+
return multiplier, offset
|
| 95 |
+
|
| 96 |
+
def training_batch(
|
| 97 |
+
self,
|
| 98 |
+
*,
|
| 99 |
+
micro_step: int,
|
| 100 |
+
batch_size: int,
|
| 101 |
+
rank: int,
|
| 102 |
+
world_size: int,
|
| 103 |
+
seed: int,
|
| 104 |
+
sample_offset: int = 0,
|
| 105 |
+
device: torch.device,
|
| 106 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 107 |
+
if sample_offset < 0:
|
| 108 |
+
raise ValueError("sample_offset must be non-negative")
|
| 109 |
+
global_batch_size = batch_size * world_size
|
| 110 |
+
first_ordinal = (
|
| 111 |
+
sample_offset
|
| 112 |
+
+ micro_step * global_batch_size
|
| 113 |
+
+ rank * batch_size
|
| 114 |
+
)
|
| 115 |
+
rows: list[np.ndarray] = []
|
| 116 |
+
for local_index in range(batch_size):
|
| 117 |
+
ordinal = first_ordinal + local_index
|
| 118 |
+
epoch, position = divmod(ordinal, self.num_sequences)
|
| 119 |
+
multiplier, offset = self._affine_parameters(epoch, seed)
|
| 120 |
+
sequence_index = (multiplier * position + offset) % self.num_sequences
|
| 121 |
+
rows.append(self.sequence(sequence_index))
|
| 122 |
+
return self._to_tensors(rows, device)
|
| 123 |
+
|
| 124 |
+
def evaluation_batch(
|
| 125 |
+
self,
|
| 126 |
+
*,
|
| 127 |
+
batch_index: int,
|
| 128 |
+
batch_size: int,
|
| 129 |
+
rank: int,
|
| 130 |
+
world_size: int,
|
| 131 |
+
device: torch.device,
|
| 132 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 133 |
+
first = (batch_index * world_size + rank) * batch_size
|
| 134 |
+
rows = [
|
| 135 |
+
self.sequence((first + offset) % self.num_sequences)
|
| 136 |
+
for offset in range(batch_size)
|
| 137 |
+
]
|
| 138 |
+
return self._to_tensors(rows, device)
|
| 139 |
+
|
| 140 |
+
@staticmethod
|
| 141 |
+
def _to_tensors(
|
| 142 |
+
rows: list[np.ndarray], device: torch.device
|
| 143 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 144 |
+
batch = torch.from_numpy(np.stack(rows, axis=0))
|
| 145 |
+
inputs = batch[:, :-1].to(device=device, dtype=torch.long, non_blocking=True)
|
| 146 |
+
labels = inputs.clone()
|
| 147 |
+
return inputs, labels
|
kawk50m/chat.py
ADDED
|
@@ -0,0 +1,119 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import re
|
| 4 |
+
from collections.abc import Iterable, Sequence
|
| 5 |
+
from typing import Any
|
| 6 |
+
|
| 7 |
+
|
| 8 |
+
CHATML_BLOCK_RE = re.compile(
|
| 9 |
+
r"<\|im_start\|>\s*(system|user|assistant)\s*\n?(.*?)<\|im_end\|>",
|
| 10 |
+
re.IGNORECASE | re.DOTALL,
|
| 11 |
+
)
|
| 12 |
+
|
| 13 |
+
ROLE_HEADERS = {
|
| 14 |
+
"system": "### 지침:\n",
|
| 15 |
+
"user": "### 사용자:\n",
|
| 16 |
+
"assistant": "### 도우미:\n",
|
| 17 |
+
}
|
| 18 |
+
|
| 19 |
+
KOREAN_CHAT_TEMPLATE = """{{ bos_token }}{% for message in messages %}{% if message['role'] == 'system' %}### 지침:
|
| 20 |
+
{{ message['content'] }}
|
| 21 |
+
|
| 22 |
+
{% elif message['role'] == 'user' %}### 사용자:
|
| 23 |
+
{{ message['content'] }}
|
| 24 |
+
|
| 25 |
+
{% elif message['role'] == 'assistant' %}### 도우미:
|
| 26 |
+
{{ message['content'] }}{{ eos_token }}
|
| 27 |
+
|
| 28 |
+
{% endif %}{% endfor %}{% if add_generation_prompt %}### 도우미:
|
| 29 |
+
{% endif %}"""
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
def parse_chatml(text: str) -> list[tuple[str, str]]:
|
| 33 |
+
if not isinstance(text, str):
|
| 34 |
+
return []
|
| 35 |
+
return [
|
| 36 |
+
(role.casefold(), content.strip())
|
| 37 |
+
for role, content in CHATML_BLOCK_RE.findall(text)
|
| 38 |
+
if content.strip()
|
| 39 |
+
]
|
| 40 |
+
|
| 41 |
+
|
| 42 |
+
def format_generation_prompt(
|
| 43 |
+
user_text: str,
|
| 44 |
+
*,
|
| 45 |
+
system_text: str | None = None,
|
| 46 |
+
) -> str:
|
| 47 |
+
sections: list[str] = []
|
| 48 |
+
if system_text:
|
| 49 |
+
sections.append(ROLE_HEADERS["system"] + system_text.strip())
|
| 50 |
+
sections.append(ROLE_HEADERS["user"] + user_text.strip())
|
| 51 |
+
sections.append(ROLE_HEADERS["assistant"])
|
| 52 |
+
return "\n\n".join(sections)
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
def _encode(tokenizer: Any, text: str) -> list[int]:
|
| 56 |
+
return list(
|
| 57 |
+
tokenizer(
|
| 58 |
+
text,
|
| 59 |
+
add_special_tokens=False,
|
| 60 |
+
return_attention_mask=False,
|
| 61 |
+
return_token_type_ids=False,
|
| 62 |
+
)["input_ids"]
|
| 63 |
+
)
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
def _encode_context_block(
|
| 67 |
+
tokenizer: Any,
|
| 68 |
+
role: str,
|
| 69 |
+
content: str,
|
| 70 |
+
) -> list[int]:
|
| 71 |
+
result = _encode(tokenizer, ROLE_HEADERS[role] + content.strip())
|
| 72 |
+
if role == "assistant":
|
| 73 |
+
result.append(int(tokenizer.eos_token_id))
|
| 74 |
+
result.extend(_encode(tokenizer, "\n\n"))
|
| 75 |
+
return result
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
def build_assistant_examples(
|
| 79 |
+
blocks: Sequence[tuple[str, str]],
|
| 80 |
+
tokenizer: Any,
|
| 81 |
+
*,
|
| 82 |
+
max_length: int,
|
| 83 |
+
min_prompt_tokens: int = 32,
|
| 84 |
+
) -> Iterable[tuple[list[int], list[int]]]:
|
| 85 |
+
"""Yield one assistant-only-loss example for every assistant turn.
|
| 86 |
+
|
| 87 |
+
Earlier assistant turns remain masked context. The current assistant response
|
| 88 |
+
and its EOS token are supervised. Long prompts are cropped from the left so
|
| 89 |
+
the most recent user request and assistant header are retained.
|
| 90 |
+
"""
|
| 91 |
+
|
| 92 |
+
if max_length <= min_prompt_tokens + 2:
|
| 93 |
+
raise ValueError("max_length is too small for an SFT example")
|
| 94 |
+
if tokenizer.bos_token_id is None or tokenizer.eos_token_id is None:
|
| 95 |
+
raise ValueError("Tokenizer must define BOS and EOS token IDs")
|
| 96 |
+
|
| 97 |
+
context_ids = [int(tokenizer.bos_token_id)]
|
| 98 |
+
for role, content in blocks:
|
| 99 |
+
if role not in ROLE_HEADERS or not content.strip():
|
| 100 |
+
continue
|
| 101 |
+
if role != "assistant":
|
| 102 |
+
context_ids.extend(_encode_context_block(tokenizer, role, content))
|
| 103 |
+
continue
|
| 104 |
+
|
| 105 |
+
assistant_header = _encode(tokenizer, ROLE_HEADERS["assistant"])
|
| 106 |
+
prompt_ids = context_ids + assistant_header
|
| 107 |
+
response_ids = _encode(tokenizer, content.strip())
|
| 108 |
+
response_budget = max_length - min_prompt_tokens - 1
|
| 109 |
+
if len(response_ids) > response_budget:
|
| 110 |
+
response_ids = response_ids[:response_budget]
|
| 111 |
+
target_ids = response_ids + [int(tokenizer.eos_token_id)]
|
| 112 |
+
prompt_budget = max_length - len(target_ids)
|
| 113 |
+
prompt_tail = prompt_ids[-prompt_budget:]
|
| 114 |
+
input_ids = prompt_tail + target_ids
|
| 115 |
+
loss_mask = [0] * len(prompt_tail) + [1] * len(target_ids)
|
| 116 |
+
if response_ids and len(input_ids) <= max_length:
|
| 117 |
+
yield input_ids, loss_mask
|
| 118 |
+
|
| 119 |
+
context_ids.extend(_encode_context_block(tokenizer, role, content))
|
kawk50m/checkpoint.py
ADDED
|
@@ -0,0 +1,111 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import json
|
| 4 |
+
import os
|
| 5 |
+
import random
|
| 6 |
+
import uuid
|
| 7 |
+
from pathlib import Path
|
| 8 |
+
from typing import Any
|
| 9 |
+
|
| 10 |
+
import numpy as np
|
| 11 |
+
import torch
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
def resolve_checkpoint(resume: str | Path, output_dir: str | Path) -> Path:
|
| 15 |
+
if str(resume).lower() != "latest":
|
| 16 |
+
path = Path(resume)
|
| 17 |
+
else:
|
| 18 |
+
latest_file = Path(output_dir) / "latest_checkpoint.txt"
|
| 19 |
+
if not latest_file.is_file():
|
| 20 |
+
raise FileNotFoundError(
|
| 21 |
+
f"Cannot resume latest: {latest_file} does not exist"
|
| 22 |
+
)
|
| 23 |
+
stored = latest_file.read_text(encoding="utf-8").strip()
|
| 24 |
+
path = Path(stored)
|
| 25 |
+
if not path.is_absolute():
|
| 26 |
+
path = Path(output_dir) / path
|
| 27 |
+
if not (path / "model").is_dir() or not (path / "trainer_state.pt").is_file():
|
| 28 |
+
raise FileNotFoundError(f"Invalid checkpoint directory: {path}")
|
| 29 |
+
return path.resolve()
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
def capture_rng_state() -> dict[str, Any]:
|
| 33 |
+
state: dict[str, Any] = {
|
| 34 |
+
"python": random.getstate(),
|
| 35 |
+
"numpy": np.random.get_state(),
|
| 36 |
+
"torch": torch.get_rng_state(),
|
| 37 |
+
}
|
| 38 |
+
if torch.cuda.is_available():
|
| 39 |
+
state["cuda"] = torch.cuda.get_rng_state_all()
|
| 40 |
+
return state
|
| 41 |
+
|
| 42 |
+
|
| 43 |
+
def restore_rng_state(state: dict[str, Any]) -> None:
|
| 44 |
+
random.setstate(state["python"])
|
| 45 |
+
np.random.set_state(state["numpy"])
|
| 46 |
+
torch.set_rng_state(state["torch"])
|
| 47 |
+
if torch.cuda.is_available() and "cuda" in state:
|
| 48 |
+
torch.cuda.set_rng_state_all(state["cuda"])
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
def load_training_state(path: Path) -> dict[str, Any]:
|
| 52 |
+
try:
|
| 53 |
+
return torch.load(path, map_location="cpu", weights_only=False)
|
| 54 |
+
except TypeError:
|
| 55 |
+
return torch.load(path, map_location="cpu")
|
| 56 |
+
|
| 57 |
+
|
| 58 |
+
def save_checkpoint(
|
| 59 |
+
*,
|
| 60 |
+
output_dir: Path,
|
| 61 |
+
step: int,
|
| 62 |
+
model: Any,
|
| 63 |
+
tokenizer: Any,
|
| 64 |
+
optimizer: torch.optim.Optimizer,
|
| 65 |
+
scheduler: Any,
|
| 66 |
+
training_config: dict[str, Any],
|
| 67 |
+
world_size: int,
|
| 68 |
+
) -> Path:
|
| 69 |
+
output_dir.mkdir(parents=True, exist_ok=True)
|
| 70 |
+
final_path = output_dir / f"checkpoint-{step:08d}"
|
| 71 |
+
if final_path.exists():
|
| 72 |
+
return final_path
|
| 73 |
+
|
| 74 |
+
temporary_path = output_dir / f".checkpoint-{step:08d}-{uuid.uuid4().hex}.tmp"
|
| 75 |
+
temporary_path.mkdir(parents=False, exist_ok=False)
|
| 76 |
+
try:
|
| 77 |
+
model.save_pretrained(
|
| 78 |
+
temporary_path / "model",
|
| 79 |
+
safe_serialization=True,
|
| 80 |
+
max_shard_size="2GB",
|
| 81 |
+
)
|
| 82 |
+
tokenizer.save_pretrained(temporary_path / "tokenizer")
|
| 83 |
+
state = {
|
| 84 |
+
"step": step,
|
| 85 |
+
"world_size": world_size,
|
| 86 |
+
"optimizer": optimizer.state_dict(),
|
| 87 |
+
"scheduler": scheduler.state_dict(),
|
| 88 |
+
"rng": capture_rng_state(),
|
| 89 |
+
"training_config": training_config,
|
| 90 |
+
}
|
| 91 |
+
torch.save(state, temporary_path / "trainer_state.pt")
|
| 92 |
+
with (temporary_path / "state_summary.json").open(
|
| 93 |
+
"w", encoding="utf-8"
|
| 94 |
+
) as handle:
|
| 95 |
+
json.dump(
|
| 96 |
+
{"step": step, "world_size": world_size},
|
| 97 |
+
handle,
|
| 98 |
+
ensure_ascii=False,
|
| 99 |
+
indent=2,
|
| 100 |
+
)
|
| 101 |
+
handle.write("\n")
|
| 102 |
+
os.replace(temporary_path, final_path)
|
| 103 |
+
except BaseException:
|
| 104 |
+
# Preserve a failed temporary checkpoint for inspection rather than deleting it.
|
| 105 |
+
raise
|
| 106 |
+
|
| 107 |
+
latest_temp = output_dir / ".latest_checkpoint.tmp"
|
| 108 |
+
latest_temp.write_text(final_path.name + "\n", encoding="utf-8")
|
| 109 |
+
os.replace(latest_temp, output_dir / "latest_checkpoint.txt")
|
| 110 |
+
return final_path
|
| 111 |
+
|
kawk50m/cli.py
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import sys
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
def configure_utf8_stdio() -> None:
|
| 7 |
+
"""Make Korean/emoji CLI output reliable on legacy Windows code pages."""
|
| 8 |
+
for stream in (sys.stdout, sys.stderr):
|
| 9 |
+
reconfigure = getattr(stream, "reconfigure", None)
|
| 10 |
+
if reconfigure is not None:
|
| 11 |
+
reconfigure(encoding="utf-8", errors="replace")
|
| 12 |
+
|
kawk50m/config.py
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
from pathlib import Path
|
| 4 |
+
from typing import Any
|
| 5 |
+
|
| 6 |
+
import yaml
|
| 7 |
+
from transformers import LlamaConfig
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
def load_yaml(path: str | Path) -> dict[str, Any]:
|
| 11 |
+
config_path = Path(path)
|
| 12 |
+
with config_path.open("r", encoding="utf-8") as handle:
|
| 13 |
+
data = yaml.safe_load(handle)
|
| 14 |
+
if not isinstance(data, dict):
|
| 15 |
+
raise ValueError(f"YAML root must be a mapping: {config_path}")
|
| 16 |
+
return data
|
| 17 |
+
|
| 18 |
+
|
| 19 |
+
def _positive_int(value: Any, name: str) -> int:
|
| 20 |
+
if not isinstance(value, int) or isinstance(value, bool) or value <= 0:
|
| 21 |
+
raise ValueError(f"{name} must be a positive integer, got {value!r}")
|
| 22 |
+
return value
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
def build_llama_config(
|
| 26 |
+
model_data: dict[str, Any],
|
| 27 |
+
*,
|
| 28 |
+
tokenizer_vocab_size: int | None = None,
|
| 29 |
+
) -> tuple[LlamaConfig, int | None]:
|
| 30 |
+
"""Validate a model mapping and return a Hugging Face LlamaConfig."""
|
| 31 |
+
if "model" in model_data:
|
| 32 |
+
model_data = model_data["model"]
|
| 33 |
+
if not isinstance(model_data, dict):
|
| 34 |
+
raise ValueError("model configuration must be a mapping")
|
| 35 |
+
|
| 36 |
+
values = dict(model_data)
|
| 37 |
+
expected_parameters = values.pop("expected_parameters", None)
|
| 38 |
+
|
| 39 |
+
required = (
|
| 40 |
+
"vocab_size",
|
| 41 |
+
"hidden_size",
|
| 42 |
+
"intermediate_size",
|
| 43 |
+
"num_hidden_layers",
|
| 44 |
+
"num_attention_heads",
|
| 45 |
+
"num_key_value_heads",
|
| 46 |
+
"max_position_embeddings",
|
| 47 |
+
)
|
| 48 |
+
for name in required:
|
| 49 |
+
values[name] = _positive_int(values.get(name), name)
|
| 50 |
+
|
| 51 |
+
if values["hidden_size"] % values["num_attention_heads"] != 0:
|
| 52 |
+
raise ValueError("hidden_size must be divisible by num_attention_heads")
|
| 53 |
+
if values["num_attention_heads"] % values["num_key_value_heads"] != 0:
|
| 54 |
+
raise ValueError(
|
| 55 |
+
"num_attention_heads must be divisible by num_key_value_heads"
|
| 56 |
+
)
|
| 57 |
+
if tokenizer_vocab_size is not None and values["vocab_size"] != tokenizer_vocab_size:
|
| 58 |
+
raise ValueError(
|
| 59 |
+
"Tokenizer/model vocabulary mismatch: "
|
| 60 |
+
f"tokenizer={tokenizer_vocab_size}, model={values['vocab_size']}"
|
| 61 |
+
)
|
| 62 |
+
|
| 63 |
+
config = LlamaConfig(**values)
|
| 64 |
+
return config, expected_parameters
|
| 65 |
+
|
| 66 |
+
|
| 67 |
+
def count_unique_parameters(model: Any) -> int:
|
| 68 |
+
"""Count parameters without double-counting tied tensors."""
|
| 69 |
+
seen: set[int] = set()
|
| 70 |
+
total = 0
|
| 71 |
+
for parameter in model.parameters():
|
| 72 |
+
pointer = parameter.untyped_storage().data_ptr()
|
| 73 |
+
if pointer not in seen:
|
| 74 |
+
seen.add(pointer)
|
| 75 |
+
total += parameter.numel()
|
| 76 |
+
return total
|
| 77 |
+
|
kawk50m/io.py
ADDED
|
@@ -0,0 +1,144 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import gzip
|
| 4 |
+
import json
|
| 5 |
+
from collections.abc import Iterator, Sequence
|
| 6 |
+
from pathlib import Path
|
| 7 |
+
from typing import Any, TextIO
|
| 8 |
+
import glob
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
SUPPORTED_SUFFIXES = (".jsonl", ".jsonl.gz", ".txt", ".txt.gz")
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
def discover_paths(inputs: Sequence[str | Path]) -> list[Path]:
|
| 15 |
+
found: set[Path] = set()
|
| 16 |
+
for raw_input in inputs:
|
| 17 |
+
raw = str(raw_input)
|
| 18 |
+
path = Path(raw)
|
| 19 |
+
if path.is_dir():
|
| 20 |
+
for candidate in path.rglob("*"):
|
| 21 |
+
if candidate.is_file() and any(
|
| 22 |
+
str(candidate).lower().endswith(suffix)
|
| 23 |
+
for suffix in SUPPORTED_SUFFIXES
|
| 24 |
+
):
|
| 25 |
+
found.add(candidate.resolve())
|
| 26 |
+
continue
|
| 27 |
+
matches = [Path(item) for item in glob.glob(raw, recursive=True)]
|
| 28 |
+
if matches:
|
| 29 |
+
found.update(item.resolve() for item in matches if item.is_file())
|
| 30 |
+
elif path.is_file():
|
| 31 |
+
found.add(path.resolve())
|
| 32 |
+
return sorted(found)
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
def open_text(path: Path) -> TextIO:
|
| 36 |
+
if path.name.lower().endswith(".gz"):
|
| 37 |
+
return gzip.open(path, "rt", encoding="utf-8", errors="replace")
|
| 38 |
+
return path.open("r", encoding="utf-8", errors="replace")
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
def iter_documents(
|
| 42 |
+
paths: Sequence[Path],
|
| 43 |
+
*,
|
| 44 |
+
text_field: str = "text",
|
| 45 |
+
txt_mode: str = "line",
|
| 46 |
+
) -> Iterator[dict[str, Any]]:
|
| 47 |
+
if txt_mode not in {"line", "document"}:
|
| 48 |
+
raise ValueError("txt_mode must be 'line' or 'document'")
|
| 49 |
+
|
| 50 |
+
for path in paths:
|
| 51 |
+
lower_name = path.name.lower()
|
| 52 |
+
source_default = path.name.removesuffix(".gz")
|
| 53 |
+
if lower_name.endswith((".jsonl", ".jsonl.gz")):
|
| 54 |
+
with open_text(path) as handle:
|
| 55 |
+
for line_number, line in enumerate(handle, start=1):
|
| 56 |
+
if not line.strip():
|
| 57 |
+
continue
|
| 58 |
+
try:
|
| 59 |
+
row = json.loads(line)
|
| 60 |
+
except json.JSONDecodeError as error:
|
| 61 |
+
raise ValueError(
|
| 62 |
+
f"Invalid JSON at {path}:{line_number}: {error}"
|
| 63 |
+
) from error
|
| 64 |
+
if not isinstance(row, dict):
|
| 65 |
+
raise ValueError(
|
| 66 |
+
f"Expected a JSON object at {path}:{line_number}"
|
| 67 |
+
)
|
| 68 |
+
text = row.get(text_field)
|
| 69 |
+
if not isinstance(text, str):
|
| 70 |
+
continue
|
| 71 |
+
yield {
|
| 72 |
+
"id": row.get("id", f"{path.name}:{line_number}"),
|
| 73 |
+
"text": text,
|
| 74 |
+
"source": row.get("source", source_default),
|
| 75 |
+
"url": row.get("url"),
|
| 76 |
+
"license": row.get("license"),
|
| 77 |
+
}
|
| 78 |
+
continue
|
| 79 |
+
|
| 80 |
+
if lower_name.endswith((".txt", ".txt.gz")):
|
| 81 |
+
with open_text(path) as handle:
|
| 82 |
+
if txt_mode == "document":
|
| 83 |
+
text = handle.read()
|
| 84 |
+
if text.strip():
|
| 85 |
+
yield {
|
| 86 |
+
"id": path.name,
|
| 87 |
+
"text": text,
|
| 88 |
+
"source": source_default,
|
| 89 |
+
"url": None,
|
| 90 |
+
"license": None,
|
| 91 |
+
}
|
| 92 |
+
else:
|
| 93 |
+
for line_number, line in enumerate(handle, start=1):
|
| 94 |
+
if line.strip():
|
| 95 |
+
yield {
|
| 96 |
+
"id": f"{path.name}:{line_number}",
|
| 97 |
+
"text": line.rstrip("\n"),
|
| 98 |
+
"source": source_default,
|
| 99 |
+
"url": None,
|
| 100 |
+
"license": None,
|
| 101 |
+
}
|
| 102 |
+
|
| 103 |
+
|
| 104 |
+
class ShardedJsonlWriter:
|
| 105 |
+
def __init__(
|
| 106 |
+
self,
|
| 107 |
+
output_dir: Path,
|
| 108 |
+
split: str,
|
| 109 |
+
*,
|
| 110 |
+
documents_per_shard: int,
|
| 111 |
+
) -> None:
|
| 112 |
+
self.output_dir = output_dir
|
| 113 |
+
self.split = split
|
| 114 |
+
self.documents_per_shard = documents_per_shard
|
| 115 |
+
self.document_count = 0
|
| 116 |
+
self.shard_index = -1
|
| 117 |
+
self._handle: TextIO | None = None
|
| 118 |
+
|
| 119 |
+
def _open_next(self) -> None:
|
| 120 |
+
if self._handle is not None:
|
| 121 |
+
self._handle.close()
|
| 122 |
+
self.shard_index += 1
|
| 123 |
+
path = self.output_dir / f"{self.split}-{self.shard_index:05d}.jsonl"
|
| 124 |
+
self._handle = path.open("w", encoding="utf-8", newline="\n")
|
| 125 |
+
|
| 126 |
+
def write(self, row: dict[str, Any]) -> None:
|
| 127 |
+
if self._handle is None or self.document_count % self.documents_per_shard == 0:
|
| 128 |
+
self._open_next()
|
| 129 |
+
assert self._handle is not None
|
| 130 |
+
self._handle.write(json.dumps(row, ensure_ascii=False, separators=(",", ":")))
|
| 131 |
+
self._handle.write("\n")
|
| 132 |
+
self.document_count += 1
|
| 133 |
+
|
| 134 |
+
def close(self) -> None:
|
| 135 |
+
if self._handle is not None:
|
| 136 |
+
self._handle.close()
|
| 137 |
+
self._handle = None
|
| 138 |
+
|
| 139 |
+
def __enter__(self) -> "ShardedJsonlWriter":
|
| 140 |
+
return self
|
| 141 |
+
|
| 142 |
+
def __exit__(self, *_: object) -> None:
|
| 143 |
+
self.close()
|
| 144 |
+
|
kawk50m/sft_dataset.py
ADDED
|
@@ -0,0 +1,139 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import math
|
| 4 |
+
from collections.abc import Sequence
|
| 5 |
+
from pathlib import Path
|
| 6 |
+
|
| 7 |
+
import numpy as np
|
| 8 |
+
import torch
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
class SFTPackedDataset:
|
| 12 |
+
"""Memory-map fixed-length input IDs and assistant-only loss masks."""
|
| 13 |
+
|
| 14 |
+
def __init__(
|
| 15 |
+
self,
|
| 16 |
+
prefixes: str | Path | Sequence[str | Path],
|
| 17 |
+
sequence_length: int,
|
| 18 |
+
) -> None:
|
| 19 |
+
if sequence_length <= 0:
|
| 20 |
+
raise ValueError("sequence_length must be positive")
|
| 21 |
+
raw_prefixes = (
|
| 22 |
+
[prefixes] if isinstance(prefixes, (str, Path)) else list(prefixes)
|
| 23 |
+
)
|
| 24 |
+
self.sequence_length = sequence_length
|
| 25 |
+
self.inputs: list[np.memmap] = []
|
| 26 |
+
self.masks: list[np.memmap] = []
|
| 27 |
+
self.sequence_offsets: list[int] = []
|
| 28 |
+
running_total = 0
|
| 29 |
+
for raw_prefix in raw_prefixes:
|
| 30 |
+
prefix = Path(raw_prefix)
|
| 31 |
+
input_path = Path(str(prefix) + ".input_ids.bin")
|
| 32 |
+
mask_path = Path(str(prefix) + ".loss_mask.bin")
|
| 33 |
+
if not input_path.is_file() or not mask_path.is_file():
|
| 34 |
+
raise FileNotFoundError(
|
| 35 |
+
f"Missing SFT pair: {input_path} and {mask_path}"
|
| 36 |
+
)
|
| 37 |
+
input_stream = np.memmap(input_path, mode="r", dtype=np.uint16)
|
| 38 |
+
mask_stream = np.memmap(mask_path, mode="r", dtype=np.uint8)
|
| 39 |
+
if len(input_stream) != len(mask_stream):
|
| 40 |
+
raise ValueError(f"Input/mask length mismatch for {prefix}")
|
| 41 |
+
if len(input_stream) % sequence_length != 0:
|
| 42 |
+
raise ValueError(f"SFT data is not record-aligned: {prefix}")
|
| 43 |
+
records = len(input_stream) // sequence_length
|
| 44 |
+
if records == 0:
|
| 45 |
+
raise ValueError(f"SFT data contains no records: {prefix}")
|
| 46 |
+
self.inputs.append(input_stream)
|
| 47 |
+
self.masks.append(mask_stream)
|
| 48 |
+
running_total += records
|
| 49 |
+
self.sequence_offsets.append(running_total)
|
| 50 |
+
if not self.inputs:
|
| 51 |
+
raise ValueError("At least one SFT data prefix is required")
|
| 52 |
+
self.num_sequences = running_total
|
| 53 |
+
self._affine_cache: dict[tuple[int, int], tuple[int, int]] = {}
|
| 54 |
+
|
| 55 |
+
def __len__(self) -> int:
|
| 56 |
+
return self.num_sequences
|
| 57 |
+
|
| 58 |
+
def _locate(self, index: int) -> tuple[int, int]:
|
| 59 |
+
if not 0 <= index < self.num_sequences:
|
| 60 |
+
raise IndexError(index)
|
| 61 |
+
previous = 0
|
| 62 |
+
for stream_index, end in enumerate(self.sequence_offsets):
|
| 63 |
+
if index < end:
|
| 64 |
+
return stream_index, index - previous
|
| 65 |
+
previous = end
|
| 66 |
+
raise IndexError(index)
|
| 67 |
+
|
| 68 |
+
def sequence(self, index: int) -> tuple[np.ndarray, np.ndarray]:
|
| 69 |
+
stream_index, local_index = self._locate(index)
|
| 70 |
+
start = local_index * self.sequence_length
|
| 71 |
+
stop = start + self.sequence_length
|
| 72 |
+
return (
|
| 73 |
+
np.asarray(self.inputs[stream_index][start:stop], dtype=np.int64),
|
| 74 |
+
np.asarray(self.masks[stream_index][start:stop], dtype=np.bool_),
|
| 75 |
+
)
|
| 76 |
+
|
| 77 |
+
def _affine_parameters(self, epoch: int, seed: int) -> tuple[int, int]:
|
| 78 |
+
key = (epoch, seed)
|
| 79 |
+
if key in self._affine_cache:
|
| 80 |
+
return self._affine_cache[key]
|
| 81 |
+
rng = np.random.default_rng(seed + epoch * 1_000_003)
|
| 82 |
+
multiplier = int(rng.integers(1, max(self.num_sequences, 2)))
|
| 83 |
+
while math.gcd(multiplier, self.num_sequences) != 1:
|
| 84 |
+
multiplier += 1
|
| 85 |
+
if multiplier >= self.num_sequences:
|
| 86 |
+
multiplier = 1
|
| 87 |
+
offset = int(rng.integers(0, self.num_sequences))
|
| 88 |
+
self._affine_cache[key] = (multiplier, offset)
|
| 89 |
+
return multiplier, offset
|
| 90 |
+
|
| 91 |
+
def training_batch(
|
| 92 |
+
self,
|
| 93 |
+
*,
|
| 94 |
+
micro_step: int,
|
| 95 |
+
batch_size: int,
|
| 96 |
+
rank: int,
|
| 97 |
+
world_size: int,
|
| 98 |
+
seed: int,
|
| 99 |
+
device: torch.device,
|
| 100 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 101 |
+
global_batch_size = batch_size * world_size
|
| 102 |
+
first = micro_step * global_batch_size + rank * batch_size
|
| 103 |
+
indices = []
|
| 104 |
+
for local_index in range(batch_size):
|
| 105 |
+
ordinal = first + local_index
|
| 106 |
+
epoch, position = divmod(ordinal, self.num_sequences)
|
| 107 |
+
multiplier, offset = self._affine_parameters(epoch, seed)
|
| 108 |
+
indices.append((multiplier * position + offset) % self.num_sequences)
|
| 109 |
+
return self._to_tensors(indices, device)
|
| 110 |
+
|
| 111 |
+
def evaluation_batch(
|
| 112 |
+
self,
|
| 113 |
+
*,
|
| 114 |
+
batch_index: int,
|
| 115 |
+
batch_size: int,
|
| 116 |
+
rank: int,
|
| 117 |
+
world_size: int,
|
| 118 |
+
device: torch.device,
|
| 119 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 120 |
+
first = (batch_index * world_size + rank) * batch_size
|
| 121 |
+
indices = [
|
| 122 |
+
(first + local_index) % self.num_sequences
|
| 123 |
+
for local_index in range(batch_size)
|
| 124 |
+
]
|
| 125 |
+
return self._to_tensors(indices, device)
|
| 126 |
+
|
| 127 |
+
def _to_tensors(
|
| 128 |
+
self,
|
| 129 |
+
indices: list[int],
|
| 130 |
+
device: torch.device,
|
| 131 |
+
) -> tuple[torch.Tensor, torch.Tensor]:
|
| 132 |
+
rows = [self.sequence(index) for index in indices]
|
| 133 |
+
inputs = torch.from_numpy(np.stack([row[0] for row in rows]))
|
| 134 |
+
masks = torch.from_numpy(np.stack([row[1] for row in rows]))
|
| 135 |
+
inputs = inputs.to(device=device, dtype=torch.long, non_blocking=True)
|
| 136 |
+
masks = masks.to(device=device, dtype=torch.bool, non_blocking=True)
|
| 137 |
+
labels = inputs.clone()
|
| 138 |
+
labels.masked_fill_(~masks, -100)
|
| 139 |
+
return inputs, labels
|
kawk50m/text.py
ADDED
|
@@ -0,0 +1,337 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from __future__ import annotations
|
| 2 |
+
|
| 3 |
+
import hashlib
|
| 4 |
+
import html
|
| 5 |
+
import math
|
| 6 |
+
import re
|
| 7 |
+
import unicodedata
|
| 8 |
+
from dataclasses import asdict, dataclass
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
HANGUL_RE = re.compile(r"[가-힣]")
|
| 12 |
+
LATIN_RE = re.compile(r"[A-Za-z]")
|
| 13 |
+
CJK_RE = re.compile(r"[\u3400-\u4dbf\u4e00-\u9fff]")
|
| 14 |
+
URL_RE = re.compile(r"(?:https?://|www\.)\S+", re.IGNORECASE)
|
| 15 |
+
EMAIL_RE = re.compile(r"\b[\w.+-]+@[\w.-]+\.[A-Za-z]{2,}\b")
|
| 16 |
+
PHONE_RE = re.compile(r"(?<!\d)(?:\+?82[- ]?)?0?1[016789][ -]?\d{3,4}[ -]?\d{4}(?!\d)")
|
| 17 |
+
RRN_RE = re.compile(r"(?<!\d)\d{6}[ -]?[1-4]\d{6}(?!\d)")
|
| 18 |
+
SCRIPT_STYLE_RE = re.compile(
|
| 19 |
+
r"<(?:script|style)\b[^>]*>.*?</(?:script|style)>",
|
| 20 |
+
re.IGNORECASE | re.DOTALL,
|
| 21 |
+
)
|
| 22 |
+
HTML_TAG_RE = re.compile(r"<[^>\n]{1,500}>")
|
| 23 |
+
MULTISPACE_RE = re.compile(r"[ \t]{2,}")
|
| 24 |
+
MANY_NEWLINES_RE = re.compile(r"\n{3,}")
|
| 25 |
+
CODE_LINE_RE = re.compile(
|
| 26 |
+
r"^\s*(?:"
|
| 27 |
+
r"(?:def|class|import|from|function|const|let|var|public|private)\s+"
|
| 28 |
+
r"|#include\b|package\s+\w+|using\s+namespace\b"
|
| 29 |
+
r"|[{}[\];]{2,}|</?\w+[^>]*>"
|
| 30 |
+
r")"
|
| 31 |
+
)
|
| 32 |
+
CODE_MATH_CHARS = frozenset("{}[]<>=+*/\\^_|~`$")
|
| 33 |
+
WORD_RE = re.compile(r"[가-힣A-Za-z0-9]+")
|
| 34 |
+
LIST_LINE_RE = re.compile(
|
| 35 |
+
r"^\s*(?:[-*•·▪◦▶▷■□]|(?:\d+|[가-힣A-Za-z])[.)])\s+"
|
| 36 |
+
)
|
| 37 |
+
BOILERPLATE_RE = re.compile(
|
| 38 |
+
r"(?:로그인|회원가입|개인정보\s*처리방침|이용약관|전체\s*메뉴|"
|
| 39 |
+
r"장바구니|상품\s*후기|배송\s*조회|쿠폰|고객\s*센터|"
|
| 40 |
+
r"무단\s*전재|재배포\s*금지|관련\s*기사|실시간\s*뉴스)"
|
| 41 |
+
)
|
| 42 |
+
REPEATED_CHARACTER_RE = re.compile(r"(.)\1+")
|
| 43 |
+
NON_CONTENT_RE = re.compile(r"[\W_]+", re.UNICODE)
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
@dataclass(slots=True)
|
| 47 |
+
class FilterConfig:
|
| 48 |
+
min_chars: int = 300
|
| 49 |
+
max_chars: int = 100_000
|
| 50 |
+
min_hangul_chars: int = 50
|
| 51 |
+
min_hangul_letter_ratio: float = 0.70
|
| 52 |
+
min_hangul_visible_ratio: float = 0.55
|
| 53 |
+
max_digit_ratio: float = 0.08
|
| 54 |
+
max_symbol_ratio: float = 0.15
|
| 55 |
+
max_url_per_word: float = 0.01
|
| 56 |
+
min_unique_line_ratio: float = 0.70
|
| 57 |
+
min_unique_word_ratio: float = 0.20
|
| 58 |
+
max_short_line_ratio: float = 0.50
|
| 59 |
+
max_list_line_ratio: float = 0.25
|
| 60 |
+
max_table_line_ratio: float = 0.10
|
| 61 |
+
max_boilerplate_hits_per_1k: float = 1.5
|
| 62 |
+
max_repeated_character_run: int = 20
|
| 63 |
+
max_code_line_ratio: float = 0.25
|
| 64 |
+
max_code_math_char_ratio: float = 0.15
|
| 65 |
+
min_quality_score: float = 0.70
|
| 66 |
+
strip_html: bool = True
|
| 67 |
+
mask_pii: bool = True
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
@dataclass(slots=True)
|
| 71 |
+
class TextMetrics:
|
| 72 |
+
characters: int
|
| 73 |
+
visible_characters: int
|
| 74 |
+
hangul_characters: int
|
| 75 |
+
latin_characters: int
|
| 76 |
+
cjk_characters: int
|
| 77 |
+
digit_characters: int
|
| 78 |
+
symbol_characters: int
|
| 79 |
+
hangul_letter_ratio: float
|
| 80 |
+
hangul_visible_ratio: float
|
| 81 |
+
digit_ratio: float
|
| 82 |
+
symbol_ratio: float
|
| 83 |
+
url_per_word: float
|
| 84 |
+
unique_line_ratio: float
|
| 85 |
+
unique_word_ratio: float
|
| 86 |
+
short_line_ratio: float
|
| 87 |
+
list_line_ratio: float
|
| 88 |
+
table_line_ratio: float
|
| 89 |
+
boilerplate_hits_per_1k: float
|
| 90 |
+
repeated_character_run: int
|
| 91 |
+
code_line_ratio: float
|
| 92 |
+
code_math_char_ratio: float
|
| 93 |
+
|
| 94 |
+
def to_dict(self) -> dict[str, int | float]:
|
| 95 |
+
return asdict(self)
|
| 96 |
+
|
| 97 |
+
|
| 98 |
+
def normalize_text(
|
| 99 |
+
text: str,
|
| 100 |
+
*,
|
| 101 |
+
strip_html: bool = True,
|
| 102 |
+
mask_pii: bool = True,
|
| 103 |
+
) -> str:
|
| 104 |
+
if not isinstance(text, str):
|
| 105 |
+
raise TypeError(f"text must be str, got {type(text).__name__}")
|
| 106 |
+
|
| 107 |
+
text = unicodedata.normalize("NFC", text)
|
| 108 |
+
text = text.replace("\r\n", "\n").replace("\r", "\n")
|
| 109 |
+
text = text.replace("\u00a0", " ").replace("\u200b", "")
|
| 110 |
+
text = text.replace("\u200c", "").replace("\u200d", "")
|
| 111 |
+
text = "".join(
|
| 112 |
+
character
|
| 113 |
+
for character in text
|
| 114 |
+
if character in "\n\t" or unicodedata.category(character) != "Cc"
|
| 115 |
+
)
|
| 116 |
+
|
| 117 |
+
if strip_html:
|
| 118 |
+
text = SCRIPT_STYLE_RE.sub(" ", text)
|
| 119 |
+
text = HTML_TAG_RE.sub(" ", text)
|
| 120 |
+
text = html.unescape(text)
|
| 121 |
+
|
| 122 |
+
if mask_pii:
|
| 123 |
+
text = EMAIL_RE.sub("<EMAIL>", text)
|
| 124 |
+
text = PHONE_RE.sub("<PHONE>", text)
|
| 125 |
+
text = RRN_RE.sub("<RRN>", text)
|
| 126 |
+
|
| 127 |
+
lines = [MULTISPACE_RE.sub(" ", line).strip() for line in text.splitlines()]
|
| 128 |
+
text = "\n".join(line for line in lines if line)
|
| 129 |
+
text = MANY_NEWLINES_RE.sub("\n\n", text)
|
| 130 |
+
return text.strip()
|
| 131 |
+
|
| 132 |
+
|
| 133 |
+
def measure_text(text: str) -> TextMetrics:
|
| 134 |
+
visible = [character for character in text if not character.isspace()]
|
| 135 |
+
hangul_count = len(HANGUL_RE.findall(text))
|
| 136 |
+
latin_count = len(LATIN_RE.findall(text))
|
| 137 |
+
cjk_count = len(CJK_RE.findall(text))
|
| 138 |
+
digit_count = sum(character.isdigit() for character in visible)
|
| 139 |
+
symbol_count = sum(
|
| 140 |
+
unicodedata.category(character).startswith(("P", "S"))
|
| 141 |
+
for character in visible
|
| 142 |
+
)
|
| 143 |
+
letter_count = hangul_count + latin_count + cjk_count
|
| 144 |
+
lines = [line for line in text.splitlines() if line.strip()]
|
| 145 |
+
words = text.split()
|
| 146 |
+
content_words = WORD_RE.findall(text.casefold())
|
| 147 |
+
code_lines = sum(bool(CODE_LINE_RE.search(line)) for line in lines)
|
| 148 |
+
code_math_count = sum(character in CODE_MATH_CHARS for character in visible)
|
| 149 |
+
short_lines = sum(len(line.strip()) < 20 for line in lines)
|
| 150 |
+
list_lines = sum(bool(LIST_LINE_RE.search(line)) for line in lines)
|
| 151 |
+
table_lines = sum(line.count("|") >= 2 for line in lines)
|
| 152 |
+
boilerplate_hits = len(BOILERPLATE_RE.findall(text))
|
| 153 |
+
repeated_runs = [
|
| 154 |
+
len(match.group(0)) for match in REPEATED_CHARACTER_RE.finditer(text)
|
| 155 |
+
]
|
| 156 |
+
|
| 157 |
+
return TextMetrics(
|
| 158 |
+
characters=len(text),
|
| 159 |
+
visible_characters=len(visible),
|
| 160 |
+
hangul_characters=hangul_count,
|
| 161 |
+
latin_characters=latin_count,
|
| 162 |
+
cjk_characters=cjk_count,
|
| 163 |
+
digit_characters=digit_count,
|
| 164 |
+
symbol_characters=symbol_count,
|
| 165 |
+
hangul_letter_ratio=hangul_count / max(letter_count, 1),
|
| 166 |
+
hangul_visible_ratio=hangul_count / max(len(visible), 1),
|
| 167 |
+
digit_ratio=digit_count / max(len(visible), 1),
|
| 168 |
+
symbol_ratio=symbol_count / max(len(visible), 1),
|
| 169 |
+
url_per_word=len(URL_RE.findall(text)) / max(len(words), 1),
|
| 170 |
+
unique_line_ratio=len(set(lines)) / max(len(lines), 1),
|
| 171 |
+
unique_word_ratio=len(set(content_words)) / max(len(content_words), 1),
|
| 172 |
+
short_line_ratio=short_lines / max(len(lines), 1),
|
| 173 |
+
list_line_ratio=list_lines / max(len(lines), 1),
|
| 174 |
+
table_line_ratio=table_lines / max(len(lines), 1),
|
| 175 |
+
boilerplate_hits_per_1k=boilerplate_hits * 1000 / max(len(text), 1),
|
| 176 |
+
repeated_character_run=max(repeated_runs, default=1),
|
| 177 |
+
code_line_ratio=code_lines / max(len(lines), 1),
|
| 178 |
+
code_math_char_ratio=code_math_count / max(len(visible), 1),
|
| 179 |
+
)
|
| 180 |
+
|
| 181 |
+
|
| 182 |
+
def rejection_reason(
|
| 183 |
+
text: str,
|
| 184 |
+
config: FilterConfig,
|
| 185 |
+
*,
|
| 186 |
+
metrics: TextMetrics | None = None,
|
| 187 |
+
) -> str | None:
|
| 188 |
+
metrics = metrics or measure_text(text)
|
| 189 |
+
if metrics.characters < config.min_chars:
|
| 190 |
+
return "too_short"
|
| 191 |
+
if metrics.characters > config.max_chars:
|
| 192 |
+
return "too_long"
|
| 193 |
+
if metrics.hangul_characters < config.min_hangul_chars:
|
| 194 |
+
return "too_little_hangul"
|
| 195 |
+
if metrics.hangul_letter_ratio < config.min_hangul_letter_ratio:
|
| 196 |
+
return "low_hangul_letter_ratio"
|
| 197 |
+
if metrics.hangul_visible_ratio < config.min_hangul_visible_ratio:
|
| 198 |
+
return "low_hangul_visible_ratio"
|
| 199 |
+
if metrics.digit_ratio > config.max_digit_ratio:
|
| 200 |
+
return "digit_heavy"
|
| 201 |
+
if metrics.symbol_ratio > config.max_symbol_ratio:
|
| 202 |
+
return "symbol_heavy"
|
| 203 |
+
if metrics.repeated_character_run > config.max_repeated_character_run:
|
| 204 |
+
return "repeated_character_run"
|
| 205 |
+
if metrics.url_per_word > config.max_url_per_word:
|
| 206 |
+
return "too_many_urls"
|
| 207 |
+
if (
|
| 208 |
+
len(text.splitlines()) >= 5
|
| 209 |
+
and metrics.unique_line_ratio < config.min_unique_line_ratio
|
| 210 |
+
):
|
| 211 |
+
return "repeated_lines"
|
| 212 |
+
if len(WORD_RE.findall(text)) >= 100 and (
|
| 213 |
+
metrics.unique_word_ratio < config.min_unique_word_ratio
|
| 214 |
+
):
|
| 215 |
+
return "low_unique_word_ratio"
|
| 216 |
+
if (
|
| 217 |
+
len(text.splitlines()) >= 8
|
| 218 |
+
and metrics.short_line_ratio > config.max_short_line_ratio
|
| 219 |
+
):
|
| 220 |
+
return "short_line_heavy"
|
| 221 |
+
if (
|
| 222 |
+
len(text.splitlines()) >= 8
|
| 223 |
+
and metrics.list_line_ratio > config.max_list_line_ratio
|
| 224 |
+
):
|
| 225 |
+
return "list_heavy"
|
| 226 |
+
if (
|
| 227 |
+
len(text.splitlines()) >= 4
|
| 228 |
+
and metrics.table_line_ratio > config.max_table_line_ratio
|
| 229 |
+
):
|
| 230 |
+
return "table_heavy"
|
| 231 |
+
if metrics.boilerplate_hits_per_1k > config.max_boilerplate_hits_per_1k:
|
| 232 |
+
return "boilerplate_heavy"
|
| 233 |
+
if (
|
| 234 |
+
len(text.splitlines()) >= 4
|
| 235 |
+
and metrics.code_line_ratio > config.max_code_line_ratio
|
| 236 |
+
):
|
| 237 |
+
return "code_heavy"
|
| 238 |
+
if metrics.code_math_char_ratio > config.max_code_math_char_ratio:
|
| 239 |
+
return "code_or_math_heavy"
|
| 240 |
+
if quality_score(metrics, config) < config.min_quality_score:
|
| 241 |
+
return "low_quality_score"
|
| 242 |
+
return None
|
| 243 |
+
|
| 244 |
+
|
| 245 |
+
def quality_score(metrics: TextMetrics, config: FilterConfig) -> float:
|
| 246 |
+
"""Return a conservative 0..1 heuristic used for corpus auditing."""
|
| 247 |
+
hangul_score = min(
|
| 248 |
+
1.0,
|
| 249 |
+
metrics.hangul_visible_ratio
|
| 250 |
+
/ max(config.min_hangul_visible_ratio * 1.5, 1e-9),
|
| 251 |
+
)
|
| 252 |
+
digit_score = max(
|
| 253 |
+
0.0,
|
| 254 |
+
1.0
|
| 255 |
+
- metrics.digit_ratio / max(config.max_digit_ratio * 2.0, 1e-9),
|
| 256 |
+
)
|
| 257 |
+
symbol_score = max(
|
| 258 |
+
0.0,
|
| 259 |
+
1.0
|
| 260 |
+
- metrics.symbol_ratio / max(config.max_symbol_ratio * 2.0, 1e-9),
|
| 261 |
+
)
|
| 262 |
+
line_score = min(1.0, metrics.unique_line_ratio)
|
| 263 |
+
word_score = min(1.0, metrics.unique_word_ratio / 0.50)
|
| 264 |
+
score = (
|
| 265 |
+
0.35 * hangul_score
|
| 266 |
+
+ 0.20 * digit_score
|
| 267 |
+
+ 0.15 * symbol_score
|
| 268 |
+
+ 0.15 * line_score
|
| 269 |
+
+ 0.15 * word_score
|
| 270 |
+
)
|
| 271 |
+
return max(0.0, min(1.0, score))
|
| 272 |
+
|
| 273 |
+
|
| 274 |
+
def near_duplicate_signature(text: str, *, sketch_size: int = 4) -> str:
|
| 275 |
+
"""Build a bottom-k paragraph/chunk sketch for conservative near deduplication."""
|
| 276 |
+
if sketch_size <= 0:
|
| 277 |
+
raise ValueError("sketch_size must be positive")
|
| 278 |
+
|
| 279 |
+
canonical_blocks: list[str] = []
|
| 280 |
+
for paragraph in re.split(r"\n+", text):
|
| 281 |
+
canonical = NON_CONTENT_RE.sub("", paragraph.casefold())
|
| 282 |
+
if len(canonical) < 40:
|
| 283 |
+
continue
|
| 284 |
+
if len(canonical) <= 320:
|
| 285 |
+
canonical_blocks.append(canonical)
|
| 286 |
+
continue
|
| 287 |
+
for start in range(0, len(canonical), 240):
|
| 288 |
+
chunk = canonical[start : start + 320]
|
| 289 |
+
if len(chunk) >= 80:
|
| 290 |
+
canonical_blocks.append(chunk)
|
| 291 |
+
|
| 292 |
+
if not canonical_blocks:
|
| 293 |
+
canonical = NON_CONTENT_RE.sub("", text.casefold())
|
| 294 |
+
canonical_blocks = [
|
| 295 |
+
canonical[start : start + 320]
|
| 296 |
+
for start in range(0, len(canonical), 240)
|
| 297 |
+
if len(canonical[start : start + 320]) >= 40
|
| 298 |
+
]
|
| 299 |
+
|
| 300 |
+
fingerprints = sorted(
|
| 301 |
+
{
|
| 302 |
+
hashlib.blake2b(
|
| 303 |
+
block.encode("utf-8"), digest_size=8, person=b"kawk-v2"
|
| 304 |
+
).hexdigest()
|
| 305 |
+
for block in canonical_blocks
|
| 306 |
+
}
|
| 307 |
+
)
|
| 308 |
+
length_bucket = int(math.log2(max(len(text), 1)) * 4)
|
| 309 |
+
selected = fingerprints[:sketch_size]
|
| 310 |
+
return f"{length_bucket}:" + ":".join(selected)
|
| 311 |
+
|
| 312 |
+
|
| 313 |
+
def canonical_hash(text: str) -> str:
|
| 314 |
+
canonical = " ".join(text.casefold().split())
|
| 315 |
+
return hashlib.sha256(canonical.encode("utf-8")).hexdigest()
|
| 316 |
+
|
| 317 |
+
|
| 318 |
+
def split_for_hash(
|
| 319 |
+
document_hash: str,
|
| 320 |
+
*,
|
| 321 |
+
validation_ratio: float,
|
| 322 |
+
test_ratio: float,
|
| 323 |
+
seed: int,
|
| 324 |
+
) -> str:
|
| 325 |
+
if validation_ratio < 0 or test_ratio < 0:
|
| 326 |
+
raise ValueError("split ratios must be non-negative")
|
| 327 |
+
if validation_ratio + test_ratio >= 1:
|
| 328 |
+
raise ValueError("validation_ratio + test_ratio must be less than 1")
|
| 329 |
+
digest = hashlib.blake2b(
|
| 330 |
+
f"{seed}:{document_hash}".encode("ascii"), digest_size=8
|
| 331 |
+
).digest()
|
| 332 |
+
value = int.from_bytes(digest, "big") / 2**64
|
| 333 |
+
if value < test_ratio:
|
| 334 |
+
return "test"
|
| 335 |
+
if value < test_ratio + validation_ratio:
|
| 336 |
+
return "validation"
|
| 337 |
+
return "train"
|
pyproject.toml
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[build-system]
|
| 2 |
+
requires = ["setuptools>=69"]
|
| 3 |
+
build-backend = "setuptools.build_meta"
|
| 4 |
+
|
| 5 |
+
[project]
|
| 6 |
+
name = "kawk50m"
|
| 7 |
+
version = "0.1.0"
|
| 8 |
+
description = "Korean-first 50M parameter Llama pretraining toolkit"
|
| 9 |
+
requires-python = ">=3.10"
|
| 10 |
+
dependencies = [
|
| 11 |
+
"torch>=2.4,<3",
|
| 12 |
+
"transformers>=4.48,<5",
|
| 13 |
+
"sentencepiece>=0.2,<1",
|
| 14 |
+
"protobuf>=4.25,<7",
|
| 15 |
+
"PyYAML>=6,<7",
|
| 16 |
+
"numpy>=1.26,<3",
|
| 17 |
+
"safetensors>=0.4,<1",
|
| 18 |
+
"tqdm>=4.66,<5",
|
| 19 |
+
]
|
| 20 |
+
|
| 21 |
+
[tool.setuptools.packages.find]
|
| 22 |
+
include = ["kawk50m*"]
|
| 23 |
+
|
| 24 |
+
[tool.pytest.ini_options]
|
| 25 |
+
testpaths = ["tests"]
|
| 26 |
+
addopts = "-q"
|
reports/fixed-ko-cpt-3b-5k/byte-evaluation.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415",
|
| 3 |
+
"tokenizer": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415/tokenizer",
|
| 4 |
+
"inputs": [
|
| 5 |
+
"/workspace/KAWK50M/data/clean-v2-full/validation-00000.jsonl"
|
| 6 |
+
],
|
| 7 |
+
"documents": 500,
|
| 8 |
+
"characters": 792512,
|
| 9 |
+
"utf8_bytes": 1888545,
|
| 10 |
+
"predicted_tokens": 418575,
|
| 11 |
+
"context_length": 5120,
|
| 12 |
+
"token_nll": 2.8587026298034006,
|
| 13 |
+
"token_perplexity": 17.438887560946455,
|
| 14 |
+
"nats_per_character": 1.5098590977423163,
|
| 15 |
+
"bits_per_utf8_byte": 0.9140910746909807,
|
| 16 |
+
"note": "Compare bits_per_utf8_byte only when checkpoints are evaluated on the exact same ordered raw-text documents."
|
| 17 |
+
}
|
reports/fixed-ko-cpt-3b-5k/evaluation.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415",
|
| 3 |
+
"tokenizer": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415/tokenizer",
|
| 4 |
+
"validation_loss": 2.8582549810409548,
|
| 5 |
+
"validation_perplexity": 17.431082811534903,
|
| 6 |
+
"evaluated_tokens": 409600,
|
| 7 |
+
"samples": [
|
| 8 |
+
{
|
| 9 |
+
"prompt": "대한민국의 수도는",
|
| 10 |
+
"continuation": "그 자체로 하나의 문화유산과 문화역사를 품고 있다.\n그 중에서도 ‘조선왕조실록’은 조선왕조실록을 대표하는 자료로써, 역사적 사실을 간직한 역사의 현장이며, 그 유적들이 현대에 이르러서는 더욱 획기적으로 발전하고 있는 것으로 평가되고 있다. 특히 1970년대에 들어와서 그 실황이 확증되면서 6.25전쟁으로 인해 유포된 조선왕조실록이 다시 활기를 되찾게 되었다. 이렇듯 조선왕조실록은 현재도 계속되고 있으며, 13세기 전반부터 18세기 후반까지 그 유물이 수집되어져 오고 있다.\n그러나 이러한 사실들은 아직까지 제대로 보고된 바가 없다. 다만 일부 자료만이 3D영상자료로 대체될 뿐이며, 그 중에서도 현재의 유물들은 모두 ‘한국사’라는 책자에서 확인가능하다.",
|
| 11 |
+
"full_text": "대한민국의 수도는그 자체로 하나의 문화유산과 문화역사를 품고 있다.\n그 중에서도 ‘조선왕조실록’은 조선왕조실록을 대표하는 자료로써, 역사적 사실을 간직한 역사의 현장이며, 그 유적들이 현대에 이르러서는 더욱 획기적으로 발전하고 있는 것으로 평가되고 있다. 특히 1970년대에 들어와서 그 실황이 확증되면서 6.25전쟁으로 인해 유포된 조선왕조실록이 다시 활기를 되찾게 되었다. 이렇듯 조선왕조실록은 현재도 계속되고 있으며, 13세기 전반부터 18세기 후반까지 그 유물이 수집되어져 오고 있다.\n그러나 이러한 사실들은 아직까지 제대로 보고된 바가 없다. 다만 일부 자료만이 3D영상자료로 대체될 뿐이며, 그 중에서도 현재의 유물들은 모두 ‘한국사’라는 책자에서 확인가능하다."
|
| 12 |
+
},
|
| 13 |
+
{
|
| 14 |
+
"prompt": "오늘 아침에 일어나서",
|
| 15 |
+
"continuation": "13일 동안 잠을 잘 수가 없었다. 그런데 새벽에 일어나서 내가 잠이 안 들어 잠들기가 쉽지 않았다. 새벽에 일어나서 10분 정도 자고 일어나면 일어나기 어려울 정도의 피곤함이 몰려왔다. 그래서 나는 새벽에 일어나서 15분 정도 자는 것으로 새벽에 일어날 수 있는 시간을 잡으려 했다. 그리고 4시쯤에 일어나서 2시간 정도 자고 일어나니 잠이 안 들었다. 아침에 일어나서 15분 정도 자고 일어나니 피곤함이 싹 사라져 있었다.\n그런데 새벽에 일어난 것이 너무너무 힘들었다. 아침에 일어나서 10분 정도 자고 일어나니 피곤함도 많이 사라졌다. 밤에 일어나서 잠을 못 잤는데 새벽에 일어나서 10분 정도 자고 일어나니 피곤함도 많이 사라졌다. 그래서 7월의 마지막 주말에는 낮잠을 잔다고 생각했는데, 오늘도 잠을 못",
|
| 16 |
+
"full_text": "오늘 아침에 일어나서13일 동안 잠을 잘 수가 없었다. 그런데 새벽에 일어나서 내가 잠이 안 들어 잠들기가 쉽지 않았다. 새벽에 일어나서 10분 정도 자고 일어나면 일어나기 어려울 정도의 피곤함이 몰려왔다. 그래서 나는 새벽에 일어나서 15분 정도 자는 것으로 새벽에 일어날 수 있는 시간을 잡으려 했다. 그리고 4시쯤에 일어나서 2시간 정도 자고 일어나니 잠이 안 들었다. 아침에 일어나서 15분 정도 자고 일어나니 피곤함이 싹 사라져 있었다.\n그런데 새벽에 일어난 것이 너무너무 힘들었다. 아침에 일어나서 10분 정도 자고 일어나니 피곤함도 많이 사라졌다. 밤에 일어나서 잠을 못 잤는데 새벽에 일어나서 10분 정도 자고 일어나니 피곤함도 많이 사라졌다. 그래서 7월의 마지막 주말에는 낮잠을 잔다고 생각했는데, 오늘도 잠을 못"
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"prompt": "인공지능 언어 모델은",
|
| 20 |
+
"continuation": "딥러닝을 사용하여 컴퓨터가 학습한 패턴의 모델을 학습하는 모델이다.\n이 모델은 여러 사람이 사용하는 컴퓨터에 의존하지 않고, 사용자가 입력한 데이터와 기계학습 알고리즘이 서로 상호 작용할 수 있는 학습 환경인 딥러닝을 기반으로 한다.\n* **예시:** A는 “아무도 없는 환경에서 10만 개의 데이터 세트를 분석하고, 9억 개의 데이터를 학습하고 있습니다.”\n**[요약]**\n해당 모델은 AI가 학습한 패턴을 머신 러닝으로 학습하여, 컴퓨터가 학습한 패턴의 모델을 학습할 때 발생하는 문제나 코딩 오류를 사전에 방지한다. 또한 학습된 ��이터셋을 분석하여 학습 모델을 훈련할 수 있다.",
|
| 21 |
+
"full_text": "인공지능 언어 모델은딥러닝을 사용하여 컴퓨터가 학습한 패턴의 모델을 학습하는 모델이다.\n이 모델은 여러 사람이 사용하는 컴퓨터에 의존하지 않고, 사용자가 입력한 데이터와 기계학습 알고리즘이 서로 상호 작용할 수 있는 학습 환경인 딥러닝을 기반으로 한다.\n* **예시:** A는 “아무도 없는 환경에서 10만 개의 데이터 세트를 분석하고, 9억 개의 데이터를 학습하고 있습니다.”\n**[요약]**\n해당 모델은 AI가 학습한 패턴을 머신 러닝으로 학습하여, 컴퓨터가 학습한 패턴의 모델을 학습할 때 발생하는 문제나 코딩 오류를 사전에 방지한다. 또한 학습된 데이터셋을 분석하여 학습 모델을 훈련할 수 있다."
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"prompt": "비가 내리는 날에는",
|
| 25 |
+
"continuation": "햇빛이 반사되어 차분한 오후의 분위기를 연출한다.\n[경기도민일보 인천=김용국기자] 인천 남동구는 지난 12일 관내 사회복지시설을 대상으로 아동학대 예방과 위기청소년 지원에 대한 인식 제고를 위한 ‘아동학대 예방교육’을 실시했다.이번 교육은 아동학대 예방을 위해 가정 내 아동학대를 인식하고 아동학대 사례관리 및 상담사례 공유, 아동보호자 상담 및 보호조치 방법 등을 안내하는 내용으로 진행됐다.이날 교육을 받은 한 아동복지시설은 “아이들에게 도움을 주기 위해 아동학대가 의심될 때, ‘아빠는 엄마다’라는 말을 하는 경우가 많다”며 “언제든지 아동을 찾을 수 있도록 가정 내 아동학대 예방을 위한 지속적인 관심을 가져주시길 바란다”라고 전했다.",
|
| 26 |
+
"full_text": "비가 내리는 날에는햇빛이 반사되어 차분한 오후의 분위기를 연출한다.\n[경기도민일보 인천=김용국기자] 인천 남동구는 지난 12일 관내 사회복지시설을 대상으로 아동학대 예방과 위기청소년 지원에 대한 인식 제고를 위한 ‘아동학대 예방교육’을 실시했다.이번 교육은 아동학대 예방을 위해 가정 내 아동학대를 인식하고 아동학대 사례관리 및 상담사례 공유, 아동보호자 상담 및 보호조치 방법 등을 안내하는 내용으로 진행됐다.이날 교육을 받은 한 아동복지시설은 “아이들에게 도움을 주기 위해 아동학대가 의심될 때, ‘아빠는 엄마다’라는 말을 하는 경우가 많다”며 “언제든지 아동을 찾을 수 있도록 가정 내 아동학대 예방을 위한 지속적인 관심을 가져주시길 바란다”라고 전했다."
|
| 27 |
+
}
|
| 28 |
+
],
|
| 29 |
+
"note": "This is a base-model continuation evaluation, not instruction QA."
|
| 30 |
+
}
|
reports/fixed-ko-cpt-3b-5k/final_checkpoint.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415
|
reports/fixed-ko-cpt-3b-5k/gate.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"passed": true,
|
| 3 |
+
"candidate_loss": 2.8582549810409548,
|
| 4 |
+
"baseline_loss": null,
|
| 5 |
+
"loss_improvement": null,
|
| 6 |
+
"sample_count": 4
|
| 7 |
+
}
|
reports/fixed-ko-cpt-3b-5k/position-bucket-evaluation.json
ADDED
|
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415",
|
| 3 |
+
"validation_bin": "/workspace/KAWK50M/data/tokenized-v2-cpt-3b/validation.bin",
|
| 4 |
+
"sequence_length": 5120,
|
| 5 |
+
"bucket_size": 1024,
|
| 6 |
+
"evaluated_tokens": 102380,
|
| 7 |
+
"loss": 2.9305559042446037,
|
| 8 |
+
"perplexity": 18.738044160189084,
|
| 9 |
+
"buckets": [
|
| 10 |
+
{
|
| 11 |
+
"start_position": 1,
|
| 12 |
+
"end_position": 1024,
|
| 13 |
+
"tokens": 20480,
|
| 14 |
+
"loss": 2.9409491062164306,
|
| 15 |
+
"perplexity": 18.933807981996924
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"start_position": 1025,
|
| 19 |
+
"end_position": 2048,
|
| 20 |
+
"tokens": 20480,
|
| 21 |
+
"loss": 2.8075664043426514,
|
| 22 |
+
"perplexity": 16.56954553925456
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"start_position": 2049,
|
| 26 |
+
"end_position": 3072,
|
| 27 |
+
"tokens": 20480,
|
| 28 |
+
"loss": 2.9897894978523256,
|
| 29 |
+
"perplexity": 19.88149695324014
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"start_position": 3073,
|
| 33 |
+
"end_position": 4096,
|
| 34 |
+
"tokens": 20480,
|
| 35 |
+
"loss": 2.9651451110839844,
|
| 36 |
+
"perplexity": 19.397517832091484
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"start_position": 4097,
|
| 40 |
+
"end_position": 5119,
|
| 41 |
+
"tokens": 20460,
|
| 42 |
+
"loss": 2.9493477531425647,
|
| 43 |
+
"perplexity": 19.093495993155035
|
| 44 |
+
}
|
| 45 |
+
]
|
| 46 |
+
}
|
reports/fixed-ko-cpt-3b-5k/train.log
ADDED
|
@@ -0,0 +1,532 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"parameters": 51542528,
|
| 3 |
+
"device": "cuda:0",
|
| 4 |
+
"world_size": 1,
|
| 5 |
+
"sequence_length": 5120,
|
| 6 |
+
"training_sample_offset": 0,
|
| 7 |
+
"tokens_per_optimizer_step": 122880,
|
| 8 |
+
"max_steps": 24415,
|
| 9 |
+
"planned_tokens": 3000115200,
|
| 10 |
+
"start_step": 0,
|
| 11 |
+
"initialized_from": "/workspace/KAWK50M/runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00038147",
|
| 12 |
+
"precision": "bf16",
|
| 13 |
+
"compile": true
|
| 14 |
+
}
|
| 15 |
+
{"step": 1, "max_steps": 24415, "train_loss": 8.144108, "learning_rate": 3.0737704918032784e-07, "grad_norm": 65.61449432373047, "tokens_seen": 122880, "tokens_per_second": 36373}
|
| 16 |
+
{"step": 50, "max_steps": 24415, "train_loss": 5.261077, "learning_rate": 1.5368852459016393e-05, "grad_norm": 2.527670383453369, "tokens_seen": 6144000, "tokens_per_second": 115796}
|
| 17 |
+
{"step": 100, "max_steps": 24415, "train_loss": 4.557536, "learning_rate": 3.073770491803279e-05, "grad_norm": 2.5450265407562256, "tokens_seen": 12288000, "tokens_per_second": 115654}
|
| 18 |
+
{"step": 150, "max_steps": 24415, "train_loss": 3.998088, "learning_rate": 4.610655737704918e-05, "grad_norm": 2.4517602920532227, "tokens_seen": 18432000, "tokens_per_second": 115644}
|
| 19 |
+
{"step": 200, "max_steps": 24415, "train_loss": 3.411813, "learning_rate": 6.147540983606557e-05, "grad_norm": 1.9274886846542358, "tokens_seen": 24576000, "tokens_per_second": 115607}
|
| 20 |
+
{"step": 250, "max_steps": 24415, "train_loss": 3.153503, "learning_rate": 7.684426229508196e-05, "grad_norm": 1.9783965349197388, "tokens_seen": 30720000, "tokens_per_second": 115566}
|
| 21 |
+
{"step": 300, "max_steps": 24415, "train_loss": 3.199131, "learning_rate": 9.221311475409836e-05, "grad_norm": 1.3728467226028442, "tokens_seen": 36864000, "tokens_per_second": 115581}
|
| 22 |
+
{"step": 350, "max_steps": 24415, "train_loss": 3.044291, "learning_rate": 0.00010758196721311473, "grad_norm": 1.6034692525863647, "tokens_seen": 43008000, "tokens_per_second": 115596}
|
| 23 |
+
{"step": 400, "max_steps": 24415, "train_loss": 3.029246, "learning_rate": 0.00012295081967213115, "grad_norm": 1.4586750268936157, "tokens_seen": 49152000, "tokens_per_second": 115629}
|
| 24 |
+
{"step": 450, "max_steps": 24415, "train_loss": 2.942982, "learning_rate": 0.00013831967213114753, "grad_norm": 0.9969210028648376, "tokens_seen": 55296000, "tokens_per_second": 115644}
|
| 25 |
+
{"step": 500, "max_steps": 24415, "train_loss": 3.001114, "learning_rate": 0.00014999991621632118, "grad_norm": 1.1246283054351807, "tokens_seen": 61440000, "tokens_per_second": 115624}
|
| 26 |
+
{"step": 550, "max_steps": 24415, "train_loss": 2.900805, "learning_rate": 0.00014999776345312908, "grad_norm": 1.0763027667999268, "tokens_seen": 67584000, "tokens_per_second": 115617}
|
| 27 |
+
{"step": 600, "max_steps": 24415, "train_loss": 2.919337, "learning_rate": 0.00014999270164066104, "grad_norm": 0.8842090368270874, "tokens_seen": 73728000, "tokens_per_second": 115626}
|
| 28 |
+
{"step": 650, "max_steps": 24415, "train_loss": 2.896405, "learning_rate": 0.00014998473099707337, "grad_norm": 0.8379791378974915, "tokens_seen": 79872000, "tokens_per_second": 115637}
|
| 29 |
+
{"step": 700, "max_steps": 24415, "train_loss": 2.799349, "learning_rate": 0.00014997385186588848, "grad_norm": 0.8146543502807617, "tokens_seen": 86016000, "tokens_per_second": 115635}
|
| 30 |
+
{"step": 750, "max_steps": 24415, "train_loss": 2.928234, "learning_rate": 0.00014996006471598011, "grad_norm": 0.8336044549942017, "tokens_seen": 92160000, "tokens_per_second": 115613}
|
| 31 |
+
{"step": 800, "max_steps": 24415, "train_loss": 2.825323, "learning_rate": 0.00014994337014155313, "grad_norm": 0.9470697045326233, "tokens_seen": 98304000, "tokens_per_second": 115612}
|
| 32 |
+
{"step": 850, "max_steps": 24415, "train_loss": 2.879508, "learning_rate": 0.00014992376886211787, "grad_norm": 0.8570364713668823, "tokens_seen": 104448000, "tokens_per_second": 115616}
|
| 33 |
+
{"step": 900, "max_steps": 24415, "train_loss": 2.905575, "learning_rate": 0.0001499012617224592, "grad_norm": 0.7632439136505127, "tokens_seen": 110592000, "tokens_per_second": 115640}
|
| 34 |
+
{"step": 950, "max_steps": 24415, "train_loss": 2.864619, "learning_rate": 0.00014987584969259997, "grad_norm": 0.7789876461029053, "tokens_seen": 116736000, "tokens_per_second": 115639}
|
| 35 |
+
{"step": 1000, "max_steps": 24415, "train_loss": 2.931551, "learning_rate": 0.0001498475338677595, "grad_norm": 0.8311377167701721, "tokens_seen": 122880000, "tokens_per_second": 115646}
|
| 36 |
+
{"step": 1000, "validation_loss": 2.96824, "validation_perplexity": 19.4577}
|
| 37 |
+
{"step": 1050, "max_steps": 24415, "train_loss": 2.925203, "learning_rate": 0.00014981631546830604, "grad_norm": 0.7916623950004578, "tokens_seen": 129024000, "tokens_per_second": 109465}
|
| 38 |
+
{"step": 1100, "max_steps": 24415, "train_loss": 2.899991, "learning_rate": 0.00014978219583970429, "grad_norm": 0.7494450211524963, "tokens_seen": 135168000, "tokens_per_second": 115661}
|
| 39 |
+
{"step": 1150, "max_steps": 24415, "train_loss": 2.859691, "learning_rate": 0.00014974517645245758, "grad_norm": 0.8097645044326782, "tokens_seen": 141312000, "tokens_per_second": 115689}
|
| 40 |
+
{"step": 1200, "max_steps": 24415, "train_loss": 2.78326, "learning_rate": 0.00014970525890204428, "grad_norm": 0.8151496052742004, "tokens_seen": 147456000, "tokens_per_second": 115671}
|
| 41 |
+
{"step": 1250, "max_steps": 24415, "train_loss": 2.912475, "learning_rate": 0.00014966244490884908, "grad_norm": 0.7340596914291382, "tokens_seen": 153600000, "tokens_per_second": 115662}
|
| 42 |
+
{"step": 1300, "max_steps": 24415, "train_loss": 2.836194, "learning_rate": 0.0001496167363180889, "grad_norm": 0.869087278842926, "tokens_seen": 159744000, "tokens_per_second": 115640}
|
| 43 |
+
{"step": 1350, "max_steps": 24415, "train_loss": 2.880843, "learning_rate": 0.00014956813509973336, "grad_norm": 0.9172841310501099, "tokens_seen": 165888000, "tokens_per_second": 115639}
|
| 44 |
+
{"step": 1400, "max_steps": 24415, "train_loss": 2.907881, "learning_rate": 0.00014951664334841988, "grad_norm": 0.8426065444946289, "tokens_seen": 172032000, "tokens_per_second": 115641}
|
| 45 |
+
{"step": 1450, "max_steps": 24415, "train_loss": 2.75049, "learning_rate": 0.00014946226328336332, "grad_norm": 0.7333223819732666, "tokens_seen": 178176000, "tokens_per_second": 115649}
|
| 46 |
+
{"step": 1500, "max_steps": 24415, "train_loss": 2.902985, "learning_rate": 0.00014940499724826046, "grad_norm": 0.6550875306129456, "tokens_seen": 184320000, "tokens_per_second": 115621}
|
| 47 |
+
{"step": 1550, "max_steps": 24415, "train_loss": 2.808746, "learning_rate": 0.00014934484771118885, "grad_norm": 0.9193676114082336, "tokens_seen": 190464000, "tokens_per_second": 115628}
|
| 48 |
+
{"step": 1600, "max_steps": 24415, "train_loss": 2.813932, "learning_rate": 0.00014928181726450062, "grad_norm": 0.6951905488967896, "tokens_seen": 196608000, "tokens_per_second": 115653}
|
| 49 |
+
{"step": 1650, "max_steps": 24415, "train_loss": 2.858552, "learning_rate": 0.0001492159086247106, "grad_norm": 0.8277786374092102, "tokens_seen": 202752000, "tokens_per_second": 115640}
|
| 50 |
+
{"step": 1700, "max_steps": 24415, "train_loss": 2.82544, "learning_rate": 0.00014914712463237923, "grad_norm": 0.6503491997718811, "tokens_seen": 208896000, "tokens_per_second": 115643}
|
| 51 |
+
{"step": 1750, "max_steps": 24415, "train_loss": 2.93095, "learning_rate": 0.00014907546825199033, "grad_norm": 0.6746272444725037, "tokens_seen": 215040000, "tokens_per_second": 115593}
|
| 52 |
+
{"step": 1800, "max_steps": 24415, "train_loss": 2.918631, "learning_rate": 0.00014900094257182313, "grad_norm": 0.681274950504303, "tokens_seen": 221184000, "tokens_per_second": 115624}
|
| 53 |
+
{"step": 1850, "max_steps": 24415, "train_loss": 2.732643, "learning_rate": 0.0001489235508038192, "grad_norm": 1.0000346899032593, "tokens_seen": 227328000, "tokens_per_second": 115644}
|
| 54 |
+
{"step": 1900, "max_steps": 24415, "train_loss": 2.784161, "learning_rate": 0.00014884329628344416, "grad_norm": 0.7131108045578003, "tokens_seen": 233472000, "tokens_per_second": 115619}
|
| 55 |
+
{"step": 1950, "max_steps": 24415, "train_loss": 2.867771, "learning_rate": 0.0001487601824695438, "grad_norm": 0.6672623157501221, "tokens_seen": 239616000, "tokens_per_second": 115612}
|
| 56 |
+
{"step": 2000, "max_steps": 24415, "train_loss": 2.88868, "learning_rate": 0.000148674212944195, "grad_norm": 0.6188545227050781, "tokens_seen": 245760000, "tokens_per_second": 115613}
|
| 57 |
+
{"step": 2000, "validation_loss": 2.957756, "validation_perplexity": 19.2547}
|
| 58 |
+
{"step": 2050, "max_steps": 24415, "train_loss": 2.848936, "learning_rate": 0.00014858539141255145, "grad_norm": 0.6768792867660522, "tokens_seen": 251904000, "tokens_per_second": 113459}
|
| 59 |
+
{"step": 2100, "max_steps": 24415, "train_loss": 2.911329, "learning_rate": 0.00014849372170268386, "grad_norm": 0.7390404343605042, "tokens_seen": 258048000, "tokens_per_second": 115657}
|
| 60 |
+
{"step": 2150, "max_steps": 24415, "train_loss": 2.7573, "learning_rate": 0.00014839920776541505, "grad_norm": 0.7121871709823608, "tokens_seen": 264192000, "tokens_per_second": 115647}
|
| 61 |
+
{"step": 2200, "max_steps": 24415, "train_loss": 2.712324, "learning_rate": 0.00014830185367414961, "grad_norm": 0.6592078804969788, "tokens_seen": 270336000, "tokens_per_second": 115660}
|
| 62 |
+
{"step": 2250, "max_steps": 24415, "train_loss": 2.879639, "learning_rate": 0.00014820166362469842, "grad_norm": 0.7398443222045898, "tokens_seen": 276480000, "tokens_per_second": 115657}
|
| 63 |
+
{"step": 2300, "max_steps": 24415, "train_loss": 2.785928, "learning_rate": 0.0001480986419350977, "grad_norm": 0.7555476427078247, "tokens_seen": 282624000, "tokens_per_second": 115677}
|
| 64 |
+
{"step": 2350, "max_steps": 24415, "train_loss": 2.75432, "learning_rate": 0.00014799279304542314, "grad_norm": 0.7762490510940552, "tokens_seen": 288768000, "tokens_per_second": 115702}
|
| 65 |
+
{"step": 2400, "max_steps": 24415, "train_loss": 2.971488, "learning_rate": 0.00014788412151759816, "grad_norm": 0.64515620470047, "tokens_seen": 294912000, "tokens_per_second": 115659}
|
| 66 |
+
{"step": 2450, "max_steps": 24415, "train_loss": 2.837723, "learning_rate": 0.00014777263203519768, "grad_norm": 0.7510269284248352, "tokens_seen": 301056000, "tokens_per_second": 115677}
|
| 67 |
+
{"step": 2500, "max_steps": 24415, "train_loss": 2.888861, "learning_rate": 0.00014765832940324614, "grad_norm": 0.6924780607223511, "tokens_seen": 307200000, "tokens_per_second": 115656}
|
| 68 |
+
{"step": 2550, "max_steps": 24415, "train_loss": 2.793462, "learning_rate": 0.00014754121854801018, "grad_norm": 0.69148188829422, "tokens_seen": 313344000, "tokens_per_second": 115646}
|
| 69 |
+
{"step": 2600, "max_steps": 24415, "train_loss": 2.906426, "learning_rate": 0.00014742130451678674, "grad_norm": 0.7049274444580078, "tokens_seen": 319488000, "tokens_per_second": 115640}
|
| 70 |
+
{"step": 2650, "max_steps": 24415, "train_loss": 2.913443, "learning_rate": 0.00014729859247768516, "grad_norm": 0.682880163192749, "tokens_seen": 325632000, "tokens_per_second": 115638}
|
| 71 |
+
{"step": 2700, "max_steps": 24415, "train_loss": 2.857629, "learning_rate": 0.00014717308771940472, "grad_norm": 0.6135403513908386, "tokens_seen": 331776000, "tokens_per_second": 115655}
|
| 72 |
+
{"step": 2750, "max_steps": 24415, "train_loss": 2.883287, "learning_rate": 0.00014704479565100643, "grad_norm": 0.6782547831535339, "tokens_seen": 337920000, "tokens_per_second": 115658}
|
| 73 |
+
{"step": 2800, "max_steps": 24415, "train_loss": 2.860058, "learning_rate": 0.0001469137218016802, "grad_norm": 0.7186660766601562, "tokens_seen": 344064000, "tokens_per_second": 115661}
|
| 74 |
+
{"step": 2850, "max_steps": 24415, "train_loss": 2.853855, "learning_rate": 0.00014677987182050632, "grad_norm": 0.6767726540565491, "tokens_seen": 350208000, "tokens_per_second": 115653}
|
| 75 |
+
{"step": 2900, "max_steps": 24415, "train_loss": 2.86035, "learning_rate": 0.00014664325147621213, "grad_norm": 0.6750748753547668, "tokens_seen": 356352000, "tokens_per_second": 115655}
|
| 76 |
+
{"step": 2950, "max_steps": 24415, "train_loss": 2.855024, "learning_rate": 0.00014650386665692323, "grad_norm": 0.7292869091033936, "tokens_seen": 362496000, "tokens_per_second": 115648}
|
| 77 |
+
{"step": 3000, "max_steps": 24415, "train_loss": 2.808876, "learning_rate": 0.00014636172336990991, "grad_norm": 0.6982874274253845, "tokens_seen": 368640000, "tokens_per_second": 115667}
|
| 78 |
+
{"step": 3000, "validation_loss": 2.950226, "validation_perplexity": 19.1103}
|
| 79 |
+
{"step": 3050, "max_steps": 24415, "train_loss": 2.787679, "learning_rate": 0.00014621682774132818, "grad_norm": 0.7349603772163391, "tokens_seen": 374784000, "tokens_per_second": 113471}
|
| 80 |
+
{"step": 3100, "max_steps": 24415, "train_loss": 2.748469, "learning_rate": 0.00014606918601595562, "grad_norm": 0.8711739182472229, "tokens_seen": 380928000, "tokens_per_second": 115645}
|
| 81 |
+
{"step": 3150, "max_steps": 24415, "train_loss": 2.822111, "learning_rate": 0.00014591880455692242, "grad_norm": 0.722649097442627, "tokens_seen": 387072000, "tokens_per_second": 115647}
|
| 82 |
+
{"step": 3200, "max_steps": 24415, "train_loss": 2.784047, "learning_rate": 0.00014576568984543702, "grad_norm": 0.675203800201416, "tokens_seen": 393216000, "tokens_per_second": 115674}
|
| 83 |
+
{"step": 3250, "max_steps": 24415, "train_loss": 2.812503, "learning_rate": 0.00014560984848050682, "grad_norm": 0.6620163321495056, "tokens_seen": 399360000, "tokens_per_second": 115674}
|
| 84 |
+
{"step": 3300, "max_steps": 24415, "train_loss": 2.851004, "learning_rate": 0.00014545128717865377, "grad_norm": 0.6590436100959778, "tokens_seen": 405504000, "tokens_per_second": 115663}
|
| 85 |
+
{"step": 3350, "max_steps": 24415, "train_loss": 2.666078, "learning_rate": 0.0001452900127736249, "grad_norm": 0.7361322045326233, "tokens_seen": 411648000, "tokens_per_second": 115664}
|
| 86 |
+
{"step": 3400, "max_steps": 24415, "train_loss": 2.847474, "learning_rate": 0.00014512603221609774, "grad_norm": 0.6425556540489197, "tokens_seen": 417792000, "tokens_per_second": 115646}
|
| 87 |
+
{"step": 3450, "max_steps": 24415, "train_loss": 2.894404, "learning_rate": 0.00014495935257338095, "grad_norm": 0.687091052532196, "tokens_seen": 423936000, "tokens_per_second": 115651}
|
| 88 |
+
{"step": 3500, "max_steps": 24415, "train_loss": 2.785236, "learning_rate": 0.00014478998102910944, "grad_norm": 0.6465118527412415, "tokens_seen": 430080000, "tokens_per_second": 115671}
|
| 89 |
+
{"step": 3550, "max_steps": 24415, "train_loss": 2.781479, "learning_rate": 0.00014461792488293496, "grad_norm": 0.7730088829994202, "tokens_seen": 436224000, "tokens_per_second": 115674}
|
| 90 |
+
{"step": 3600, "max_steps": 24415, "train_loss": 2.919193, "learning_rate": 0.00014444319155021152, "grad_norm": 0.6525118947029114, "tokens_seen": 442368000, "tokens_per_second": 115663}
|
| 91 |
+
{"step": 3650, "max_steps": 24415, "train_loss": 2.847713, "learning_rate": 0.00014426578856167568, "grad_norm": 0.737187922000885, "tokens_seen": 448512000, "tokens_per_second": 115661}
|
| 92 |
+
{"step": 3700, "max_steps": 24415, "train_loss": 2.915229, "learning_rate": 0.00014408572356312203, "grad_norm": 0.6132097244262695, "tokens_seen": 454656000, "tokens_per_second": 115686}
|
| 93 |
+
{"step": 3750, "max_steps": 24415, "train_loss": 2.85277, "learning_rate": 0.00014390300431507366, "grad_norm": 0.6486119031906128, "tokens_seen": 460800000, "tokens_per_second": 115701}
|
| 94 |
+
{"step": 3800, "max_steps": 24415, "train_loss": 2.921737, "learning_rate": 0.00014371763869244775, "grad_norm": 0.8457760214805603, "tokens_seen": 466944000, "tokens_per_second": 115692}
|
| 95 |
+
{"step": 3850, "max_steps": 24415, "train_loss": 2.825371, "learning_rate": 0.00014352963468421614, "grad_norm": 0.7568368315696716, "tokens_seen": 473088000, "tokens_per_second": 115654}
|
| 96 |
+
{"step": 3900, "max_steps": 24415, "train_loss": 2.780189, "learning_rate": 0.00014333900039306097, "grad_norm": 0.7013327479362488, "tokens_seen": 479232000, "tokens_per_second": 115642}
|
| 97 |
+
{"step": 3950, "max_steps": 24415, "train_loss": 2.902723, "learning_rate": 0.00014314574403502555, "grad_norm": 0.6669299602508545, "tokens_seen": 485376000, "tokens_per_second": 115658}
|
| 98 |
+
{"step": 4000, "max_steps": 24415, "train_loss": 2.750378, "learning_rate": 0.0001429498739391602, "grad_norm": 0.6626065969467163, "tokens_seen": 491520000, "tokens_per_second": 115649}
|
| 99 |
+
{"step": 4000, "validation_loss": 2.948161, "validation_perplexity": 19.0709}
|
| 100 |
+
{"step": 4050, "max_steps": 24415, "train_loss": 2.874746, "learning_rate": 0.0001427513985471633, "grad_norm": 0.7452462315559387, "tokens_seen": 497664000, "tokens_per_second": 113529}
|
| 101 |
+
{"step": 4100, "max_steps": 24415, "train_loss": 2.850423, "learning_rate": 0.0001425503264130175, "grad_norm": 0.6568276286125183, "tokens_seen": 503808000, "tokens_per_second": 115665}
|
| 102 |
+
{"step": 4150, "max_steps": 24415, "train_loss": 2.802352, "learning_rate": 0.0001423466662026209, "grad_norm": 0.6452341675758362, "tokens_seen": 509952000, "tokens_per_second": 115661}
|
| 103 |
+
{"step": 4200, "max_steps": 24415, "train_loss": 2.848085, "learning_rate": 0.00014214042669341388, "grad_norm": 0.6572521924972534, "tokens_seen": 516096000, "tokens_per_second": 115640}
|
| 104 |
+
{"step": 4250, "max_steps": 24415, "train_loss": 2.969786, "learning_rate": 0.00014193161677400048, "grad_norm": 0.7090214490890503, "tokens_seen": 522240000, "tokens_per_second": 115650}
|
| 105 |
+
{"step": 4300, "max_steps": 24415, "train_loss": 2.826574, "learning_rate": 0.00014172024544376543, "grad_norm": 0.6470843553543091, "tokens_seen": 528384000, "tokens_per_second": 115670}
|
| 106 |
+
{"step": 4350, "max_steps": 24415, "train_loss": 2.935181, "learning_rate": 0.00014150632181248642, "grad_norm": 0.6939204335212708, "tokens_seen": 534528000, "tokens_per_second": 115678}
|
| 107 |
+
{"step": 4400, "max_steps": 24415, "train_loss": 2.870769, "learning_rate": 0.00014128985509994122, "grad_norm": 0.6737850308418274, "tokens_seen": 540672000, "tokens_per_second": 115669}
|
| 108 |
+
{"step": 4450, "max_steps": 24415, "train_loss": 3.00332, "learning_rate": 0.00014107085463551064, "grad_norm": 0.6614698171615601, "tokens_seen": 546816000, "tokens_per_second": 115649}
|
| 109 |
+
{"step": 4500, "max_steps": 24415, "train_loss": 2.696218, "learning_rate": 0.00014084932985777618, "grad_norm": 0.6942455768585205, "tokens_seen": 552960000, "tokens_per_second": 115658}
|
| 110 |
+
{"step": 4550, "max_steps": 24415, "train_loss": 2.828182, "learning_rate": 0.00014062529031411331, "grad_norm": 0.6730613708496094, "tokens_seen": 559104000, "tokens_per_second": 115665}
|
| 111 |
+
{"step": 4600, "max_steps": 24415, "train_loss": 2.920113, "learning_rate": 0.00014039874566028015, "grad_norm": 0.6104084253311157, "tokens_seen": 565248000, "tokens_per_second": 115640}
|
| 112 |
+
{"step": 4650, "max_steps": 24415, "train_loss": 2.754896, "learning_rate": 0.00014016970566000102, "grad_norm": 0.6483374834060669, "tokens_seen": 571392000, "tokens_per_second": 115647}
|
| 113 |
+
{"step": 4700, "max_steps": 24415, "train_loss": 2.813306, "learning_rate": 0.00013993818018454598, "grad_norm": 0.6836441159248352, "tokens_seen": 577536000, "tokens_per_second": 115660}
|
| 114 |
+
{"step": 4750, "max_steps": 24415, "train_loss": 2.903481, "learning_rate": 0.00013970417921230514, "grad_norm": 0.656783401966095, "tokens_seen": 583680000, "tokens_per_second": 115651}
|
| 115 |
+
{"step": 4800, "max_steps": 24415, "train_loss": 3.033902, "learning_rate": 0.00013946771282835873, "grad_norm": 0.7004533410072327, "tokens_seen": 589824000, "tokens_per_second": 115658}
|
| 116 |
+
{"step": 4850, "max_steps": 24415, "train_loss": 2.867407, "learning_rate": 0.0001392287912240424, "grad_norm": 0.6720044612884521, "tokens_seen": 595968000, "tokens_per_second": 115647}
|
| 117 |
+
{"step": 4900, "max_steps": 24415, "train_loss": 2.976752, "learning_rate": 0.00013898742469650797, "grad_norm": 0.7038782835006714, "tokens_seen": 602112000, "tokens_per_second": 115666}
|
| 118 |
+
{"step": 4950, "max_steps": 24415, "train_loss": 2.880877, "learning_rate": 0.00013874362364827976, "grad_norm": 0.6553775072097778, "tokens_seen": 608256000, "tokens_per_second": 115658}
|
| 119 |
+
{"step": 5000, "max_steps": 24415, "train_loss": 2.872165, "learning_rate": 0.00013849739858680607, "grad_norm": 0.6521583795547485, "tokens_seen": 614400000, "tokens_per_second": 115656}
|
| 120 |
+
{"step": 5000, "validation_loss": 2.944971, "validation_perplexity": 19.0101}
|
| 121 |
+
Saved checkpoint: runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00005000
|
| 122 |
+
{"step": 5050, "max_steps": 24415, "train_loss": 2.799103, "learning_rate": 0.00013824876012400648, "grad_norm": 0.6628615260124207, "tokens_seen": 620544000, "tokens_per_second": 111699}
|
| 123 |
+
{"step": 5100, "max_steps": 24415, "train_loss": 2.91504, "learning_rate": 0.00013799771897581443, "grad_norm": 0.6126322746276855, "tokens_seen": 626688000, "tokens_per_second": 115663}
|
| 124 |
+
{"step": 5150, "max_steps": 24415, "train_loss": 2.882699, "learning_rate": 0.00013774428596171538, "grad_norm": 0.6810411214828491, "tokens_seen": 632832000, "tokens_per_second": 115645}
|
| 125 |
+
{"step": 5200, "max_steps": 24415, "train_loss": 2.850346, "learning_rate": 0.0001374884720042805, "grad_norm": 0.6813170909881592, "tokens_seen": 638976000, "tokens_per_second": 115652}
|
| 126 |
+
{"step": 5250, "max_steps": 24415, "train_loss": 2.856393, "learning_rate": 0.00013723028812869602, "grad_norm": 0.619038462638855, "tokens_seen": 645120000, "tokens_per_second": 115651}
|
| 127 |
+
{"step": 5300, "max_steps": 24415, "train_loss": 2.843284, "learning_rate": 0.00013696974546228781, "grad_norm": 0.6991771459579468, "tokens_seen": 651264000, "tokens_per_second": 115651}
|
| 128 |
+
{"step": 5350, "max_steps": 24415, "train_loss": 2.908741, "learning_rate": 0.00013670685523404214, "grad_norm": 0.6153818368911743, "tokens_seen": 657408000, "tokens_per_second": 115647}
|
| 129 |
+
{"step": 5400, "max_steps": 24415, "train_loss": 2.931608, "learning_rate": 0.00013644162877412153, "grad_norm": 0.6771771907806396, "tokens_seen": 663552000, "tokens_per_second": 115640}
|
| 130 |
+
{"step": 5450, "max_steps": 24415, "train_loss": 2.815716, "learning_rate": 0.00013617407751337636, "grad_norm": 0.8512451648712158, "tokens_seen": 669696000, "tokens_per_second": 115659}
|
| 131 |
+
{"step": 5500, "max_steps": 24415, "train_loss": 2.867514, "learning_rate": 0.00013590421298285252, "grad_norm": 0.6467471122741699, "tokens_seen": 675840000, "tokens_per_second": 115656}
|
| 132 |
+
{"step": 5550, "max_steps": 24415, "train_loss": 2.826002, "learning_rate": 0.0001356320468132941, "grad_norm": 0.6404849290847778, "tokens_seen": 681984000, "tokens_per_second": 115648}
|
| 133 |
+
{"step": 5600, "max_steps": 24415, "train_loss": 2.87931, "learning_rate": 0.0001353575907346424, "grad_norm": 0.6391315460205078, "tokens_seen": 688128000, "tokens_per_second": 115627}
|
| 134 |
+
{"step": 5650, "max_steps": 24415, "train_loss": 2.75582, "learning_rate": 0.00013508085657553008, "grad_norm": 0.6846815943717957, "tokens_seen": 694272000, "tokens_per_second": 115641}
|
| 135 |
+
{"step": 5700, "max_steps": 24415, "train_loss": 2.772559, "learning_rate": 0.00013480185626277175, "grad_norm": 0.7212074995040894, "tokens_seen": 700416000, "tokens_per_second": 115620}
|
| 136 |
+
{"step": 5750, "max_steps": 24415, "train_loss": 2.822725, "learning_rate": 0.0001345206018208496, "grad_norm": 0.599242627620697, "tokens_seen": 706560000, "tokens_per_second": 115654}
|
| 137 |
+
{"step": 5800, "max_steps": 24415, "train_loss": 2.858205, "learning_rate": 0.0001342371053713954, "grad_norm": 0.7058699727058411, "tokens_seen": 712704000, "tokens_per_second": 115668}
|
| 138 |
+
{"step": 5850, "max_steps": 24415, "train_loss": 2.982627, "learning_rate": 0.00013395137913266788, "grad_norm": 0.6416256427764893, "tokens_seen": 718848000, "tokens_per_second": 115671}
|
| 139 |
+
{"step": 5900, "max_steps": 24415, "train_loss": 2.781853, "learning_rate": 0.00013366343541902634, "grad_norm": 0.7527068853378296, "tokens_seen": 724992000, "tokens_per_second": 115660}
|
| 140 |
+
{"step": 5950, "max_steps": 24415, "train_loss": 2.83377, "learning_rate": 0.0001333732866403997, "grad_norm": 0.6535189151763916, "tokens_seen": 731136000, "tokens_per_second": 115677}
|
| 141 |
+
{"step": 6000, "max_steps": 24415, "train_loss": 2.767776, "learning_rate": 0.00013308094530175193, "grad_norm": 0.8586512207984924, "tokens_seen": 737280000, "tokens_per_second": 115664}
|
| 142 |
+
{"step": 6000, "validation_loss": 2.940371, "validation_perplexity": 18.9229}
|
| 143 |
+
{"step": 6050, "max_steps": 24415, "train_loss": 2.885991, "learning_rate": 0.00013278642400254285, "grad_norm": 0.6209027767181396, "tokens_seen": 743424000, "tokens_per_second": 113498}
|
| 144 |
+
{"step": 6100, "max_steps": 24415, "train_loss": 2.79999, "learning_rate": 0.00013248973543618516, "grad_norm": 0.622372031211853, "tokens_seen": 749568000, "tokens_per_second": 115650}
|
| 145 |
+
{"step": 6150, "max_steps": 24415, "train_loss": 2.847553, "learning_rate": 0.00013219089238949755, "grad_norm": 0.6702292561531067, "tokens_seen": 755712000, "tokens_per_second": 115648}
|
| 146 |
+
{"step": 6200, "max_steps": 24415, "train_loss": 2.88014, "learning_rate": 0.00013188990774215341, "grad_norm": 0.7778111696243286, "tokens_seen": 761856000, "tokens_per_second": 115667}
|
| 147 |
+
{"step": 6250, "max_steps": 24415, "train_loss": 2.876659, "learning_rate": 0.0001315867944661258, "grad_norm": 0.648591160774231, "tokens_seen": 768000000, "tokens_per_second": 115647}
|
| 148 |
+
{"step": 6300, "max_steps": 24415, "train_loss": 2.835785, "learning_rate": 0.00013128156562512843, "grad_norm": 0.6847606897354126, "tokens_seen": 774144000, "tokens_per_second": 115637}
|
| 149 |
+
{"step": 6350, "max_steps": 24415, "train_loss": 2.748451, "learning_rate": 0.00013097423437405254, "grad_norm": 0.7733544707298279, "tokens_seen": 780288000, "tokens_per_second": 115644}
|
| 150 |
+
{"step": 6400, "max_steps": 24415, "train_loss": 2.864969, "learning_rate": 0.00013066481395840004, "grad_norm": 0.6182960867881775, "tokens_seen": 786432000, "tokens_per_second": 115636}
|
| 151 |
+
{"step": 6450, "max_steps": 24415, "train_loss": 2.830928, "learning_rate": 0.0001303533177137126, "grad_norm": 0.6351659893989563, "tokens_seen": 792576000, "tokens_per_second": 115642}
|
| 152 |
+
{"step": 6500, "max_steps": 24415, "train_loss": 2.778158, "learning_rate": 0.00013003975906499685, "grad_norm": 0.6366986632347107, "tokens_seen": 798720000, "tokens_per_second": 115654}
|
| 153 |
+
{"step": 6550, "max_steps": 24415, "train_loss": 2.641845, "learning_rate": 0.00012972415152614594, "grad_norm": 0.9436250329017639, "tokens_seen": 804864000, "tokens_per_second": 115641}
|
| 154 |
+
{"step": 6600, "max_steps": 24415, "train_loss": 2.877358, "learning_rate": 0.00012940650869935693, "grad_norm": 0.6089520454406738, "tokens_seen": 811008000, "tokens_per_second": 115625}
|
| 155 |
+
{"step": 6650, "max_steps": 24415, "train_loss": 2.90905, "learning_rate": 0.00012908684427454472, "grad_norm": 0.6610862612724304, "tokens_seen": 817152000, "tokens_per_second": 115629}
|
| 156 |
+
{"step": 6700, "max_steps": 24415, "train_loss": 2.787947, "learning_rate": 0.0001287651720287519, "grad_norm": 0.7656141519546509, "tokens_seen": 823296000, "tokens_per_second": 115623}
|
| 157 |
+
{"step": 6750, "max_steps": 24415, "train_loss": 2.715172, "learning_rate": 0.00012844150582555508, "grad_norm": 0.6442705988883972, "tokens_seen": 829440000, "tokens_per_second": 115626}
|
| 158 |
+
{"step": 6800, "max_steps": 24415, "train_loss": 2.770439, "learning_rate": 0.00012811585961446735, "grad_norm": 0.6757336258888245, "tokens_seen": 835584000, "tokens_per_second": 115638}
|
| 159 |
+
{"step": 6850, "max_steps": 24415, "train_loss": 2.883263, "learning_rate": 0.00012778824743033703, "grad_norm": 0.624424159526825, "tokens_seen": 841728000, "tokens_per_second": 115659}
|
| 160 |
+
{"step": 6900, "max_steps": 24415, "train_loss": 2.866001, "learning_rate": 0.0001274586833927429, "grad_norm": 0.6462403535842896, "tokens_seen": 847872000, "tokens_per_second": 115663}
|
| 161 |
+
{"step": 6950, "max_steps": 24415, "train_loss": 2.759226, "learning_rate": 0.00012712718170538562, "grad_norm": 0.8008264899253845, "tokens_seen": 854016000, "tokens_per_second": 115657}
|
| 162 |
+
{"step": 7000, "max_steps": 24415, "train_loss": 2.854571, "learning_rate": 0.00012679375665547546, "grad_norm": 0.6955330967903137, "tokens_seen": 860160000, "tokens_per_second": 115619}
|
| 163 |
+
{"step": 7000, "validation_loss": 2.933616, "validation_perplexity": 18.7955}
|
| 164 |
+
{"step": 7050, "max_steps": 24415, "train_loss": 2.797329, "learning_rate": 0.00012645842261311676, "grad_norm": 0.7074103951454163, "tokens_seen": 866304000, "tokens_per_second": 113444}
|
| 165 |
+
{"step": 7100, "max_steps": 24415, "train_loss": 2.811077, "learning_rate": 0.0001261211940306884, "grad_norm": 0.6589587926864624, "tokens_seen": 872448000, "tokens_per_second": 115588}
|
| 166 |
+
{"step": 7150, "max_steps": 24415, "train_loss": 2.85789, "learning_rate": 0.00012578208544222104, "grad_norm": 0.6195741891860962, "tokens_seen": 878592000, "tokens_per_second": 115604}
|
| 167 |
+
{"step": 7200, "max_steps": 24415, "train_loss": 2.951097, "learning_rate": 0.00012544111146277072, "grad_norm": 0.7131823897361755, "tokens_seen": 884736000, "tokens_per_second": 115624}
|
| 168 |
+
{"step": 7250, "max_steps": 24415, "train_loss": 2.760288, "learning_rate": 0.0001250982867877889, "grad_norm": 0.6765594482421875, "tokens_seen": 890880000, "tokens_per_second": 115599}
|
| 169 |
+
{"step": 7300, "max_steps": 24415, "train_loss": 2.820308, "learning_rate": 0.0001247536261924892, "grad_norm": 0.6766009330749512, "tokens_seen": 897024000, "tokens_per_second": 115633}
|
| 170 |
+
{"step": 7350, "max_steps": 24415, "train_loss": 2.978635, "learning_rate": 0.00012440714453121053, "grad_norm": 0.7272297143936157, "tokens_seen": 903168000, "tokens_per_second": 115719}
|
| 171 |
+
{"step": 7400, "max_steps": 24415, "train_loss": 2.880204, "learning_rate": 0.00012405885673677698, "grad_norm": 0.6244488954544067, "tokens_seen": 909312000, "tokens_per_second": 115676}
|
| 172 |
+
{"step": 7450, "max_steps": 24415, "train_loss": 2.904444, "learning_rate": 0.0001237087778198541, "grad_norm": 0.6648251414299011, "tokens_seen": 915456000, "tokens_per_second": 115711}
|
| 173 |
+
{"step": 7500, "max_steps": 24415, "train_loss": 2.761484, "learning_rate": 0.0001233569228683021, "grad_norm": 0.6898263096809387, "tokens_seen": 921600000, "tokens_per_second": 115655}
|
| 174 |
+
{"step": 7550, "max_steps": 24415, "train_loss": 2.747628, "learning_rate": 0.0001230033070465256, "grad_norm": 0.9069489240646362, "tokens_seen": 927744000, "tokens_per_second": 115602}
|
| 175 |
+
{"step": 7600, "max_steps": 24415, "train_loss": 2.856875, "learning_rate": 0.00012264794559481993, "grad_norm": 0.6639710664749146, "tokens_seen": 933888000, "tokens_per_second": 115599}
|
| 176 |
+
{"step": 7650, "max_steps": 24415, "train_loss": 2.803813, "learning_rate": 0.00012229085382871438, "grad_norm": 0.6735744476318359, "tokens_seen": 940032000, "tokens_per_second": 115599}
|
| 177 |
+
{"step": 7700, "max_steps": 24415, "train_loss": 2.850425, "learning_rate": 0.00012193204713831213, "grad_norm": 0.6658360958099365, "tokens_seen": 946176000, "tokens_per_second": 115599}
|
| 178 |
+
{"step": 7750, "max_steps": 24415, "train_loss": 2.651114, "learning_rate": 0.00012157154098762691, "grad_norm": 0.696196436882019, "tokens_seen": 952320000, "tokens_per_second": 115594}
|
| 179 |
+
{"step": 7800, "max_steps": 24415, "train_loss": 2.810525, "learning_rate": 0.00012120935091391663, "grad_norm": 0.6725903749465942, "tokens_seen": 958464000, "tokens_per_second": 115602}
|
| 180 |
+
{"step": 7850, "max_steps": 24415, "train_loss": 2.753515, "learning_rate": 0.00012084549252701364, "grad_norm": 0.6148095726966858, "tokens_seen": 964608000, "tokens_per_second": 115589}
|
| 181 |
+
{"step": 7900, "max_steps": 24415, "train_loss": 2.743506, "learning_rate": 0.000120479981508652, "grad_norm": 0.6364421844482422, "tokens_seen": 970752000, "tokens_per_second": 115599}
|
| 182 |
+
{"step": 7950, "max_steps": 24415, "train_loss": 2.852497, "learning_rate": 0.00012011283361179166, "grad_norm": 0.6017224788665771, "tokens_seen": 976896000, "tokens_per_second": 115577}
|
| 183 |
+
{"step": 8000, "max_steps": 24415, "train_loss": 2.799852, "learning_rate": 0.00011974406465993947, "grad_norm": 0.6454333066940308, "tokens_seen": 983040000, "tokens_per_second": 115618}
|
| 184 |
+
{"step": 8000, "validation_loss": 2.930552, "validation_perplexity": 18.738}
|
| 185 |
+
{"step": 8050, "max_steps": 24415, "train_loss": 2.713696, "learning_rate": 0.00011937369054646724, "grad_norm": 0.7488113045692444, "tokens_seen": 989184000, "tokens_per_second": 113452}
|
| 186 |
+
{"step": 8100, "max_steps": 24415, "train_loss": 2.814722, "learning_rate": 0.00011900172723392684, "grad_norm": 0.5979893803596497, "tokens_seen": 995328000, "tokens_per_second": 115611}
|
| 187 |
+
{"step": 8150, "max_steps": 24415, "train_loss": 2.784884, "learning_rate": 0.00011862819075336208, "grad_norm": 0.6830214262008667, "tokens_seen": 1001472000, "tokens_per_second": 115602}
|
| 188 |
+
{"step": 8200, "max_steps": 24415, "train_loss": 2.891899, "learning_rate": 0.00011825309720361792, "grad_norm": 0.6417458653450012, "tokens_seen": 1007616000, "tokens_per_second": 115604}
|
| 189 |
+
{"step": 8250, "max_steps": 24415, "train_loss": 2.750473, "learning_rate": 0.00011787646275064659, "grad_norm": 0.6115127205848694, "tokens_seen": 1013760000, "tokens_per_second": 115613}
|
| 190 |
+
{"step": 8300, "max_steps": 24415, "train_loss": 2.830243, "learning_rate": 0.00011749830362681082, "grad_norm": 0.6408085227012634, "tokens_seen": 1019904000, "tokens_per_second": 115590}
|
| 191 |
+
{"step": 8350, "max_steps": 24415, "train_loss": 2.837465, "learning_rate": 0.00011711863613018439, "grad_norm": 0.6117357015609741, "tokens_seen": 1026048000, "tokens_per_second": 115559}
|
| 192 |
+
{"step": 8400, "max_steps": 24415, "train_loss": 2.806834, "learning_rate": 0.00011673747662384954, "grad_norm": 0.6075411438941956, "tokens_seen": 1032192000, "tokens_per_second": 115583}
|
| 193 |
+
{"step": 8450, "max_steps": 24415, "train_loss": 2.785034, "learning_rate": 0.00011635484153519188, "grad_norm": 0.6879437565803528, "tokens_seen": 1038336000, "tokens_per_second": 115682}
|
| 194 |
+
{"step": 8500, "max_steps": 24415, "train_loss": 2.706253, "learning_rate": 0.00011597074735519223, "grad_norm": 0.6767988204956055, "tokens_seen": 1044480000, "tokens_per_second": 115683}
|
| 195 |
+
{"step": 8550, "max_steps": 24415, "train_loss": 2.755598, "learning_rate": 0.00011558521063771618, "grad_norm": 0.6101839542388916, "tokens_seen": 1050624000, "tokens_per_second": 115676}
|
| 196 |
+
{"step": 8600, "max_steps": 24415, "train_loss": 2.693565, "learning_rate": 0.00011519824799880022, "grad_norm": 0.7641435265541077, "tokens_seen": 1056768000, "tokens_per_second": 115653}
|
| 197 |
+
{"step": 8650, "max_steps": 24415, "train_loss": 2.87077, "learning_rate": 0.0001148098761159361, "grad_norm": 0.6561196446418762, "tokens_seen": 1062912000, "tokens_per_second": 115597}
|
| 198 |
+
{"step": 8700, "max_steps": 24415, "train_loss": 2.79678, "learning_rate": 0.00011442011172735157, "grad_norm": 0.6285972595214844, "tokens_seen": 1069056000, "tokens_per_second": 115544}
|
| 199 |
+
{"step": 8750, "max_steps": 24415, "train_loss": 2.886481, "learning_rate": 0.00011402897163128938, "grad_norm": 0.6041844487190247, "tokens_seen": 1075200000, "tokens_per_second": 115560}
|
| 200 |
+
{"step": 8800, "max_steps": 24415, "train_loss": 2.832706, "learning_rate": 0.00011363647268528311, "grad_norm": 0.6426976919174194, "tokens_seen": 1081344000, "tokens_per_second": 115582}
|
| 201 |
+
{"step": 8850, "max_steps": 24415, "train_loss": 2.83513, "learning_rate": 0.00011324263180543064, "grad_norm": 0.7424055337905884, "tokens_seen": 1087488000, "tokens_per_second": 115581}
|
| 202 |
+
{"step": 8900, "max_steps": 24415, "train_loss": 2.816452, "learning_rate": 0.00011284746596566514, "grad_norm": 0.7485107779502869, "tokens_seen": 1093632000, "tokens_per_second": 115567}
|
| 203 |
+
{"step": 8950, "max_steps": 24415, "train_loss": 2.680914, "learning_rate": 0.00011245099219702354, "grad_norm": 0.6200953722000122, "tokens_seen": 1099776000, "tokens_per_second": 115559}
|
| 204 |
+
{"step": 9000, "max_steps": 24415, "train_loss": 2.7827, "learning_rate": 0.00011205322758691243, "grad_norm": 0.6235812306404114, "tokens_seen": 1105920000, "tokens_per_second": 115537}
|
| 205 |
+
{"step": 9000, "validation_loss": 2.923473, "validation_perplexity": 18.6058}
|
| 206 |
+
{"step": 9050, "max_steps": 24415, "train_loss": 2.857782, "learning_rate": 0.0001116541892783717, "grad_norm": 0.6333112120628357, "tokens_seen": 1112064000, "tokens_per_second": 113341}
|
| 207 |
+
{"step": 9100, "max_steps": 24415, "train_loss": 2.739355, "learning_rate": 0.00011125389446933568, "grad_norm": 0.6599845290184021, "tokens_seen": 1118208000, "tokens_per_second": 115497}
|
| 208 |
+
{"step": 9150, "max_steps": 24415, "train_loss": 2.742508, "learning_rate": 0.0001108523604118919, "grad_norm": 0.628777265548706, "tokens_seen": 1124352000, "tokens_per_second": 115493}
|
| 209 |
+
{"step": 9200, "max_steps": 24415, "train_loss": 2.775906, "learning_rate": 0.0001104496044115376, "grad_norm": 0.6229897737503052, "tokens_seen": 1130496000, "tokens_per_second": 115528}
|
| 210 |
+
{"step": 9250, "max_steps": 24415, "train_loss": 2.848124, "learning_rate": 0.00011004564382643386, "grad_norm": 0.6016040444374084, "tokens_seen": 1136640000, "tokens_per_second": 115619}
|
| 211 |
+
{"step": 9300, "max_steps": 24415, "train_loss": 2.68203, "learning_rate": 0.00010964049606665756, "grad_norm": 0.6977869868278503, "tokens_seen": 1142784000, "tokens_per_second": 115661}
|
| 212 |
+
{"step": 9350, "max_steps": 24415, "train_loss": 2.952381, "learning_rate": 0.00010923417859345088, "grad_norm": 0.7186052203178406, "tokens_seen": 1148928000, "tokens_per_second": 115641}
|
| 213 |
+
{"step": 9400, "max_steps": 24415, "train_loss": 2.80347, "learning_rate": 0.00010882670891846888, "grad_norm": 0.6344025731086731, "tokens_seen": 1155072000, "tokens_per_second": 115618}
|
| 214 |
+
{"step": 9450, "max_steps": 24415, "train_loss": 2.885839, "learning_rate": 0.00010841810460302477, "grad_norm": 0.606657862663269, "tokens_seen": 1161216000, "tokens_per_second": 115545}
|
| 215 |
+
{"step": 9500, "max_steps": 24415, "train_loss": 2.723036, "learning_rate": 0.00010800838325733296, "grad_norm": 0.6441790461540222, "tokens_seen": 1167360000, "tokens_per_second": 115506}
|
| 216 |
+
{"step": 9550, "max_steps": 24415, "train_loss": 2.873209, "learning_rate": 0.00010759756253975014, "grad_norm": 0.6934483051300049, "tokens_seen": 1173504000, "tokens_per_second": 115502}
|
| 217 |
+
{"step": 9600, "max_steps": 24415, "train_loss": 2.900885, "learning_rate": 0.0001071856601560143, "grad_norm": 0.6176964044570923, "tokens_seen": 1179648000, "tokens_per_second": 115514}
|
| 218 |
+
{"step": 9650, "max_steps": 24415, "train_loss": 2.848973, "learning_rate": 0.0001067726938584815, "grad_norm": 0.6434008479118347, "tokens_seen": 1185792000, "tokens_per_second": 115454}
|
| 219 |
+
{"step": 9700, "max_steps": 24415, "train_loss": 2.97445, "learning_rate": 0.00010635868144536079, "grad_norm": 0.612534761428833, "tokens_seen": 1191936000, "tokens_per_second": 115480}
|
| 220 |
+
{"step": 9750, "max_steps": 24415, "train_loss": 2.820954, "learning_rate": 0.00010594364075994735, "grad_norm": 0.7931225299835205, "tokens_seen": 1198080000, "tokens_per_second": 115473}
|
| 221 |
+
{"step": 9800, "max_steps": 24415, "train_loss": 2.845421, "learning_rate": 0.00010552758968985312, "grad_norm": 0.6815776228904724, "tokens_seen": 1204224000, "tokens_per_second": 115428}
|
| 222 |
+
{"step": 9850, "max_steps": 24415, "train_loss": 2.838773, "learning_rate": 0.00010511054616623621, "grad_norm": 0.5951747894287109, "tokens_seen": 1210368000, "tokens_per_second": 115374}
|
| 223 |
+
{"step": 9900, "max_steps": 24415, "train_loss": 2.772448, "learning_rate": 0.00010469252816302786, "grad_norm": 0.6392014622688293, "tokens_seen": 1216512000, "tokens_per_second": 115407}
|
| 224 |
+
{"step": 9950, "max_steps": 24415, "train_loss": 2.7407, "learning_rate": 0.00010427355369615789, "grad_norm": 0.6149389147758484, "tokens_seen": 1222656000, "tokens_per_second": 115586}
|
| 225 |
+
{"step": 10000, "max_steps": 24415, "train_loss": 2.783382, "learning_rate": 0.00010385364082277821, "grad_norm": 0.6641293168067932, "tokens_seen": 1228800000, "tokens_per_second": 115629}
|
| 226 |
+
{"step": 10000, "validation_loss": 2.918266, "validation_perplexity": 18.5092}
|
| 227 |
+
Saved checkpoint: runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00010000
|
| 228 |
+
{"step": 10050, "max_steps": 24415, "train_loss": 2.775162, "learning_rate": 0.00010343280764048466, "grad_norm": 0.7611445188522339, "tokens_seen": 1234944000, "tokens_per_second": 111850}
|
| 229 |
+
{"step": 10100, "max_steps": 24415, "train_loss": 2.774081, "learning_rate": 0.00010301107228653688, "grad_norm": 0.6389397382736206, "tokens_seen": 1241088000, "tokens_per_second": 115663}
|
| 230 |
+
{"step": 10150, "max_steps": 24415, "train_loss": 2.850143, "learning_rate": 0.0001025884529370768, "grad_norm": 0.648176908493042, "tokens_seen": 1247232000, "tokens_per_second": 115662}
|
| 231 |
+
{"step": 10200, "max_steps": 24415, "train_loss": 2.785788, "learning_rate": 0.00010216496780634514, "grad_norm": 0.6346315145492554, "tokens_seen": 1253376000, "tokens_per_second": 115662}
|
| 232 |
+
{"step": 10250, "max_steps": 24415, "train_loss": 2.827677, "learning_rate": 0.00010174063514589646, "grad_norm": 0.6351058483123779, "tokens_seen": 1259520000, "tokens_per_second": 115682}
|
| 233 |
+
{"step": 10300, "max_steps": 24415, "train_loss": 2.861314, "learning_rate": 0.00010131547324381256, "grad_norm": 0.7274054884910583, "tokens_seen": 1265664000, "tokens_per_second": 115658}
|
| 234 |
+
{"step": 10350, "max_steps": 24415, "train_loss": 2.80999, "learning_rate": 0.00010088950042391428, "grad_norm": 0.6244896054267883, "tokens_seen": 1271808000, "tokens_per_second": 115603}
|
| 235 |
+
{"step": 10400, "max_steps": 24415, "train_loss": 2.8915, "learning_rate": 0.00010046273504497171, "grad_norm": 0.6021413803100586, "tokens_seen": 1277952000, "tokens_per_second": 115526}
|
| 236 |
+
{"step": 10450, "max_steps": 24415, "train_loss": 2.839523, "learning_rate": 0.00010003519549991305, "grad_norm": 0.6517390608787537, "tokens_seen": 1284096000, "tokens_per_second": 115511}
|
| 237 |
+
{"step": 10500, "max_steps": 24415, "train_loss": 2.868553, "learning_rate": 9.960690021503191e-05, "grad_norm": 0.6538746356964111, "tokens_seen": 1290240000, "tokens_per_second": 115540}
|
| 238 |
+
{"step": 10550, "max_steps": 24415, "train_loss": 2.958018, "learning_rate": 9.917786764919303e-05, "grad_norm": 0.681606113910675, "tokens_seen": 1296384000, "tokens_per_second": 115549}
|
| 239 |
+
{"step": 10600, "max_steps": 24415, "train_loss": 2.696034, "learning_rate": 9.874811629303682e-05, "grad_norm": 0.6564909815788269, "tokens_seen": 1302528000, "tokens_per_second": 115552}
|
| 240 |
+
{"step": 10650, "max_steps": 24415, "train_loss": 2.781479, "learning_rate": 9.831766466818254e-05, "grad_norm": 0.6478615999221802, "tokens_seen": 1308672000, "tokens_per_second": 115511}
|
| 241 |
+
{"step": 10700, "max_steps": 24415, "train_loss": 2.929315, "learning_rate": 9.788653132642987e-05, "grad_norm": 0.6069942116737366, "tokens_seen": 1314816000, "tokens_per_second": 115469}
|
| 242 |
+
{"step": 10750, "max_steps": 24415, "train_loss": 2.940819, "learning_rate": 9.745473484895944e-05, "grad_norm": 0.6308217644691467, "tokens_seen": 1320960000, "tokens_per_second": 115430}
|
| 243 |
+
{"step": 10800, "max_steps": 24415, "train_loss": 2.799978, "learning_rate": 9.702229384553203e-05, "grad_norm": 0.6280612349510193, "tokens_seen": 1327104000, "tokens_per_second": 115436}
|
| 244 |
+
{"step": 10850, "max_steps": 24415, "train_loss": 2.869563, "learning_rate": 9.658922695368647e-05, "grad_norm": 0.6281547546386719, "tokens_seen": 1333248000, "tokens_per_second": 115442}
|
| 245 |
+
{"step": 10900, "max_steps": 24415, "train_loss": 2.887809, "learning_rate": 9.615555283793642e-05, "grad_norm": 0.6927506327629089, "tokens_seen": 1339392000, "tokens_per_second": 115582}
|
| 246 |
+
{"step": 10950, "max_steps": 24415, "train_loss": 2.833543, "learning_rate": 9.572129018896597e-05, "grad_norm": 0.6338732242584229, "tokens_seen": 1345536000, "tokens_per_second": 115663}
|
| 247 |
+
{"step": 11000, "max_steps": 24415, "train_loss": 2.729088, "learning_rate": 9.528645772282403e-05, "grad_norm": 0.6513716578483582, "tokens_seen": 1351680000, "tokens_per_second": 115694}
|
| 248 |
+
{"step": 11000, "validation_loss": 2.911769, "validation_perplexity": 18.3893}
|
| 249 |
+
{"step": 11050, "max_steps": 24415, "train_loss": 2.870943, "learning_rate": 9.485107418011784e-05, "grad_norm": 0.6725945472717285, "tokens_seen": 1357824000, "tokens_per_second": 113478}
|
| 250 |
+
{"step": 11100, "max_steps": 24415, "train_loss": 2.791466, "learning_rate": 9.44151583252051e-05, "grad_norm": 0.6462335586547852, "tokens_seen": 1363968000, "tokens_per_second": 115642}
|
| 251 |
+
{"step": 11150, "max_steps": 24415, "train_loss": 2.850855, "learning_rate": 9.397872894538538e-05, "grad_norm": 0.6676380634307861, "tokens_seen": 1370112000, "tokens_per_second": 115671}
|
| 252 |
+
{"step": 11200, "max_steps": 24415, "train_loss": 2.849724, "learning_rate": 9.354180485009036e-05, "grad_norm": 0.5953037142753601, "tokens_seen": 1376256000, "tokens_per_second": 115721}
|
| 253 |
+
{"step": 11250, "max_steps": 24415, "train_loss": 2.752258, "learning_rate": 9.310440487007324e-05, "grad_norm": 0.6712871193885803, "tokens_seen": 1382400000, "tokens_per_second": 115674}
|
| 254 |
+
{"step": 11300, "max_steps": 24415, "train_loss": 2.850544, "learning_rate": 9.26665478565971e-05, "grad_norm": 0.6593591570854187, "tokens_seen": 1388544000, "tokens_per_second": 115636}
|
| 255 |
+
{"step": 11350, "max_steps": 24415, "train_loss": 2.827023, "learning_rate": 9.222825268062237e-05, "grad_norm": 0.6266449689865112, "tokens_seen": 1394688000, "tokens_per_second": 115630}
|
| 256 |
+
{"step": 11400, "max_steps": 24415, "train_loss": 2.742964, "learning_rate": 9.178953823199372e-05, "grad_norm": 0.6214085817337036, "tokens_seen": 1400832000, "tokens_per_second": 115692}
|
| 257 |
+
{"step": 11450, "max_steps": 24415, "train_loss": 2.870918, "learning_rate": 9.135042341862582e-05, "grad_norm": 0.634956955909729, "tokens_seen": 1406976000, "tokens_per_second": 115698}
|
| 258 |
+
{"step": 11500, "max_steps": 24415, "train_loss": 2.695957, "learning_rate": 9.091092716568831e-05, "grad_norm": 0.6031589508056641, "tokens_seen": 1413120000, "tokens_per_second": 115660}
|
| 259 |
+
{"step": 11550, "max_steps": 24415, "train_loss": 2.973763, "learning_rate": 9.04710684147905e-05, "grad_norm": 0.6234080791473389, "tokens_seen": 1419264000, "tokens_per_second": 115694}
|
| 260 |
+
{"step": 11600, "max_steps": 24415, "train_loss": 2.770549, "learning_rate": 9.003086612316455e-05, "grad_norm": 0.6769762635231018, "tokens_seen": 1425408000, "tokens_per_second": 115705}
|
| 261 |
+
{"step": 11650, "max_steps": 24415, "train_loss": 2.84782, "learning_rate": 8.959033926284892e-05, "grad_norm": 0.7006694078445435, "tokens_seen": 1431552000, "tokens_per_second": 115663}
|
| 262 |
+
{"step": 11700, "max_steps": 24415, "train_loss": 2.885127, "learning_rate": 8.914950681987033e-05, "grad_norm": 0.6738485097885132, "tokens_seen": 1437696000, "tokens_per_second": 115643}
|
| 263 |
+
{"step": 11750, "max_steps": 24415, "train_loss": 2.778524, "learning_rate": 8.870838779342573e-05, "grad_norm": 0.6949870586395264, "tokens_seen": 1443840000, "tokens_per_second": 115675}
|
| 264 |
+
{"step": 11800, "max_steps": 24415, "train_loss": 2.804691, "learning_rate": 8.82670011950633e-05, "grad_norm": 0.6358240842819214, "tokens_seen": 1449984000, "tokens_per_second": 115694}
|
| 265 |
+
{"step": 11850, "max_steps": 24415, "train_loss": 2.793259, "learning_rate": 8.78253660478632e-05, "grad_norm": 0.6506847143173218, "tokens_seen": 1456128000, "tokens_per_second": 115639}
|
| 266 |
+
{"step": 11900, "max_steps": 24415, "train_loss": 2.802632, "learning_rate": 8.738350138561769e-05, "grad_norm": 0.653716504573822, "tokens_seen": 1462272000, "tokens_per_second": 115615}
|
| 267 |
+
{"step": 11950, "max_steps": 24415, "train_loss": 2.823897, "learning_rate": 8.69414262520107e-05, "grad_norm": 0.6153150200843811, "tokens_seen": 1468416000, "tokens_per_second": 115660}
|
| 268 |
+
{"step": 12000, "max_steps": 24415, "train_loss": 2.723172, "learning_rate": 8.649915969979722e-05, "grad_norm": 0.6362366080284119, "tokens_seen": 1474560000, "tokens_per_second": 115652}
|
| 269 |
+
{"step": 12000, "validation_loss": 2.905931, "validation_perplexity": 18.2823}
|
| 270 |
+
{"step": 12050, "max_steps": 24415, "train_loss": 2.788503, "learning_rate": 8.605672078998203e-05, "grad_norm": 0.6385388970375061, "tokens_seen": 1480704000, "tokens_per_second": 113464}
|
| 271 |
+
{"step": 12100, "max_steps": 24415, "train_loss": 2.863936, "learning_rate": 8.561412859099828e-05, "grad_norm": 0.6147498488426208, "tokens_seen": 1486848000, "tokens_per_second": 115615}
|
| 272 |
+
{"step": 12150, "max_steps": 24415, "train_loss": 2.9177, "learning_rate": 8.517140217788565e-05, "grad_norm": 0.6115130186080933, "tokens_seen": 1492992000, "tokens_per_second": 115654}
|
| 273 |
+
{"step": 12200, "max_steps": 24415, "train_loss": 2.721272, "learning_rate": 8.472856063146826e-05, "grad_norm": 0.6808973550796509, "tokens_seen": 1499136000, "tokens_per_second": 115618}
|
| 274 |
+
{"step": 12250, "max_steps": 24415, "train_loss": 2.686518, "learning_rate": 8.428562303753227e-05, "grad_norm": 0.7563024759292603, "tokens_seen": 1505280000, "tokens_per_second": 115582}
|
| 275 |
+
{"step": 12300, "max_steps": 24415, "train_loss": 2.819199, "learning_rate": 8.384260848600334e-05, "grad_norm": 0.6369250416755676, "tokens_seen": 1511424000, "tokens_per_second": 115612}
|
| 276 |
+
{"step": 12350, "max_steps": 24415, "train_loss": 2.606734, "learning_rate": 8.339953607012387e-05, "grad_norm": 0.6867128610610962, "tokens_seen": 1517568000, "tokens_per_second": 115660}
|
| 277 |
+
{"step": 12400, "max_steps": 24415, "train_loss": 2.724569, "learning_rate": 8.295642488563015e-05, "grad_norm": 0.6900272965431213, "tokens_seen": 1523712000, "tokens_per_second": 115654}
|
| 278 |
+
{"step": 12450, "max_steps": 24415, "train_loss": 2.916764, "learning_rate": 8.251329402992934e-05, "grad_norm": 0.6223289966583252, "tokens_seen": 1529856000, "tokens_per_second": 115611}
|
| 279 |
+
{"step": 12500, "max_steps": 24415, "train_loss": 2.714432, "learning_rate": 8.20701626012764e-05, "grad_norm": 0.6510653495788574, "tokens_seen": 1536000000, "tokens_per_second": 115652}
|
| 280 |
+
{"step": 12550, "max_steps": 24415, "train_loss": 2.846118, "learning_rate": 8.162704969795096e-05, "grad_norm": 0.6865654587745667, "tokens_seen": 1542144000, "tokens_per_second": 115665}
|
| 281 |
+
{"step": 12600, "max_steps": 24415, "train_loss": 2.918616, "learning_rate": 8.118397441743427e-05, "grad_norm": 0.661815881729126, "tokens_seen": 1548288000, "tokens_per_second": 115631}
|
| 282 |
+
{"step": 12650, "max_steps": 24415, "train_loss": 2.86565, "learning_rate": 8.074095585558602e-05, "grad_norm": 0.6402216553688049, "tokens_seen": 1554432000, "tokens_per_second": 115621}
|
| 283 |
+
{"step": 12700, "max_steps": 24415, "train_loss": 2.907798, "learning_rate": 8.029801310582152e-05, "grad_norm": 0.6475313901901245, "tokens_seen": 1560576000, "tokens_per_second": 115649}
|
| 284 |
+
{"step": 12750, "max_steps": 24415, "train_loss": 2.718009, "learning_rate": 7.98551652582886e-05, "grad_norm": 0.6396787762641907, "tokens_seen": 1566720000, "tokens_per_second": 115666}
|
| 285 |
+
{"step": 12800, "max_steps": 24415, "train_loss": 2.888366, "learning_rate": 7.941243139904506e-05, "grad_norm": 0.8234379291534424, "tokens_seen": 1572864000, "tokens_per_second": 115629}
|
| 286 |
+
{"step": 12850, "max_steps": 24415, "train_loss": 2.671792, "learning_rate": 7.896983060923585e-05, "grad_norm": 0.6498484015464783, "tokens_seen": 1579008000, "tokens_per_second": 115603}
|
| 287 |
+
{"step": 12900, "max_steps": 24415, "train_loss": 2.773788, "learning_rate": 7.852738196427095e-05, "grad_norm": 0.6323695778846741, "tokens_seen": 1585152000, "tokens_per_second": 115614}
|
| 288 |
+
{"step": 12950, "max_steps": 24415, "train_loss": 2.8168, "learning_rate": 7.8085104533003e-05, "grad_norm": 0.6435253024101257, "tokens_seen": 1591296000, "tokens_per_second": 115633}
|
| 289 |
+
{"step": 13000, "max_steps": 24415, "train_loss": 2.724554, "learning_rate": 7.764301737690578e-05, "grad_norm": 0.5999656915664673, "tokens_seen": 1597440000, "tokens_per_second": 115589}
|
| 290 |
+
{"step": 13000, "validation_loss": 2.901231, "validation_perplexity": 18.1965}
|
| 291 |
+
{"step": 13050, "max_steps": 24415, "train_loss": 2.780255, "learning_rate": 7.720113954925233e-05, "grad_norm": 0.6173715591430664, "tokens_seen": 1603584000, "tokens_per_second": 113429}
|
| 292 |
+
{"step": 13100, "max_steps": 24415, "train_loss": 2.756385, "learning_rate": 7.675949009429407e-05, "grad_norm": 0.6280876398086548, "tokens_seen": 1609728000, "tokens_per_second": 115638}
|
| 293 |
+
{"step": 13150, "max_steps": 24415, "train_loss": 2.79896, "learning_rate": 7.63180880464398e-05, "grad_norm": 0.7241061329841614, "tokens_seen": 1615872000, "tokens_per_second": 115638}
|
| 294 |
+
{"step": 13200, "max_steps": 24415, "train_loss": 2.841177, "learning_rate": 7.587695242943563e-05, "grad_norm": 0.6539769768714905, "tokens_seen": 1622016000, "tokens_per_second": 115619}
|
| 295 |
+
{"step": 13250, "max_steps": 24415, "train_loss": 2.778755, "learning_rate": 7.543610225554469e-05, "grad_norm": 0.6095080971717834, "tokens_seen": 1628160000, "tokens_per_second": 115628}
|
| 296 |
+
{"step": 13300, "max_steps": 24415, "train_loss": 2.882869, "learning_rate": 7.499555652472813e-05, "grad_norm": 0.6407991647720337, "tokens_seen": 1634304000, "tokens_per_second": 115666}
|
| 297 |
+
{"step": 13350, "max_steps": 24415, "train_loss": 2.947798, "learning_rate": 7.455533422382599e-05, "grad_norm": 0.6137878894805908, "tokens_seen": 1640448000, "tokens_per_second": 115646}
|
| 298 |
+
{"step": 13400, "max_steps": 24415, "train_loss": 2.744945, "learning_rate": 7.411545432573904e-05, "grad_norm": 0.76528000831604, "tokens_seen": 1646592000, "tokens_per_second": 115618}
|
| 299 |
+
{"step": 13450, "max_steps": 24415, "train_loss": 2.75651, "learning_rate": 7.367593578861094e-05, "grad_norm": 0.7015501856803894, "tokens_seen": 1652736000, "tokens_per_second": 115635}
|
| 300 |
+
{"step": 13500, "max_steps": 24415, "train_loss": 2.878917, "learning_rate": 7.323679755501133e-05, "grad_norm": 0.5917170643806458, "tokens_seen": 1658880000, "tokens_per_second": 115676}
|
| 301 |
+
{"step": 13550, "max_steps": 24415, "train_loss": 2.757949, "learning_rate": 7.279805855111927e-05, "grad_norm": 0.6195739507675171, "tokens_seen": 1665024000, "tokens_per_second": 115639}
|
| 302 |
+
{"step": 13600, "max_steps": 24415, "train_loss": 2.804666, "learning_rate": 7.235973768590771e-05, "grad_norm": 0.7056005001068115, "tokens_seen": 1671168000, "tokens_per_second": 115632}
|
| 303 |
+
{"step": 13650, "max_steps": 24415, "train_loss": 2.70874, "learning_rate": 7.192185385032843e-05, "grad_norm": 0.9107558727264404, "tokens_seen": 1677312000, "tokens_per_second": 115641}
|
| 304 |
+
{"step": 13700, "max_steps": 24415, "train_loss": 2.779933, "learning_rate": 7.148442591649797e-05, "grad_norm": 0.6652510166168213, "tokens_seen": 1683456000, "tokens_per_second": 115682}
|
| 305 |
+
{"step": 13750, "max_steps": 24415, "train_loss": 2.603291, "learning_rate": 7.104747273688412e-05, "grad_norm": 0.6849772334098816, "tokens_seen": 1689600000, "tokens_per_second": 115664}
|
| 306 |
+
{"step": 13800, "max_steps": 24415, "train_loss": 2.811565, "learning_rate": 7.061101314349354e-05, "grad_norm": 0.6300348043441772, "tokens_seen": 1695744000, "tokens_per_second": 115635}
|
| 307 |
+
{"step": 13850, "max_steps": 24415, "train_loss": 2.741807, "learning_rate": 7.01750659470601e-05, "grad_norm": 0.645389974117279, "tokens_seen": 1701888000, "tokens_per_second": 115581}
|
| 308 |
+
{"step": 13900, "max_steps": 24415, "train_loss": 2.81428, "learning_rate": 6.97396499362341e-05, "grad_norm": 0.6963964104652405, "tokens_seen": 1708032000, "tokens_per_second": 115551}
|
| 309 |
+
{"step": 13950, "max_steps": 24415, "train_loss": 2.861082, "learning_rate": 6.930478387677264e-05, "grad_norm": 0.6477249264717102, "tokens_seen": 1714176000, "tokens_per_second": 115599}
|
| 310 |
+
{"step": 14000, "max_steps": 24415, "train_loss": 2.859481, "learning_rate": 6.887048651073072e-05, "grad_norm": 0.6993773579597473, "tokens_seen": 1720320000, "tokens_per_second": 115607}
|
| 311 |
+
{"step": 14000, "validation_loss": 2.895466, "validation_perplexity": 18.0919}
|
| 312 |
+
{"step": 14050, "max_steps": 24415, "train_loss": 2.814404, "learning_rate": 6.843677655565353e-05, "grad_norm": 0.6627041101455688, "tokens_seen": 1726464000, "tokens_per_second": 113444}
|
| 313 |
+
{"step": 14100, "max_steps": 24415, "train_loss": 2.721513, "learning_rate": 6.800367270376977e-05, "grad_norm": 0.6665034294128418, "tokens_seen": 1732608000, "tokens_per_second": 115615}
|
| 314 |
+
{"step": 14150, "max_steps": 24415, "train_loss": 2.683664, "learning_rate": 6.7571193621186e-05, "grad_norm": 0.6050234436988831, "tokens_seen": 1738752000, "tokens_per_second": 115613}
|
| 315 |
+
{"step": 14200, "max_steps": 24415, "train_loss": 2.886776, "learning_rate": 6.713935794708225e-05, "grad_norm": 0.6370565891265869, "tokens_seen": 1744896000, "tokens_per_second": 115620}
|
| 316 |
+
{"step": 14250, "max_steps": 24415, "train_loss": 2.712796, "learning_rate": 6.670818429290858e-05, "grad_norm": 0.6131051182746887, "tokens_seen": 1751040000, "tokens_per_second": 115631}
|
| 317 |
+
{"step": 14300, "max_steps": 24415, "train_loss": 2.78503, "learning_rate": 6.6277691241583e-05, "grad_norm": 0.612759530544281, "tokens_seen": 1757184000, "tokens_per_second": 115644}
|
| 318 |
+
{"step": 14350, "max_steps": 24415, "train_loss": 2.740587, "learning_rate": 6.584789734669072e-05, "grad_norm": 0.6165915131568909, "tokens_seen": 1763328000, "tokens_per_second": 115642}
|
| 319 |
+
{"step": 14400, "max_steps": 24415, "train_loss": 2.870152, "learning_rate": 6.541882113168419e-05, "grad_norm": 0.6879968643188477, "tokens_seen": 1769472000, "tokens_per_second": 115640}
|
| 320 |
+
{"step": 14450, "max_steps": 24415, "train_loss": 2.838705, "learning_rate": 6.499048108908511e-05, "grad_norm": 0.6072972416877747, "tokens_seen": 1775616000, "tokens_per_second": 115633}
|
| 321 |
+
{"step": 14500, "max_steps": 24415, "train_loss": 2.752175, "learning_rate": 6.456289567968726e-05, "grad_norm": 0.6414802670478821, "tokens_seen": 1781760000, "tokens_per_second": 115631}
|
| 322 |
+
{"step": 14550, "max_steps": 24415, "train_loss": 2.849361, "learning_rate": 6.41360833317609e-05, "grad_norm": 0.6170805096626282, "tokens_seen": 1787904000, "tokens_per_second": 115623}
|
| 323 |
+
{"step": 14600, "max_steps": 24415, "train_loss": 2.782098, "learning_rate": 6.371006244025843e-05, "grad_norm": 0.6256428360939026, "tokens_seen": 1794048000, "tokens_per_second": 115589}
|
| 324 |
+
{"step": 14650, "max_steps": 24415, "train_loss": 2.936581, "learning_rate": 6.32848513660219e-05, "grad_norm": 0.6580077409744263, "tokens_seen": 1800192000, "tokens_per_second": 115594}
|
| 325 |
+
{"step": 14700, "max_steps": 24415, "train_loss": 2.845289, "learning_rate": 6.286046843499127e-05, "grad_norm": 0.6480897665023804, "tokens_seen": 1806336000, "tokens_per_second": 115546}
|
| 326 |
+
{"step": 14750, "max_steps": 24415, "train_loss": 2.850777, "learning_rate": 6.243693193741497e-05, "grad_norm": 0.6406828761100769, "tokens_seen": 1812480000, "tokens_per_second": 115569}
|
| 327 |
+
{"step": 14800, "max_steps": 24415, "train_loss": 2.725981, "learning_rate": 6.20142601270614e-05, "grad_norm": 0.7072864770889282, "tokens_seen": 1818624000, "tokens_per_second": 115645}
|
| 328 |
+
{"step": 14850, "max_steps": 24415, "train_loss": 2.690922, "learning_rate": 6.15924712204323e-05, "grad_norm": 0.7200900912284851, "tokens_seen": 1824768000, "tokens_per_second": 115676}
|
| 329 |
+
{"step": 14900, "max_steps": 24415, "train_loss": 2.866086, "learning_rate": 6.117158339597754e-05, "grad_norm": 0.6389758586883545, "tokens_seen": 1830912000, "tokens_per_second": 115629}
|
| 330 |
+
{"step": 14950, "max_steps": 24415, "train_loss": 2.697815, "learning_rate": 6.075161479331188e-05, "grad_norm": 0.6718475818634033, "tokens_seen": 1837056000, "tokens_per_second": 115626}
|
| 331 |
+
{"step": 15000, "max_steps": 24415, "train_loss": 2.823217, "learning_rate": 6.033258351243286e-05, "grad_norm": 0.6866132616996765, "tokens_seen": 1843200000, "tokens_per_second": 115586}
|
| 332 |
+
{"step": 15000, "validation_loss": 2.890042, "validation_perplexity": 17.9941}
|
| 333 |
+
Saved checkpoint: runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00015000
|
| 334 |
+
{"step": 15050, "max_steps": 24415, "train_loss": 2.902357, "learning_rate": 5.991450761294112e-05, "grad_norm": 0.6383296251296997, "tokens_seen": 1849344000, "tokens_per_second": 111864}
|
| 335 |
+
{"step": 15100, "max_steps": 24415, "train_loss": 2.827427, "learning_rate": 5.9497405113261604e-05, "grad_norm": 0.6769825220108032, "tokens_seen": 1855488000, "tokens_per_second": 115560}
|
| 336 |
+
{"step": 15150, "max_steps": 24415, "train_loss": 2.633896, "learning_rate": 5.908129398986747e-05, "grad_norm": 0.6125101447105408, "tokens_seen": 1861632000, "tokens_per_second": 115565}
|
| 337 |
+
{"step": 15200, "max_steps": 24415, "train_loss": 2.883741, "learning_rate": 5.866619217650488e-05, "grad_norm": 0.6056382060050964, "tokens_seen": 1867776000, "tokens_per_second": 115555}
|
| 338 |
+
{"step": 15250, "max_steps": 24415, "train_loss": 2.786928, "learning_rate": 5.8252117563420435e-05, "grad_norm": 0.6179369688034058, "tokens_seen": 1873920000, "tokens_per_second": 115554}
|
| 339 |
+
{"step": 15300, "max_steps": 24415, "train_loss": 2.869814, "learning_rate": 5.783908799658988e-05, "grad_norm": 0.589368462562561, "tokens_seen": 1880064000, "tokens_per_second": 115511}
|
| 340 |
+
{"step": 15350, "max_steps": 24415, "train_loss": 2.79135, "learning_rate": 5.7427121276949194e-05, "grad_norm": 0.5995170474052429, "tokens_seen": 1886208000, "tokens_per_second": 115546}
|
| 341 |
+
{"step": 15400, "max_steps": 24415, "train_loss": 2.735143, "learning_rate": 5.701623515962719e-05, "grad_norm": 0.6787047982215881, "tokens_seen": 1892352000, "tokens_per_second": 115503}
|
| 342 |
+
{"step": 15450, "max_steps": 24415, "train_loss": 2.699596, "learning_rate": 5.660644735318049e-05, "grad_norm": 0.6562452912330627, "tokens_seen": 1898496000, "tokens_per_second": 115512}
|
| 343 |
+
{"step": 15500, "max_steps": 24415, "train_loss": 2.916876, "learning_rate": 5.61977755188301e-05, "grad_norm": 0.6687349081039429, "tokens_seen": 1904640000, "tokens_per_second": 115433}
|
| 344 |
+
{"step": 15550, "max_steps": 24415, "train_loss": 2.802557, "learning_rate": 5.579023726970045e-05, "grad_norm": 0.6334400177001953, "tokens_seen": 1910784000, "tokens_per_second": 115546}
|
| 345 |
+
{"step": 15600, "max_steps": 24415, "train_loss": 2.887684, "learning_rate": 5.538385017006011e-05, "grad_norm": 0.6181633472442627, "tokens_seen": 1916928000, "tokens_per_second": 115636}
|
| 346 |
+
{"step": 15650, "max_steps": 24415, "train_loss": 2.784732, "learning_rate": 5.4978631734564984e-05, "grad_norm": 0.6535164713859558, "tokens_seen": 1923072000, "tokens_per_second": 115668}
|
| 347 |
+
{"step": 15700, "max_steps": 24415, "train_loss": 2.804852, "learning_rate": 5.457459942750329e-05, "grad_norm": 0.802883505821228, "tokens_seen": 1929216000, "tokens_per_second": 115642}
|
| 348 |
+
{"step": 15750, "max_steps": 24415, "train_loss": 2.736158, "learning_rate": 5.4171770662043e-05, "grad_norm": 0.664082944393158, "tokens_seen": 1935360000, "tokens_per_second": 115632}
|
| 349 |
+
{"step": 15800, "max_steps": 24415, "train_loss": 2.798161, "learning_rate": 5.37701627994812e-05, "grad_norm": 0.6006529927253723, "tokens_seen": 1941504000, "tokens_per_second": 115551}
|
| 350 |
+
{"step": 15850, "max_steps": 24415, "train_loss": 2.810937, "learning_rate": 5.336979314849607e-05, "grad_norm": 0.6848961710929871, "tokens_seen": 1947648000, "tokens_per_second": 115517}
|
| 351 |
+
{"step": 15900, "max_steps": 24415, "train_loss": 2.86459, "learning_rate": 5.297067896440069e-05, "grad_norm": 0.5799385905265808, "tokens_seen": 1953792000, "tokens_per_second": 115555}
|
| 352 |
+
{"step": 15950, "max_steps": 24415, "train_loss": 2.862726, "learning_rate": 5.257283744839956e-05, "grad_norm": 0.6372018456459045, "tokens_seen": 1959936000, "tokens_per_second": 115553}
|
| 353 |
+
{"step": 16000, "max_steps": 24415, "train_loss": 2.638452, "learning_rate": 5.2176285746847004e-05, "grad_norm": 0.622450590133667, "tokens_seen": 1966080000, "tokens_per_second": 115521}
|
| 354 |
+
{"step": 16000, "validation_loss": 2.883631, "validation_perplexity": 17.8791}
|
| 355 |
+
{"step": 16050, "max_steps": 24415, "train_loss": 2.759337, "learning_rate": 5.178104095050848e-05, "grad_norm": 0.6415831446647644, "tokens_seen": 1972224000, "tokens_per_second": 113344}
|
| 356 |
+
{"step": 16100, "max_steps": 24415, "train_loss": 2.793138, "learning_rate": 5.13871200938238e-05, "grad_norm": 0.5900813341140747, "tokens_seen": 1978368000, "tokens_per_second": 115459}
|
| 357 |
+
{"step": 16150, "max_steps": 24415, "train_loss": 2.735165, "learning_rate": 5.0994540154172984e-05, "grad_norm": 0.6464166641235352, "tokens_seen": 1984512000, "tokens_per_second": 115447}
|
| 358 |
+
{"step": 16200, "max_steps": 24415, "train_loss": 2.77423, "learning_rate": 5.060331805114466e-05, "grad_norm": 0.6363893151283264, "tokens_seen": 1990656000, "tokens_per_second": 115414}
|
| 359 |
+
{"step": 16250, "max_steps": 24415, "train_loss": 2.813646, "learning_rate": 5.0213470645806766e-05, "grad_norm": 0.6006516218185425, "tokens_seen": 1996800000, "tokens_per_second": 115415}
|
| 360 |
+
{"step": 16300, "max_steps": 24415, "train_loss": 2.751798, "learning_rate": 4.982501473997987e-05, "grad_norm": 0.6130074858665466, "tokens_seen": 2002944000, "tokens_per_second": 115556}
|
| 361 |
+
{"step": 16350, "max_steps": 24415, "train_loss": 2.581523, "learning_rate": 4.9437967075513156e-05, "grad_norm": 0.6909277439117432, "tokens_seen": 2009088000, "tokens_per_second": 115608}
|
| 362 |
+
{"step": 16400, "max_steps": 24415, "train_loss": 2.735869, "learning_rate": 4.905234433356263e-05, "grad_norm": 0.6061579585075378, "tokens_seen": 2015232000, "tokens_per_second": 115618}
|
| 363 |
+
{"step": 16450, "max_steps": 24415, "train_loss": 2.807998, "learning_rate": 4.8668163133872515e-05, "grad_norm": 0.6679533123970032, "tokens_seen": 2021376000, "tokens_per_second": 115633}
|
| 364 |
+
{"step": 16500, "max_steps": 24415, "train_loss": 2.751439, "learning_rate": 4.8285440034058625e-05, "grad_norm": 0.6001737117767334, "tokens_seen": 2027520000, "tokens_per_second": 115605}
|
| 365 |
+
{"step": 16550, "max_steps": 24415, "train_loss": 2.812106, "learning_rate": 4.790419152889506e-05, "grad_norm": 0.6463189721107483, "tokens_seen": 2033664000, "tokens_per_second": 115646}
|
| 366 |
+
{"step": 16600, "max_steps": 24415, "train_loss": 2.806943, "learning_rate": 4.752443404960306e-05, "grad_norm": 0.589819073677063, "tokens_seen": 2039808000, "tokens_per_second": 115664}
|
| 367 |
+
{"step": 16650, "max_steps": 24415, "train_loss": 2.696105, "learning_rate": 4.7146183963143024e-05, "grad_norm": 0.6867057085037231, "tokens_seen": 2045952000, "tokens_per_second": 115627}
|
| 368 |
+
{"step": 16700, "max_steps": 24415, "train_loss": 2.667582, "learning_rate": 4.676945757150907e-05, "grad_norm": 0.6482738852500916, "tokens_seen": 2052096000, "tokens_per_second": 115647}
|
| 369 |
+
{"step": 16750, "max_steps": 24415, "train_loss": 2.577223, "learning_rate": 4.639427111102632e-05, "grad_norm": 0.6611011028289795, "tokens_seen": 2058240000, "tokens_per_second": 115680}
|
| 370 |
+
{"step": 16800, "max_steps": 24415, "train_loss": 2.715673, "learning_rate": 4.6020640751651346e-05, "grad_norm": 0.7082952260971069, "tokens_seen": 2064384000, "tokens_per_second": 115686}
|
| 371 |
+
{"step": 16850, "max_steps": 24415, "train_loss": 2.767459, "learning_rate": 4.564858259627514e-05, "grad_norm": 0.612030565738678, "tokens_seen": 2070528000, "tokens_per_second": 115666}
|
| 372 |
+
{"step": 16900, "max_steps": 24415, "train_loss": 2.778458, "learning_rate": 4.527811268002908e-05, "grad_norm": 0.6326088905334473, "tokens_seen": 2076672000, "tokens_per_second": 115642}
|
| 373 |
+
{"step": 16950, "max_steps": 24415, "train_loss": 2.707249, "learning_rate": 4.4909246969594e-05, "grad_norm": 0.6098898649215698, "tokens_seen": 2082816000, "tokens_per_second": 115640}
|
| 374 |
+
{"step": 17000, "max_steps": 24415, "train_loss": 2.895086, "learning_rate": 4.454200136251182e-05, "grad_norm": 0.5895346403121948, "tokens_seen": 2088960000, "tokens_per_second": 115704}
|
| 375 |
+
{"step": 17000, "validation_loss": 2.878743, "validation_perplexity": 17.7919}
|
| 376 |
+
{"step": 17050, "max_steps": 24415, "train_loss": 2.733071, "learning_rate": 4.417639168650056e-05, "grad_norm": 0.7047613859176636, "tokens_seen": 2095104000, "tokens_per_second": 113501}
|
| 377 |
+
{"step": 17100, "max_steps": 24415, "train_loss": 2.790722, "learning_rate": 4.3812433698772194e-05, "grad_norm": 0.6512798070907593, "tokens_seen": 2101248000, "tokens_per_second": 115668}
|
| 378 |
+
{"step": 17150, "max_steps": 24415, "train_loss": 2.787556, "learning_rate": 4.345014308535338e-05, "grad_norm": 0.6246313452720642, "tokens_seen": 2107392000, "tokens_per_second": 115608}
|
| 379 |
+
{"step": 17200, "max_steps": 24415, "train_loss": 2.820263, "learning_rate": 4.308953546040962e-05, "grad_norm": 0.6640097498893738, "tokens_seen": 2113536000, "tokens_per_second": 115484}
|
| 380 |
+
{"step": 17250, "max_steps": 24415, "train_loss": 2.645744, "learning_rate": 4.273062636557215e-05, "grad_norm": 0.6085199117660522, "tokens_seen": 2119680000, "tokens_per_second": 115510}
|
| 381 |
+
{"step": 17300, "max_steps": 24415, "train_loss": 2.775353, "learning_rate": 4.2373431269268245e-05, "grad_norm": 0.7274950742721558, "tokens_seen": 2125824000, "tokens_per_second": 115548}
|
| 382 |
+
{"step": 17350, "max_steps": 24415, "train_loss": 2.774427, "learning_rate": 4.201796556605439e-05, "grad_norm": 0.6599135994911194, "tokens_seen": 2131968000, "tokens_per_second": 115467}
|
| 383 |
+
{"step": 17400, "max_steps": 24415, "train_loss": 2.784656, "learning_rate": 4.166424457595302e-05, "grad_norm": 0.6395982503890991, "tokens_seen": 2138112000, "tokens_per_second": 115483}
|
| 384 |
+
{"step": 17450, "max_steps": 24415, "train_loss": 2.846176, "learning_rate": 4.131228354379201e-05, "grad_norm": 0.6397977471351624, "tokens_seen": 2144256000, "tokens_per_second": 115471}
|
| 385 |
+
{"step": 17500, "max_steps": 24415, "train_loss": 2.925753, "learning_rate": 4.096209763854791e-05, "grad_norm": 0.6196093559265137, "tokens_seen": 2150400000, "tokens_per_second": 115394}
|
| 386 |
+
{"step": 17550, "max_steps": 24415, "train_loss": 2.74457, "learning_rate": 4.061370195269189e-05, "grad_norm": 0.6290102601051331, "tokens_seen": 2156544000, "tokens_per_second": 115373}
|
| 387 |
+
{"step": 17600, "max_steps": 24415, "train_loss": 2.70976, "learning_rate": 4.0267111501539575e-05, "grad_norm": 0.6314865350723267, "tokens_seen": 2162688000, "tokens_per_second": 115455}
|
| 388 |
+
{"step": 17650, "max_steps": 24415, "train_loss": 2.808493, "learning_rate": 3.992234122260362e-05, "grad_norm": 0.6693544983863831, "tokens_seen": 2168832000, "tokens_per_second": 115637}
|
| 389 |
+
{"step": 17700, "max_steps": 24415, "train_loss": 2.664855, "learning_rate": 3.957940597495023e-05, "grad_norm": 0.6201000809669495, "tokens_seen": 2174976000, "tokens_per_second": 115672}
|
| 390 |
+
{"step": 17750, "max_steps": 24415, "train_loss": 2.847698, "learning_rate": 3.923832053855847e-05, "grad_norm": 0.6048368811607361, "tokens_seen": 2181120000, "tokens_per_second": 115660}
|
| 391 |
+
{"step": 17800, "max_steps": 24415, "train_loss": 2.631377, "learning_rate": 3.889909961368344e-05, "grad_norm": 0.5928443670272827, "tokens_seen": 2187264000, "tokens_per_second": 115661}
|
| 392 |
+
{"step": 17850, "max_steps": 24415, "train_loss": 2.685062, "learning_rate": 3.85617578202227e-05, "grad_norm": 0.6104403734207153, "tokens_seen": 2193408000, "tokens_per_second": 115674}
|
| 393 |
+
{"step": 17900, "max_steps": 24415, "train_loss": 2.718334, "learning_rate": 3.8226309697086135e-05, "grad_norm": 0.6966298818588257, "tokens_seen": 2199552000, "tokens_per_second": 115707}
|
| 394 |
+
{"step": 17950, "max_steps": 24415, "train_loss": 2.695795, "learning_rate": 3.789276970156935e-05, "grad_norm": 0.6385568380355835, "tokens_seen": 2205696000, "tokens_per_second": 115677}
|
| 395 |
+
{"step": 18000, "max_steps": 24415, "train_loss": 2.832335, "learning_rate": 3.756115220873061e-05, "grad_norm": 0.6135808825492859, "tokens_seen": 2211840000, "tokens_per_second": 115656}
|
| 396 |
+
{"step": 18000, "validation_loss": 2.874696, "validation_perplexity": 17.72}
|
| 397 |
+
{"step": 18050, "max_steps": 24415, "train_loss": 2.773723, "learning_rate": 3.723147151077128e-05, "grad_norm": 0.6073247194290161, "tokens_seen": 2217984000, "tokens_per_second": 113427}
|
| 398 |
+
{"step": 18100, "max_steps": 24415, "train_loss": 2.742398, "learning_rate": 3.69037418164199e-05, "grad_norm": 0.6289048790931702, "tokens_seen": 2224128000, "tokens_per_second": 115487}
|
| 399 |
+
{"step": 18150, "max_steps": 24415, "train_loss": 2.613473, "learning_rate": 3.657797725031969e-05, "grad_norm": 0.6467992663383484, "tokens_seen": 2230272000, "tokens_per_second": 115556}
|
| 400 |
+
{"step": 18200, "max_steps": 24415, "train_loss": 2.748806, "learning_rate": 3.6254191852419966e-05, "grad_norm": 0.6730474829673767, "tokens_seen": 2236416000, "tokens_per_second": 115559}
|
| 401 |
+
{"step": 18250, "max_steps": 24415, "train_loss": 2.794285, "learning_rate": 3.593239957737087e-05, "grad_norm": 0.6612525582313538, "tokens_seen": 2242560000, "tokens_per_second": 115560}
|
| 402 |
+
{"step": 18300, "max_steps": 24415, "train_loss": 2.714843, "learning_rate": 3.561261429392213e-05, "grad_norm": 0.6353873610496521, "tokens_seen": 2248704000, "tokens_per_second": 115549}
|
| 403 |
+
{"step": 18350, "max_steps": 24415, "train_loss": 2.779381, "learning_rate": 3.5294849784325127e-05, "grad_norm": 0.5724009871482849, "tokens_seen": 2254848000, "tokens_per_second": 115511}
|
| 404 |
+
{"step": 18400, "max_steps": 24415, "train_loss": 2.766123, "learning_rate": 3.49791197437391e-05, "grad_norm": 0.6544906497001648, "tokens_seen": 2260992000, "tokens_per_second": 115551}
|
| 405 |
+
{"step": 18450, "max_steps": 24415, "train_loss": 2.741471, "learning_rate": 3.4665437779640766e-05, "grad_norm": 0.6024874448776245, "tokens_seen": 2267136000, "tokens_per_second": 115478}
|
| 406 |
+
{"step": 18500, "max_steps": 24415, "train_loss": 2.701006, "learning_rate": 3.435381741123796e-05, "grad_norm": 0.6421911716461182, "tokens_seen": 2273280000, "tokens_per_second": 115466}
|
| 407 |
+
{"step": 18550, "max_steps": 24415, "train_loss": 2.679634, "learning_rate": 3.404427206888688e-05, "grad_norm": 0.7828880548477173, "tokens_seen": 2279424000, "tokens_per_second": 115518}
|
| 408 |
+
{"step": 18600, "max_steps": 24415, "train_loss": 2.690722, "learning_rate": 3.373681509351333e-05, "grad_norm": 0.7262030243873596, "tokens_seen": 2285568000, "tokens_per_second": 115651}
|
| 409 |
+
{"step": 18650, "max_steps": 24415, "train_loss": 2.683981, "learning_rate": 3.343145973603772e-05, "grad_norm": 0.6637866497039795, "tokens_seen": 2291712000, "tokens_per_second": 115684}
|
| 410 |
+
{"step": 18700, "max_steps": 24415, "train_loss": 2.810566, "learning_rate": 3.3128219156804e-05, "grad_norm": 0.6260550618171692, "tokens_seen": 2297856000, "tokens_per_second": 115682}
|
| 411 |
+
{"step": 18750, "max_steps": 24415, "train_loss": 2.779154, "learning_rate": 3.282710642501238e-05, "grad_norm": 0.6040164828300476, "tokens_seen": 2304000000, "tokens_per_second": 115669}
|
| 412 |
+
{"step": 18800, "max_steps": 24415, "train_loss": 2.774612, "learning_rate": 3.2528134518156235e-05, "grad_norm": 0.591951847076416, "tokens_seen": 2310144000, "tokens_per_second": 115609}
|
| 413 |
+
{"step": 18850, "max_steps": 24415, "train_loss": 2.887604, "learning_rate": 3.223131632146259e-05, "grad_norm": 0.7033926844596863, "tokens_seen": 2316288000, "tokens_per_second": 115536}
|
| 414 |
+
{"step": 18900, "max_steps": 24415, "train_loss": 2.66931, "learning_rate": 3.1936664627337e-05, "grad_norm": 0.6099238395690918, "tokens_seen": 2322432000, "tokens_per_second": 115559}
|
| 415 |
+
{"step": 18950, "max_steps": 24415, "train_loss": 2.735053, "learning_rate": 3.164419213481199e-05, "grad_norm": 0.625353217124939, "tokens_seen": 2328576000, "tokens_per_second": 115573}
|
| 416 |
+
{"step": 19000, "max_steps": 24415, "train_loss": 2.820366, "learning_rate": 3.135391144899995e-05, "grad_norm": 0.6075039505958557, "tokens_seen": 2334720000, "tokens_per_second": 115571}
|
| 417 |
+
{"step": 19000, "validation_loss": 2.871389, "validation_perplexity": 17.6615}
|
| 418 |
+
{"step": 19050, "max_steps": 24415, "train_loss": 2.712609, "learning_rate": 3.106583508054977e-05, "grad_norm": 0.5940249562263489, "tokens_seen": 2340864000, "tokens_per_second": 113400}
|
| 419 |
+
{"step": 19100, "max_steps": 24415, "train_loss": 2.764279, "learning_rate": 3.077997544510766e-05, "grad_norm": 0.6305410861968994, "tokens_seen": 2347008000, "tokens_per_second": 115566}
|
| 420 |
+
{"step": 19150, "max_steps": 24415, "train_loss": 2.63859, "learning_rate": 3.049634486278204e-05, "grad_norm": 0.6639687418937683, "tokens_seen": 2353152000, "tokens_per_second": 115546}
|
| 421 |
+
{"step": 19200, "max_steps": 24415, "train_loss": 2.781917, "learning_rate": 3.0214955557612656e-05, "grad_norm": 0.6226011514663696, "tokens_seen": 2359296000, "tokens_per_second": 115559}
|
| 422 |
+
{"step": 19250, "max_steps": 24415, "train_loss": 2.786218, "learning_rate": 2.9935819657043577e-05, "grad_norm": 0.5818777680397034, "tokens_seen": 2365440000, "tokens_per_second": 115539}
|
| 423 |
+
{"step": 19300, "max_steps": 24415, "train_loss": 2.766167, "learning_rate": 2.9658949191400703e-05, "grad_norm": 0.5710976123809814, "tokens_seen": 2371584000, "tokens_per_second": 115537}
|
| 424 |
+
{"step": 19350, "max_steps": 24415, "train_loss": 2.858842, "learning_rate": 2.9384356093373116e-05, "grad_norm": 0.6166850328445435, "tokens_seen": 2377728000, "tokens_per_second": 115525}
|
| 425 |
+
{"step": 19400, "max_steps": 24415, "train_loss": 2.762764, "learning_rate": 2.9112052197498966e-05, "grad_norm": 0.6117745637893677, "tokens_seen": 2383872000, "tokens_per_second": 115610}
|
| 426 |
+
{"step": 19450, "max_steps": 24415, "train_loss": 2.673555, "learning_rate": 2.8842049239655226e-05, "grad_norm": 0.6982988119125366, "tokens_seen": 2390016000, "tokens_per_second": 115678}
|
| 427 |
+
{"step": 19500, "max_steps": 24415, "train_loss": 2.801038, "learning_rate": 2.8574358856552077e-05, "grad_norm": 0.5987298488616943, "tokens_seen": 2396160000, "tokens_per_second": 115661}
|
| 428 |
+
{"step": 19550, "max_steps": 24415, "train_loss": 2.860203, "learning_rate": 2.830899258523124e-05, "grad_norm": 0.6768466234207153, "tokens_seen": 2402304000, "tokens_per_second": 115647}
|
| 429 |
+
{"step": 19600, "max_steps": 24415, "train_loss": 2.757229, "learning_rate": 2.8045961862568865e-05, "grad_norm": 0.7006207704544067, "tokens_seen": 2408448000, "tokens_per_second": 115601}
|
| 430 |
+
{"step": 19650, "max_steps": 24415, "train_loss": 2.685526, "learning_rate": 2.7785278024782547e-05, "grad_norm": 0.614594042301178, "tokens_seen": 2414592000, "tokens_per_second": 115549}
|
| 431 |
+
{"step": 19700, "max_steps": 24415, "train_loss": 2.704333, "learning_rate": 2.752695230694275e-05, "grad_norm": 0.6088916659355164, "tokens_seen": 2420736000, "tokens_per_second": 115564}
|
| 432 |
+
{"step": 19750, "max_steps": 24415, "train_loss": 2.555395, "learning_rate": 2.727099584248864e-05, "grad_norm": 0.6791955828666687, "tokens_seen": 2426880000, "tokens_per_second": 115597}
|
| 433 |
+
{"step": 19800, "max_steps": 24415, "train_loss": 2.652574, "learning_rate": 2.701741966274819e-05, "grad_norm": 0.6295214295387268, "tokens_seen": 2433024000, "tokens_per_second": 115581}
|
| 434 |
+
{"step": 19850, "max_steps": 24415, "train_loss": 2.805314, "learning_rate": 2.6766234696462827e-05, "grad_norm": 0.5992781519889832, "tokens_seen": 2439168000, "tokens_per_second": 115620}
|
| 435 |
+
{"step": 19900, "max_steps": 24415, "train_loss": 2.779102, "learning_rate": 2.6517451769316337e-05, "grad_norm": 0.5922258496284485, "tokens_seen": 2445312000, "tokens_per_second": 115585}
|
| 436 |
+
{"step": 19950, "max_steps": 24415, "train_loss": 2.606874, "learning_rate": 2.6271081603468357e-05, "grad_norm": 0.6390422582626343, "tokens_seen": 2451456000, "tokens_per_second": 115571}
|
| 437 |
+
{"step": 20000, "max_steps": 24415, "train_loss": 2.831885, "learning_rate": 2.6027134817092206e-05, "grad_norm": 0.6103304028511047, "tokens_seen": 2457600000, "tokens_per_second": 115561}
|
| 438 |
+
{"step": 20000, "validation_loss": 2.86737, "validation_perplexity": 17.5907}
|
| 439 |
+
Saved checkpoint: runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00020000
|
| 440 |
+
{"step": 20050, "max_steps": 24415, "train_loss": 2.725566, "learning_rate": 2.5785621923917355e-05, "grad_norm": 0.6002763509750366, "tokens_seen": 2463744000, "tokens_per_second": 111852}
|
| 441 |
+
{"step": 20100, "max_steps": 24415, "train_loss": 2.882972, "learning_rate": 2.5546553332776153e-05, "grad_norm": 0.6204449534416199, "tokens_seen": 2469888000, "tokens_per_second": 115589}
|
| 442 |
+
{"step": 20150, "max_steps": 24415, "train_loss": 2.607282, "learning_rate": 2.5309939347155386e-05, "grad_norm": 0.6274742484092712, "tokens_seen": 2476032000, "tokens_per_second": 115585}
|
| 443 |
+
{"step": 20200, "max_steps": 24415, "train_loss": 2.657714, "learning_rate": 2.5075790164752117e-05, "grad_norm": 0.7395855188369751, "tokens_seen": 2482176000, "tokens_per_second": 115572}
|
| 444 |
+
{"step": 20250, "max_steps": 24415, "train_loss": 2.80093, "learning_rate": 2.4844115877034186e-05, "grad_norm": 0.6137728095054626, "tokens_seen": 2488320000, "tokens_per_second": 115570}
|
| 445 |
+
{"step": 20300, "max_steps": 24415, "train_loss": 2.673792, "learning_rate": 2.4614926468805284e-05, "grad_norm": 0.6156178712844849, "tokens_seen": 2494464000, "tokens_per_second": 115583}
|
| 446 |
+
{"step": 20350, "max_steps": 24415, "train_loss": 2.843575, "learning_rate": 2.4388231817774697e-05, "grad_norm": 0.704043447971344, "tokens_seen": 2500608000, "tokens_per_second": 115584}
|
| 447 |
+
{"step": 20400, "max_steps": 24415, "train_loss": 2.882671, "learning_rate": 2.4164041694131442e-05, "grad_norm": 0.6543155908584595, "tokens_seen": 2506752000, "tokens_per_second": 115685}
|
| 448 |
+
{"step": 20450, "max_steps": 24415, "train_loss": 2.713107, "learning_rate": 2.3942365760123353e-05, "grad_norm": 0.6198663115501404, "tokens_seen": 2512896000, "tokens_per_second": 115752}
|
| 449 |
+
{"step": 20500, "max_steps": 24415, "train_loss": 2.695461, "learning_rate": 2.372321356964054e-05, "grad_norm": 0.6344349384307861, "tokens_seen": 2519040000, "tokens_per_second": 115728}
|
| 450 |
+
{"step": 20550, "max_steps": 24415, "train_loss": 2.793643, "learning_rate": 2.3506594567803688e-05, "grad_norm": 0.5910429358482361, "tokens_seen": 2525184000, "tokens_per_second": 115695}
|
| 451 |
+
{"step": 20600, "max_steps": 24415, "train_loss": 2.825883, "learning_rate": 2.3292518090556933e-05, "grad_norm": 0.587986946105957, "tokens_seen": 2531328000, "tokens_per_second": 115675}
|
| 452 |
+
{"step": 20650, "max_steps": 24415, "train_loss": 2.76228, "learning_rate": 2.3080993364265586e-05, "grad_norm": 0.672764778137207, "tokens_seen": 2537472000, "tokens_per_second": 115646}
|
| 453 |
+
{"step": 20700, "max_steps": 24415, "train_loss": 2.885475, "learning_rate": 2.2872029505318372e-05, "grad_norm": 0.6077940464019775, "tokens_seen": 2543616000, "tokens_per_second": 115661}
|
| 454 |
+
{"step": 20750, "max_steps": 24415, "train_loss": 2.764343, "learning_rate": 2.2665635519734637e-05, "grad_norm": 0.6056102514266968, "tokens_seen": 2549760000, "tokens_per_second": 115652}
|
| 455 |
+
{"step": 20800, "max_steps": 24415, "train_loss": 2.62957, "learning_rate": 2.2461820302776146e-05, "grad_norm": 0.6368304491043091, "tokens_seen": 2555904000, "tokens_per_second": 115642}
|
| 456 |
+
{"step": 20850, "max_steps": 24415, "train_loss": 2.761528, "learning_rate": 2.2260592638563736e-05, "grad_norm": 0.6025370359420776, "tokens_seen": 2562048000, "tokens_per_second": 115624}
|
| 457 |
+
{"step": 20900, "max_steps": 24415, "train_loss": 2.83255, "learning_rate": 2.2061961199698666e-05, "grad_norm": 0.6237384676933289, "tokens_seen": 2568192000, "tokens_per_second": 115615}
|
| 458 |
+
{"step": 20950, "max_steps": 24415, "train_loss": 2.722114, "learning_rate": 2.186593454688897e-05, "grad_norm": 0.6229617595672607, "tokens_seen": 2574336000, "tokens_per_second": 115642}
|
| 459 |
+
{"step": 21000, "max_steps": 24415, "train_loss": 2.781463, "learning_rate": 2.167252112858037e-05, "grad_norm": 0.5941027402877808, "tokens_seen": 2580480000, "tokens_per_second": 115625}
|
| 460 |
+
{"step": 21000, "validation_loss": 2.864438, "validation_perplexity": 17.5392}
|
| 461 |
+
{"step": 21050, "max_steps": 24415, "train_loss": 2.72432, "learning_rate": 2.148172928059226e-05, "grad_norm": 0.6507408022880554, "tokens_seen": 2586624000, "tokens_per_second": 113491}
|
| 462 |
+
{"step": 21100, "max_steps": 24415, "train_loss": 2.67957, "learning_rate": 2.1293567225758365e-05, "grad_norm": 0.6184972524642944, "tokens_seen": 2592768000, "tokens_per_second": 115608}
|
| 463 |
+
{"step": 21150, "max_steps": 24415, "train_loss": 2.896593, "learning_rate": 2.1108043073572438e-05, "grad_norm": 0.5926744937896729, "tokens_seen": 2598912000, "tokens_per_second": 115619}
|
| 464 |
+
{"step": 21200, "max_steps": 24415, "train_loss": 2.661033, "learning_rate": 2.0925164819838655e-05, "grad_norm": 0.65669846534729, "tokens_seen": 2605056000, "tokens_per_second": 115608}
|
| 465 |
+
{"step": 21250, "max_steps": 24415, "train_loss": 2.70118, "learning_rate": 2.074494034632711e-05, "grad_norm": 0.6318796873092651, "tokens_seen": 2611200000, "tokens_per_second": 115653}
|
| 466 |
+
{"step": 21300, "max_steps": 24415, "train_loss": 2.592898, "learning_rate": 2.0567377420434022e-05, "grad_norm": 0.630044162273407, "tokens_seen": 2617344000, "tokens_per_second": 115661}
|
| 467 |
+
{"step": 21350, "max_steps": 24415, "train_loss": 2.703647, "learning_rate": 2.0392483694847048e-05, "grad_norm": 0.6456885933876038, "tokens_seen": 2623488000, "tokens_per_second": 115661}
|
| 468 |
+
{"step": 21400, "max_steps": 24415, "train_loss": 2.721252, "learning_rate": 2.0220266707215445e-05, "grad_norm": 0.5997530221939087, "tokens_seen": 2629632000, "tokens_per_second": 115649}
|
| 469 |
+
{"step": 21450, "max_steps": 24415, "train_loss": 2.855813, "learning_rate": 2.00507338798252e-05, "grad_norm": 0.6231884956359863, "tokens_seen": 2635776000, "tokens_per_second": 115664}
|
| 470 |
+
{"step": 21500, "max_steps": 24415, "train_loss": 2.733331, "learning_rate": 1.9883892519279097e-05, "grad_norm": 0.5984932780265808, "tokens_seen": 2641920000, "tokens_per_second": 115628}
|
| 471 |
+
{"step": 21550, "max_steps": 24415, "train_loss": 2.784818, "learning_rate": 1.9719749816181928e-05, "grad_norm": 0.5977855920791626, "tokens_seen": 2648064000, "tokens_per_second": 115644}
|
| 472 |
+
{"step": 21600, "max_steps": 24415, "train_loss": 2.749432, "learning_rate": 1.9558312844830467e-05, "grad_norm": 0.5922501683235168, "tokens_seen": 2654208000, "tokens_per_second": 115647}
|
| 473 |
+
{"step": 21650, "max_steps": 24415, "train_loss": 2.751568, "learning_rate": 1.939958856290866e-05, "grad_norm": 0.6253129839897156, "tokens_seen": 2660352000, "tokens_per_second": 115641}
|
| 474 |
+
{"step": 21700, "max_steps": 24415, "train_loss": 2.851732, "learning_rate": 1.9243583811187723e-05, "grad_norm": 0.6306662559509277, "tokens_seen": 2666496000, "tokens_per_second": 115673}
|
| 475 |
+
{"step": 21750, "max_steps": 24415, "train_loss": 2.777505, "learning_rate": 1.909030531323133e-05, "grad_norm": 0.6194624304771423, "tokens_seen": 2672640000, "tokens_per_second": 115657}
|
| 476 |
+
{"step": 21800, "max_steps": 24415, "train_loss": 2.689925, "learning_rate": 1.8939759675105804e-05, "grad_norm": 0.6342402100563049, "tokens_seen": 2678784000, "tokens_per_second": 115669}
|
| 477 |
+
{"step": 21850, "max_steps": 24415, "train_loss": 2.78969, "learning_rate": 1.879195338509549e-05, "grad_norm": 0.6131957173347473, "tokens_seen": 2684928000, "tokens_per_second": 115672}
|
| 478 |
+
{"step": 21900, "max_steps": 24415, "train_loss": 2.616148, "learning_rate": 1.864689281342299e-05, "grad_norm": 0.6286548972129822, "tokens_seen": 2691072000, "tokens_per_second": 115696}
|
| 479 |
+
{"step": 21950, "max_steps": 24415, "train_loss": 2.780351, "learning_rate": 1.8504584211974754e-05, "grad_norm": 0.6103248596191406, "tokens_seen": 2697216000, "tokens_per_second": 115681}
|
| 480 |
+
{"step": 22000, "max_steps": 24415, "train_loss": 2.774347, "learning_rate": 1.8365033714031534e-05, "grad_norm": 0.6407751441001892, "tokens_seen": 2703360000, "tokens_per_second": 115697}
|
| 481 |
+
{"step": 22000, "validation_loss": 2.861929, "validation_perplexity": 17.4952}
|
| 482 |
+
{"step": 22050, "max_steps": 24415, "train_loss": 2.852945, "learning_rate": 1.8228247334004103e-05, "grad_norm": 0.6118343472480774, "tokens_seen": 2709504000, "tokens_per_second": 113499}
|
| 483 |
+
{"step": 22100, "max_steps": 24415, "train_loss": 2.627281, "learning_rate": 1.8094230967173996e-05, "grad_norm": 0.6102780699729919, "tokens_seen": 2715648000, "tokens_per_second": 115659}
|
| 484 |
+
{"step": 22150, "max_steps": 24415, "train_loss": 2.767095, "learning_rate": 1.79629903894395e-05, "grad_norm": 0.6172248721122742, "tokens_seen": 2721792000, "tokens_per_second": 115656}
|
| 485 |
+
{"step": 22200, "max_steps": 24415, "train_loss": 2.762802, "learning_rate": 1.7834531257066644e-05, "grad_norm": 0.6430047154426575, "tokens_seen": 2727936000, "tokens_per_second": 115667}
|
| 486 |
+
{"step": 22250, "max_steps": 24415, "train_loss": 2.763437, "learning_rate": 1.7708859106445492e-05, "grad_norm": 0.6189215779304504, "tokens_seen": 2734080000, "tokens_per_second": 115661}
|
| 487 |
+
{"step": 22300, "max_steps": 24415, "train_loss": 2.864902, "learning_rate": 1.7585979353851473e-05, "grad_norm": 0.6051227450370789, "tokens_seen": 2740224000, "tokens_per_second": 115640}
|
| 488 |
+
{"step": 22350, "max_steps": 24415, "train_loss": 2.78242, "learning_rate": 1.746589729521201e-05, "grad_norm": 0.6032001972198486, "tokens_seen": 2746368000, "tokens_per_second": 115630}
|
| 489 |
+
{"step": 22400, "max_steps": 24415, "train_loss": 2.719381, "learning_rate": 1.7348618105878208e-05, "grad_norm": 0.6332842111587524, "tokens_seen": 2752512000, "tokens_per_second": 115634}
|
| 490 |
+
{"step": 22450, "max_steps": 24415, "train_loss": 2.830464, "learning_rate": 1.7234146840401873e-05, "grad_norm": 0.6493021845817566, "tokens_seen": 2758656000, "tokens_per_second": 115615}
|
| 491 |
+
{"step": 22500, "max_steps": 24415, "train_loss": 2.843654, "learning_rate": 1.7122488432317588e-05, "grad_norm": 0.6092942953109741, "tokens_seen": 2764800000, "tokens_per_second": 115618}
|
| 492 |
+
{"step": 22550, "max_steps": 24415, "train_loss": 2.706203, "learning_rate": 1.701364769393017e-05, "grad_norm": 0.6111382842063904, "tokens_seen": 2770944000, "tokens_per_second": 115621}
|
| 493 |
+
{"step": 22600, "max_steps": 24415, "train_loss": 2.776963, "learning_rate": 1.6907629316107217e-05, "grad_norm": 0.657183051109314, "tokens_seen": 2777088000, "tokens_per_second": 115621}
|
| 494 |
+
{"step": 22650, "max_steps": 24415, "train_loss": 2.778902, "learning_rate": 1.6804437868076928e-05, "grad_norm": 0.608390212059021, "tokens_seen": 2783232000, "tokens_per_second": 115630}
|
| 495 |
+
{"step": 22700, "max_steps": 24415, "train_loss": 2.763429, "learning_rate": 1.670407779723123e-05, "grad_norm": 0.6772242188453674, "tokens_seen": 2789376000, "tokens_per_second": 115635}
|
| 496 |
+
{"step": 22750, "max_steps": 24415, "train_loss": 2.697392, "learning_rate": 1.6606553428934048e-05, "grad_norm": 0.8745754957199097, "tokens_seen": 2795520000, "tokens_per_second": 115629}
|
| 497 |
+
{"step": 22800, "max_steps": 24415, "train_loss": 2.857977, "learning_rate": 1.6511868966334915e-05, "grad_norm": 0.6229487657546997, "tokens_seen": 2801664000, "tokens_per_second": 115645}
|
| 498 |
+
{"step": 22850, "max_steps": 24415, "train_loss": 2.568233, "learning_rate": 1.6420028490187808e-05, "grad_norm": 0.6349725723266602, "tokens_seen": 2807808000, "tokens_per_second": 115634}
|
| 499 |
+
{"step": 22900, "max_steps": 24415, "train_loss": 2.643429, "learning_rate": 1.633103595867531e-05, "grad_norm": 0.6417720317840576, "tokens_seen": 2813952000, "tokens_per_second": 115643}
|
| 500 |
+
{"step": 22950, "max_steps": 24415, "train_loss": 2.868518, "learning_rate": 1.6244895207237968e-05, "grad_norm": 0.5957947969436646, "tokens_seen": 2820096000, "tokens_per_second": 115650}
|
| 501 |
+
{"step": 23000, "max_steps": 24415, "train_loss": 2.711414, "learning_rate": 1.616160994840903e-05, "grad_norm": 0.6176562905311584, "tokens_seen": 2826240000, "tokens_per_second": 115621}
|
| 502 |
+
{"step": 23000, "validation_loss": 2.860038, "validation_perplexity": 17.4622}
|
| 503 |
+
{"step": 23050, "max_steps": 24415, "train_loss": 2.736907, "learning_rate": 1.608118377165442e-05, "grad_norm": 0.777596116065979, "tokens_seen": 2832384000, "tokens_per_second": 113486}
|
| 504 |
+
{"step": 23100, "max_steps": 24415, "train_loss": 2.700006, "learning_rate": 1.6003620143218037e-05, "grad_norm": 0.6337012052536011, "tokens_seen": 2838528000, "tokens_per_second": 115625}
|
| 505 |
+
{"step": 23150, "max_steps": 24415, "train_loss": 2.781038, "learning_rate": 1.5928922405972378e-05, "grad_norm": 0.6418102979660034, "tokens_seen": 2844672000, "tokens_per_second": 115618}
|
| 506 |
+
{"step": 23200, "max_steps": 24415, "train_loss": 2.774231, "learning_rate": 1.5857093779274477e-05, "grad_norm": 0.6108967065811157, "tokens_seen": 2850816000, "tokens_per_second": 115636}
|
| 507 |
+
{"step": 23250, "max_steps": 24415, "train_loss": 2.718765, "learning_rate": 1.5788137358827107e-05, "grad_norm": 0.6196231245994568, "tokens_seen": 2856960000, "tokens_per_second": 115648}
|
| 508 |
+
{"step": 23300, "max_steps": 24415, "train_loss": 2.835135, "learning_rate": 1.572205611654541e-05, "grad_norm": 0.5968915820121765, "tokens_seen": 2863104000, "tokens_per_second": 115646}
|
| 509 |
+
{"step": 23350, "max_steps": 24415, "train_loss": 2.845595, "learning_rate": 1.5658852900428792e-05, "grad_norm": 0.6030846834182739, "tokens_seen": 2869248000, "tokens_per_second": 115626}
|
| 510 |
+
{"step": 23400, "max_steps": 24415, "train_loss": 2.771462, "learning_rate": 1.5598530434438176e-05, "grad_norm": 0.6051156520843506, "tokens_seen": 2875392000, "tokens_per_second": 115623}
|
| 511 |
+
{"step": 23450, "max_steps": 24415, "train_loss": 2.774949, "learning_rate": 1.55410913183786e-05, "grad_norm": 0.60307776927948, "tokens_seen": 2881536000, "tokens_per_second": 115646}
|
| 512 |
+
{"step": 23500, "max_steps": 24415, "train_loss": 2.73786, "learning_rate": 1.5486538027787192e-05, "grad_norm": 0.6082613468170166, "tokens_seen": 2887680000, "tokens_per_second": 115662}
|
| 513 |
+
{"step": 23550, "max_steps": 24415, "train_loss": 2.694792, "learning_rate": 1.5434872913826446e-05, "grad_norm": 0.6187634468078613, "tokens_seen": 2893824000, "tokens_per_second": 115659}
|
| 514 |
+
{"step": 23600, "max_steps": 24415, "train_loss": 2.712125, "learning_rate": 1.5386098203182932e-05, "grad_norm": 0.6332436800003052, "tokens_seen": 2899968000, "tokens_per_second": 115666}
|
| 515 |
+
{"step": 23650, "max_steps": 24415, "train_loss": 2.709451, "learning_rate": 1.5340215997971285e-05, "grad_norm": 0.6231958866119385, "tokens_seen": 2906112000, "tokens_per_second": 115629}
|
| 516 |
+
{"step": 23700, "max_steps": 24415, "train_loss": 2.908402, "learning_rate": 1.5297228275643623e-05, "grad_norm": 0.625374972820282, "tokens_seen": 2912256000, "tokens_per_second": 115644}
|
| 517 |
+
{"step": 23750, "max_steps": 24415, "train_loss": 2.825391, "learning_rate": 1.5257136888904332e-05, "grad_norm": 0.6172897815704346, "tokens_seen": 2918400000, "tokens_per_second": 115627}
|
| 518 |
+
{"step": 23800, "max_steps": 24415, "train_loss": 2.777977, "learning_rate": 1.5219943565630226e-05, "grad_norm": 0.6453419327735901, "tokens_seen": 2924544000, "tokens_per_second": 115656}
|
| 519 |
+
{"step": 23850, "max_steps": 24415, "train_loss": 2.808457, "learning_rate": 1.5185649908796045e-05, "grad_norm": 0.6297765374183655, "tokens_seen": 2930688000, "tokens_per_second": 115677}
|
| 520 |
+
{"step": 23900, "max_steps": 24415, "train_loss": 2.827693, "learning_rate": 1.5154257396405384e-05, "grad_norm": 0.6233020424842834, "tokens_seen": 2936832000, "tokens_per_second": 115662}
|
| 521 |
+
{"step": 23950, "max_steps": 24415, "train_loss": 2.777882, "learning_rate": 1.5125767381427013e-05, "grad_norm": 0.5891589522361755, "tokens_seen": 2942976000, "tokens_per_second": 115657}
|
| 522 |
+
{"step": 24000, "max_steps": 24415, "train_loss": 2.750225, "learning_rate": 1.5100181091736548e-05, "grad_norm": 0.5986984372138977, "tokens_seen": 2949120000, "tokens_per_second": 115623}
|
| 523 |
+
{"step": 24000, "validation_loss": 2.858127, "validation_perplexity": 17.4289}
|
| 524 |
+
{"step": 24050, "max_steps": 24415, "train_loss": 2.746184, "learning_rate": 1.507749963006351e-05, "grad_norm": 0.6033036708831787, "tokens_seen": 2955264000, "tokens_per_second": 113453}
|
| 525 |
+
{"step": 24100, "max_steps": 24415, "train_loss": 2.776009, "learning_rate": 1.505772397394384e-05, "grad_norm": 0.6058425307273865, "tokens_seen": 2961408000, "tokens_per_second": 115635}
|
| 526 |
+
{"step": 24150, "max_steps": 24415, "train_loss": 2.701993, "learning_rate": 1.504085497567777e-05, "grad_norm": 0.6101545095443726, "tokens_seen": 2967552000, "tokens_per_second": 115660}
|
| 527 |
+
{"step": 24200, "max_steps": 24415, "train_loss": 2.651574, "learning_rate": 1.5026893362293033e-05, "grad_norm": 0.6251341700553894, "tokens_seen": 2973696000, "tokens_per_second": 115664}
|
| 528 |
+
{"step": 24250, "max_steps": 24415, "train_loss": 2.731524, "learning_rate": 1.501583973551359e-05, "grad_norm": 0.6173787117004395, "tokens_seen": 2979840000, "tokens_per_second": 115673}
|
| 529 |
+
{"step": 24300, "max_steps": 24415, "train_loss": 2.854924, "learning_rate": 1.5007694571733676e-05, "grad_norm": 0.607036292552948, "tokens_seen": 2985984000, "tokens_per_second": 115666}
|
| 530 |
+
{"step": 24350, "max_steps": 24415, "train_loss": 2.92489, "learning_rate": 1.5002458221997248e-05, "grad_norm": 0.6901952624320984, "tokens_seen": 2992128000, "tokens_per_second": 115680}
|
| 531 |
+
{"step": 24400, "max_steps": 24415, "train_loss": 2.692121, "learning_rate": 1.5000130911982901e-05, "grad_norm": 0.6857115030288696, "tokens_seen": 2998272000, "tokens_per_second": 115647}
|
| 532 |
+
Final checkpoint: runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415
|
reports/fixed-recovery-6b/byte-evaluation.json
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/workspace/KAWK50M/runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00038147",
|
| 3 |
+
"tokenizer": "/workspace/KAWK50M/runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00038147/tokenizer",
|
| 4 |
+
"inputs": [
|
| 5 |
+
"/workspace/KAWK50M/data/clean-v2-full/validation-00000.jsonl"
|
| 6 |
+
],
|
| 7 |
+
"documents": 500,
|
| 8 |
+
"characters": 792512,
|
| 9 |
+
"utf8_bytes": 1888545,
|
| 10 |
+
"predicted_tokens": 418575,
|
| 11 |
+
"context_length": 1024,
|
| 12 |
+
"token_nll": 2.8874621830420604,
|
| 13 |
+
"token_perplexity": 17.947703769418208,
|
| 14 |
+
"nats_per_character": 1.525048810954068,
|
| 15 |
+
"bits_per_utf8_byte": 0.9232871521890333,
|
| 16 |
+
"note": "Compare bits_per_utf8_byte only when checkpoints are evaluated on the exact same ordered raw-text documents."
|
| 17 |
+
}
|
reports/fixed-recovery-6b/final_checkpoint.txt
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00038147
|
reports/fixed-recovery-6b/train.log
ADDED
|
@@ -0,0 +1,423 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"parameters": 51542528,
|
| 3 |
+
"device": "cuda:0",
|
| 4 |
+
"world_size": 1,
|
| 5 |
+
"sequence_length": 1024,
|
| 6 |
+
"training_sample_offset": 976640,
|
| 7 |
+
"tokens_per_optimizer_step": 131072,
|
| 8 |
+
"max_steps": 38147,
|
| 9 |
+
"planned_tokens": 5000003584,
|
| 10 |
+
"start_step": 0,
|
| 11 |
+
"initialized_from": "/workspace/KAWK50M/runs/kawk50m-fixed-recovery-1b-5090/checkpoint-00007630",
|
| 12 |
+
"precision": "bf16",
|
| 13 |
+
"compile": true
|
| 14 |
+
}
|
| 15 |
+
{"step": 1, "max_steps": 38147, "train_loss": 2.974643, "learning_rate": 2.6212319790301444e-07, "grad_norm": 0.4614468216896057, "tokens_seen": 131072, "tokens_per_second": 46891}
|
| 16 |
+
{"step": 100, "max_steps": 38147, "train_loss": 2.908479, "learning_rate": 2.6212319790301444e-05, "grad_norm": 0.48229220509529114, "tokens_seen": 13107200, "tokens_per_second": 266634}
|
| 17 |
+
{"step": 200, "max_steps": 38147, "train_loss": 2.938294, "learning_rate": 5.242463958060289e-05, "grad_norm": 0.48655542731285095, "tokens_seen": 26214400, "tokens_per_second": 266632}
|
| 18 |
+
{"step": 300, "max_steps": 38147, "train_loss": 2.881706, "learning_rate": 7.863695937090433e-05, "grad_norm": 0.481501966714859, "tokens_seen": 39321600, "tokens_per_second": 266703}
|
| 19 |
+
{"step": 400, "max_steps": 38147, "train_loss": 2.942861, "learning_rate": 0.00010484927916120577, "grad_norm": 0.5160865187644958, "tokens_seen": 52428800, "tokens_per_second": 266702}
|
| 20 |
+
{"step": 500, "max_steps": 38147, "train_loss": 2.871376, "learning_rate": 0.00013106159895150723, "grad_norm": 0.5494230389595032, "tokens_seen": 65536000, "tokens_per_second": 266629}
|
| 21 |
+
{"step": 600, "max_steps": 38147, "train_loss": 2.985494, "learning_rate": 0.00015727391874180866, "grad_norm": 0.560945451259613, "tokens_seen": 78643200, "tokens_per_second": 266656}
|
| 22 |
+
{"step": 700, "max_steps": 38147, "train_loss": 2.941133, "learning_rate": 0.00018348623853211012, "grad_norm": 0.552539587020874, "tokens_seen": 91750400, "tokens_per_second": 266689}
|
| 23 |
+
{"step": 800, "max_steps": 38147, "train_loss": 2.933866, "learning_rate": 0.0001999995649453435, "grad_norm": 0.5684460401535034, "tokens_seen": 104857600, "tokens_per_second": 266656}
|
| 24 |
+
{"step": 900, "max_steps": 38147, "train_loss": 2.923692, "learning_rate": 0.00019999403545855837, "grad_norm": 0.5598779320716858, "tokens_seen": 117964800, "tokens_per_second": 266536}
|
| 25 |
+
{"step": 1000, "max_steps": 38147, "train_loss": 2.863725, "learning_rate": 0.00019998215062313029, "grad_norm": 0.5328158736228943, "tokens_seen": 131072000, "tokens_per_second": 266536}
|
| 26 |
+
{"step": 1100, "max_steps": 38147, "train_loss": 2.988881, "learning_rate": 0.00019996391127836236, "grad_norm": 0.5750011205673218, "tokens_seen": 144179200, "tokens_per_second": 266535}
|
| 27 |
+
{"step": 1200, "max_steps": 38147, "train_loss": 2.960122, "learning_rate": 0.00019993931871231102, "grad_norm": 0.5375710725784302, "tokens_seen": 157286400, "tokens_per_second": 266542}
|
| 28 |
+
{"step": 1300, "max_steps": 38147, "train_loss": 2.945494, "learning_rate": 0.00019990837466169503, "grad_norm": 0.5387780070304871, "tokens_seen": 170393600, "tokens_per_second": 266548}
|
| 29 |
+
{"step": 1400, "max_steps": 38147, "train_loss": 2.960481, "learning_rate": 0.00019987108131177292, "grad_norm": 0.5903465151786804, "tokens_seen": 183500800, "tokens_per_second": 266531}
|
| 30 |
+
{"step": 1500, "max_steps": 38147, "train_loss": 2.951986, "learning_rate": 0.0001998274412961885, "grad_norm": 0.5386484861373901, "tokens_seen": 196608000, "tokens_per_second": 266531}
|
| 31 |
+
{"step": 1600, "max_steps": 38147, "train_loss": 2.979607, "learning_rate": 0.00019977745769678514, "grad_norm": 0.5697662830352783, "tokens_seen": 209715200, "tokens_per_second": 266557}
|
| 32 |
+
{"step": 1700, "max_steps": 38147, "train_loss": 2.934249, "learning_rate": 0.0001997211340433878, "grad_norm": 0.5244824290275574, "tokens_seen": 222822400, "tokens_per_second": 266522}
|
| 33 |
+
{"step": 1800, "max_steps": 38147, "train_loss": 2.874219, "learning_rate": 0.00019965847431355408, "grad_norm": 0.5934562683105469, "tokens_seen": 235929600, "tokens_per_second": 266475}
|
| 34 |
+
{"step": 1900, "max_steps": 38147, "train_loss": 3.009789, "learning_rate": 0.0001995894829322931, "grad_norm": 0.5320238471031189, "tokens_seen": 249036800, "tokens_per_second": 266516}
|
| 35 |
+
{"step": 2000, "max_steps": 38147, "train_loss": 3.027504, "learning_rate": 0.00019951416477175313, "grad_norm": 0.5308347940444946, "tokens_seen": 262144000, "tokens_per_second": 266528}
|
| 36 |
+
{"step": 2000, "validation_loss": 3.035152, "validation_perplexity": 20.8041}
|
| 37 |
+
{"step": 2100, "max_steps": 38147, "train_loss": 2.92116, "learning_rate": 0.00019943252515087738, "grad_norm": 0.5231989026069641, "tokens_seen": 275251200, "tokens_per_second": 248528}
|
| 38 |
+
{"step": 2200, "max_steps": 38147, "train_loss": 2.995961, "learning_rate": 0.00019934456983502856, "grad_norm": 0.5607808828353882, "tokens_seen": 288358400, "tokens_per_second": 266498}
|
| 39 |
+
{"step": 2300, "max_steps": 38147, "train_loss": 2.959142, "learning_rate": 0.0001992503050355816, "grad_norm": 0.552680253982544, "tokens_seen": 301465600, "tokens_per_second": 266486}
|
| 40 |
+
{"step": 2400, "max_steps": 38147, "train_loss": 2.962559, "learning_rate": 0.00019914973740948512, "grad_norm": 0.555330753326416, "tokens_seen": 314572800, "tokens_per_second": 266467}
|
| 41 |
+
{"step": 2500, "max_steps": 38147, "train_loss": 2.967474, "learning_rate": 0.00019904287405879116, "grad_norm": 0.5409094095230103, "tokens_seen": 327680000, "tokens_per_second": 266534}
|
| 42 |
+
{"step": 2600, "max_steps": 38147, "train_loss": 2.935567, "learning_rate": 0.00019892972253015371, "grad_norm": 0.560372531414032, "tokens_seen": 340787200, "tokens_per_second": 266536}
|
| 43 |
+
{"step": 2700, "max_steps": 38147, "train_loss": 3.014029, "learning_rate": 0.0001988102908142959, "grad_norm": 0.5003385543823242, "tokens_seen": 353894400, "tokens_per_second": 266545}
|
| 44 |
+
{"step": 2800, "max_steps": 38147, "train_loss": 2.960769, "learning_rate": 0.00019868458734544547, "grad_norm": 0.553016722202301, "tokens_seen": 367001600, "tokens_per_second": 266548}
|
| 45 |
+
{"step": 2900, "max_steps": 38147, "train_loss": 2.940143, "learning_rate": 0.00019855262100073925, "grad_norm": 0.49712949991226196, "tokens_seen": 380108800, "tokens_per_second": 266549}
|
| 46 |
+
{"step": 3000, "max_steps": 38147, "train_loss": 2.85196, "learning_rate": 0.00019841440109959638, "grad_norm": 0.5317330360412598, "tokens_seen": 393216000, "tokens_per_second": 266495}
|
| 47 |
+
{"step": 3100, "max_steps": 38147, "train_loss": 2.877238, "learning_rate": 0.00019826993740305993, "grad_norm": 0.537623405456543, "tokens_seen": 406323200, "tokens_per_second": 266546}
|
| 48 |
+
{"step": 3200, "max_steps": 38147, "train_loss": 2.941891, "learning_rate": 0.00019811924011310777, "grad_norm": 0.5330936312675476, "tokens_seen": 419430400, "tokens_per_second": 266551}
|
| 49 |
+
{"step": 3300, "max_steps": 38147, "train_loss": 3.023407, "learning_rate": 0.0001979623198719321, "grad_norm": 0.5457956194877625, "tokens_seen": 432537600, "tokens_per_second": 266546}
|
| 50 |
+
{"step": 3400, "max_steps": 38147, "train_loss": 3.006085, "learning_rate": 0.00019779918776118772, "grad_norm": 0.5238074660301208, "tokens_seen": 445644800, "tokens_per_second": 266553}
|
| 51 |
+
{"step": 3500, "max_steps": 38147, "train_loss": 2.975786, "learning_rate": 0.00019762985530120971, "grad_norm": 0.5653413534164429, "tokens_seen": 458752000, "tokens_per_second": 266553}
|
| 52 |
+
{"step": 3600, "max_steps": 38147, "train_loss": 2.91367, "learning_rate": 0.00019745433445019964, "grad_norm": 0.5223200917243958, "tokens_seen": 471859200, "tokens_per_second": 266552}
|
| 53 |
+
{"step": 3700, "max_steps": 38147, "train_loss": 2.930792, "learning_rate": 0.00019727263760338127, "grad_norm": 0.5368753671646118, "tokens_seen": 484966400, "tokens_per_second": 266565}
|
| 54 |
+
{"step": 3800, "max_steps": 38147, "train_loss": 3.004488, "learning_rate": 0.000197084777592125, "grad_norm": 0.5460859537124634, "tokens_seen": 498073600, "tokens_per_second": 266578}
|
| 55 |
+
{"step": 3900, "max_steps": 38147, "train_loss": 2.976407, "learning_rate": 0.00019689076768304197, "grad_norm": 0.5260820984840393, "tokens_seen": 511180800, "tokens_per_second": 266564}
|
| 56 |
+
{"step": 4000, "max_steps": 38147, "train_loss": 2.959824, "learning_rate": 0.00019669062157704693, "grad_norm": 0.5276979804039001, "tokens_seen": 524288000, "tokens_per_second": 266553}
|
| 57 |
+
{"step": 4000, "validation_loss": 3.024725, "validation_perplexity": 20.5883}
|
| 58 |
+
{"step": 4100, "max_steps": 38147, "train_loss": 3.067965, "learning_rate": 0.00019648435340839087, "grad_norm": 0.5293189883232117, "tokens_seen": 537395200, "tokens_per_second": 262466}
|
| 59 |
+
{"step": 4200, "max_steps": 38147, "train_loss": 2.972997, "learning_rate": 0.0001962719777436628, "grad_norm": 0.5041415095329285, "tokens_seen": 550502400, "tokens_per_second": 266505}
|
| 60 |
+
{"step": 4300, "max_steps": 38147, "train_loss": 2.953, "learning_rate": 0.00019605350958076104, "grad_norm": 0.5775559544563293, "tokens_seen": 563609600, "tokens_per_second": 266564}
|
| 61 |
+
{"step": 4400, "max_steps": 38147, "train_loss": 2.909318, "learning_rate": 0.00019582896434783403, "grad_norm": 0.6295892596244812, "tokens_seen": 576716800, "tokens_per_second": 266552}
|
| 62 |
+
{"step": 4500, "max_steps": 38147, "train_loss": 2.940704, "learning_rate": 0.00019559835790219094, "grad_norm": 0.512575089931488, "tokens_seen": 589824000, "tokens_per_second": 266556}
|
| 63 |
+
{"step": 4600, "max_steps": 38147, "train_loss": 2.884395, "learning_rate": 0.00019536170652918175, "grad_norm": 0.5334091186523438, "tokens_seen": 602931200, "tokens_per_second": 266554}
|
| 64 |
+
{"step": 4700, "max_steps": 38147, "train_loss": 2.972331, "learning_rate": 0.0001951190269410471, "grad_norm": 0.52705317735672, "tokens_seen": 616038400, "tokens_per_second": 266553}
|
| 65 |
+
{"step": 4800, "max_steps": 38147, "train_loss": 2.907125, "learning_rate": 0.00019487033627573833, "grad_norm": 0.5539998412132263, "tokens_seen": 629145600, "tokens_per_second": 266678}
|
| 66 |
+
{"step": 4900, "max_steps": 38147, "train_loss": 2.931528, "learning_rate": 0.0001946156520957068, "grad_norm": 0.5631355047225952, "tokens_seen": 642252800, "tokens_per_second": 266645}
|
| 67 |
+
{"step": 5000, "max_steps": 38147, "train_loss": 2.89311, "learning_rate": 0.00019435499238666415, "grad_norm": 0.5504894852638245, "tokens_seen": 655360000, "tokens_per_second": 266553}
|
| 68 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00005000
|
| 69 |
+
{"step": 5100, "max_steps": 38147, "train_loss": 3.031921, "learning_rate": 0.0001940883755563116, "grad_norm": 0.5478340983390808, "tokens_seen": 668467200, "tokens_per_second": 263557}
|
| 70 |
+
{"step": 5200, "max_steps": 38147, "train_loss": 2.886816, "learning_rate": 0.00019381582043304047, "grad_norm": 0.5800094604492188, "tokens_seen": 681574400, "tokens_per_second": 266566}
|
| 71 |
+
{"step": 5300, "max_steps": 38147, "train_loss": 2.953113, "learning_rate": 0.0001935373462646023, "grad_norm": 0.5333589911460876, "tokens_seen": 694681600, "tokens_per_second": 266559}
|
| 72 |
+
{"step": 5400, "max_steps": 38147, "train_loss": 2.936551, "learning_rate": 0.00019325297271674955, "grad_norm": 0.5394105911254883, "tokens_seen": 707788800, "tokens_per_second": 266519}
|
| 73 |
+
{"step": 5500, "max_steps": 38147, "train_loss": 3.012462, "learning_rate": 0.00019296271987184695, "grad_norm": 2.051757335662842, "tokens_seen": 720896000, "tokens_per_second": 266554}
|
| 74 |
+
{"step": 5600, "max_steps": 38147, "train_loss": 2.913292, "learning_rate": 0.0001926666082274532, "grad_norm": 0.48872488737106323, "tokens_seen": 734003200, "tokens_per_second": 266553}
|
| 75 |
+
{"step": 5700, "max_steps": 38147, "train_loss": 2.958948, "learning_rate": 0.00019236465869487346, "grad_norm": 0.5317707061767578, "tokens_seen": 747110400, "tokens_per_second": 266557}
|
| 76 |
+
{"step": 5800, "max_steps": 38147, "train_loss": 2.885794, "learning_rate": 0.00019205689259768255, "grad_norm": 0.531381368637085, "tokens_seen": 760217600, "tokens_per_second": 266527}
|
| 77 |
+
{"step": 5900, "max_steps": 38147, "train_loss": 2.895442, "learning_rate": 0.00019174333167021923, "grad_norm": 0.5155215263366699, "tokens_seen": 773324800, "tokens_per_second": 266491}
|
| 78 |
+
{"step": 6000, "max_steps": 38147, "train_loss": 2.999445, "learning_rate": 0.00019142399805605113, "grad_norm": 0.526485800743103, "tokens_seen": 786432000, "tokens_per_second": 266492}
|
| 79 |
+
{"step": 6000, "validation_loss": 3.016271, "validation_perplexity": 20.415}
|
| 80 |
+
{"step": 6100, "max_steps": 38147, "train_loss": 2.872324, "learning_rate": 0.00019109891430641113, "grad_norm": 0.5473472476005554, "tokens_seen": 799539200, "tokens_per_second": 262481}
|
| 81 |
+
{"step": 6200, "max_steps": 38147, "train_loss": 2.965532, "learning_rate": 0.00019076810337860484, "grad_norm": 0.5530160665512085, "tokens_seen": 812646400, "tokens_per_second": 266569}
|
| 82 |
+
{"step": 6300, "max_steps": 38147, "train_loss": 2.929472, "learning_rate": 0.00019043158863438905, "grad_norm": 0.5624988079071045, "tokens_seen": 825753600, "tokens_per_second": 266565}
|
| 83 |
+
{"step": 6400, "max_steps": 38147, "train_loss": 3.024289, "learning_rate": 0.00019008939383832242, "grad_norm": 0.5274127125740051, "tokens_seen": 838860800, "tokens_per_second": 266559}
|
| 84 |
+
{"step": 6500, "max_steps": 38147, "train_loss": 2.921005, "learning_rate": 0.00018974154315608673, "grad_norm": 1.0540344715118408, "tokens_seen": 851968000, "tokens_per_second": 266569}
|
| 85 |
+
{"step": 6600, "max_steps": 38147, "train_loss": 2.875171, "learning_rate": 0.00018938806115278073, "grad_norm": 0.5187219977378845, "tokens_seen": 865075200, "tokens_per_second": 266489}
|
| 86 |
+
{"step": 6700, "max_steps": 38147, "train_loss": 2.952179, "learning_rate": 0.00018902897279118503, "grad_norm": 0.5519692897796631, "tokens_seen": 878182400, "tokens_per_second": 266547}
|
| 87 |
+
{"step": 6800, "max_steps": 38147, "train_loss": 2.921197, "learning_rate": 0.00018866430342999953, "grad_norm": 0.5296387076377869, "tokens_seen": 891289600, "tokens_per_second": 266540}
|
| 88 |
+
{"step": 6900, "max_steps": 38147, "train_loss": 2.921925, "learning_rate": 0.00018829407882205238, "grad_norm": 0.5336331129074097, "tokens_seen": 904396800, "tokens_per_second": 266548}
|
| 89 |
+
{"step": 7000, "max_steps": 38147, "train_loss": 2.952497, "learning_rate": 0.00018791832511248134, "grad_norm": 0.5777912735939026, "tokens_seen": 917504000, "tokens_per_second": 266556}
|
| 90 |
+
{"step": 7100, "max_steps": 38147, "train_loss": 2.930704, "learning_rate": 0.00018753706883688757, "grad_norm": 0.5356543660163879, "tokens_seen": 930611200, "tokens_per_second": 266559}
|
| 91 |
+
{"step": 7200, "max_steps": 38147, "train_loss": 2.931836, "learning_rate": 0.0001871503369194616, "grad_norm": 0.5299394726753235, "tokens_seen": 943718400, "tokens_per_second": 266553}
|
| 92 |
+
{"step": 7300, "max_steps": 38147, "train_loss": 3.022099, "learning_rate": 0.0001867581566710818, "grad_norm": 0.5290271639823914, "tokens_seen": 956825600, "tokens_per_second": 266542}
|
| 93 |
+
{"step": 7400, "max_steps": 38147, "train_loss": 2.938763, "learning_rate": 0.00018636055578738606, "grad_norm": 0.5280903577804565, "tokens_seen": 969932800, "tokens_per_second": 266567}
|
| 94 |
+
{"step": 7500, "max_steps": 38147, "train_loss": 2.952468, "learning_rate": 0.00018595756234681554, "grad_norm": 0.5503072738647461, "tokens_seen": 983040000, "tokens_per_second": 266533}
|
| 95 |
+
{"step": 7600, "max_steps": 38147, "train_loss": 2.917225, "learning_rate": 0.00018554920480863213, "grad_norm": 0.512189507484436, "tokens_seen": 996147200, "tokens_per_second": 266547}
|
| 96 |
+
{"step": 7700, "max_steps": 38147, "train_loss": 3.027621, "learning_rate": 0.00018513551201090837, "grad_norm": 0.4862686097621918, "tokens_seen": 1009254400, "tokens_per_second": 266559}
|
| 97 |
+
{"step": 7800, "max_steps": 38147, "train_loss": 2.924481, "learning_rate": 0.00018471651316849104, "grad_norm": 0.684536874294281, "tokens_seen": 1022361600, "tokens_per_second": 266557}
|
| 98 |
+
{"step": 7900, "max_steps": 38147, "train_loss": 2.910758, "learning_rate": 0.00018429223787093813, "grad_norm": 0.5597535371780396, "tokens_seen": 1035468800, "tokens_per_second": 266492}
|
| 99 |
+
{"step": 8000, "max_steps": 38147, "train_loss": 2.882389, "learning_rate": 0.00018386271608042903, "grad_norm": 0.5164158344268799, "tokens_seen": 1048576000, "tokens_per_second": 266536}
|
| 100 |
+
{"step": 8000, "validation_loss": 3.005551, "validation_perplexity": 20.1973}
|
| 101 |
+
{"step": 8100, "max_steps": 38147, "train_loss": 2.893677, "learning_rate": 0.00018342797812964868, "grad_norm": 0.6013792157173157, "tokens_seen": 1061683200, "tokens_per_second": 262444}
|
| 102 |
+
{"step": 8200, "max_steps": 38147, "train_loss": 2.956927, "learning_rate": 0.0001829880547196456, "grad_norm": 0.509283185005188, "tokens_seen": 1074790400, "tokens_per_second": 266533}
|
| 103 |
+
{"step": 8300, "max_steps": 38147, "train_loss": 2.899491, "learning_rate": 0.00018254297691766365, "grad_norm": 0.5559905171394348, "tokens_seen": 1087897600, "tokens_per_second": 266524}
|
| 104 |
+
{"step": 8400, "max_steps": 38147, "train_loss": 2.907307, "learning_rate": 0.0001820927761549482, "grad_norm": 0.55206698179245, "tokens_seen": 1101004800, "tokens_per_second": 266528}
|
| 105 |
+
{"step": 8500, "max_steps": 38147, "train_loss": 2.953202, "learning_rate": 0.0001816374842245262, "grad_norm": 0.5241593718528748, "tokens_seen": 1114112000, "tokens_per_second": 266526}
|
| 106 |
+
{"step": 8600, "max_steps": 38147, "train_loss": 2.900405, "learning_rate": 0.0001811771332789614, "grad_norm": 0.49315375089645386, "tokens_seen": 1127219200, "tokens_per_second": 266537}
|
| 107 |
+
{"step": 8700, "max_steps": 38147, "train_loss": 2.92854, "learning_rate": 0.0001807117558280834, "grad_norm": 0.4962051212787628, "tokens_seen": 1140326400, "tokens_per_second": 266539}
|
| 108 |
+
{"step": 8800, "max_steps": 38147, "train_loss": 2.929634, "learning_rate": 0.00018024138473669184, "grad_norm": 0.5195844173431396, "tokens_seen": 1153433600, "tokens_per_second": 266526}
|
| 109 |
+
{"step": 8900, "max_steps": 38147, "train_loss": 2.947764, "learning_rate": 0.0001797660532222357, "grad_norm": 0.5545597672462463, "tokens_seen": 1166540800, "tokens_per_second": 266539}
|
| 110 |
+
{"step": 9000, "max_steps": 38147, "train_loss": 2.882339, "learning_rate": 0.00017928579485246732, "grad_norm": 0.4967882037162781, "tokens_seen": 1179648000, "tokens_per_second": 266544}
|
| 111 |
+
{"step": 9100, "max_steps": 38147, "train_loss": 2.909407, "learning_rate": 0.00017880064354307192, "grad_norm": 0.5079694390296936, "tokens_seen": 1192755200, "tokens_per_second": 266606}
|
| 112 |
+
{"step": 9200, "max_steps": 38147, "train_loss": 2.920727, "learning_rate": 0.0001783106335552725, "grad_norm": 0.5576598048210144, "tokens_seen": 1205862400, "tokens_per_second": 266686}
|
| 113 |
+
{"step": 9300, "max_steps": 38147, "train_loss": 2.946483, "learning_rate": 0.00017781579949341023, "grad_norm": 0.5156830549240112, "tokens_seen": 1218969600, "tokens_per_second": 266692}
|
| 114 |
+
{"step": 9400, "max_steps": 38147, "train_loss": 2.839191, "learning_rate": 0.00017731617630250087, "grad_norm": 0.53850257396698, "tokens_seen": 1232076800, "tokens_per_second": 266691}
|
| 115 |
+
{"step": 9500, "max_steps": 38147, "train_loss": 2.911981, "learning_rate": 0.00017681179926576682, "grad_norm": 0.5162245035171509, "tokens_seen": 1245184000, "tokens_per_second": 266693}
|
| 116 |
+
{"step": 9600, "max_steps": 38147, "train_loss": 2.913818, "learning_rate": 0.00017630270400214551, "grad_norm": 0.5505062341690063, "tokens_seen": 1258291200, "tokens_per_second": 266613}
|
| 117 |
+
{"step": 9700, "max_steps": 38147, "train_loss": 2.902928, "learning_rate": 0.0001757889264637739, "grad_norm": 0.562888503074646, "tokens_seen": 1271398400, "tokens_per_second": 266500}
|
| 118 |
+
{"step": 9800, "max_steps": 38147, "train_loss": 2.910159, "learning_rate": 0.00017527050293344976, "grad_norm": 0.4832836985588074, "tokens_seen": 1284505600, "tokens_per_second": 266541}
|
| 119 |
+
{"step": 9900, "max_steps": 38147, "train_loss": 2.925041, "learning_rate": 0.00017474747002206906, "grad_norm": 0.5571316480636597, "tokens_seen": 1297612800, "tokens_per_second": 266555}
|
| 120 |
+
{"step": 10000, "max_steps": 38147, "train_loss": 2.910882, "learning_rate": 0.00017421986466604087, "grad_norm": 0.5821037888526917, "tokens_seen": 1310720000, "tokens_per_second": 266535}
|
| 121 |
+
{"step": 10000, "validation_loss": 2.998426, "validation_perplexity": 20.0539}
|
| 122 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00010000
|
| 123 |
+
{"step": 10100, "max_steps": 38147, "train_loss": 2.918626, "learning_rate": 0.00017368772412467876, "grad_norm": 0.5134103894233704, "tokens_seen": 1323827200, "tokens_per_second": 258816}
|
| 124 |
+
{"step": 10200, "max_steps": 38147, "train_loss": 2.949979, "learning_rate": 0.0001731510859775695, "grad_norm": 0.517809271812439, "tokens_seen": 1336934400, "tokens_per_second": 266552}
|
| 125 |
+
{"step": 10300, "max_steps": 38147, "train_loss": 2.850333, "learning_rate": 0.00017260998812191928, "grad_norm": 0.5122731328010559, "tokens_seen": 1350041600, "tokens_per_second": 266656}
|
| 126 |
+
{"step": 10400, "max_steps": 38147, "train_loss": 2.913494, "learning_rate": 0.00017206446876987747, "grad_norm": 0.5601559281349182, "tokens_seen": 1363148800, "tokens_per_second": 266653}
|
| 127 |
+
{"step": 10500, "max_steps": 38147, "train_loss": 2.944695, "learning_rate": 0.000171514566445838, "grad_norm": 0.5485866069793701, "tokens_seen": 1376256000, "tokens_per_second": 266595}
|
| 128 |
+
{"step": 10600, "max_steps": 38147, "train_loss": 2.960809, "learning_rate": 0.00017096031998371883, "grad_norm": 0.5545159578323364, "tokens_seen": 1389363200, "tokens_per_second": 266695}
|
| 129 |
+
{"step": 10700, "max_steps": 38147, "train_loss": 2.922663, "learning_rate": 0.0001704017685242195, "grad_norm": 0.5326966643333435, "tokens_seen": 1402470400, "tokens_per_second": 266706}
|
| 130 |
+
{"step": 10800, "max_steps": 38147, "train_loss": 2.803973, "learning_rate": 0.00016983895151205695, "grad_norm": 0.5638318657875061, "tokens_seen": 1415577600, "tokens_per_second": 266714}
|
| 131 |
+
{"step": 10900, "max_steps": 38147, "train_loss": 2.913323, "learning_rate": 0.00016927190869318014, "grad_norm": 0.5387423634529114, "tokens_seen": 1428684800, "tokens_per_second": 266714}
|
| 132 |
+
{"step": 11000, "max_steps": 38147, "train_loss": 2.977563, "learning_rate": 0.00016870068011196296, "grad_norm": 0.5534170269966125, "tokens_seen": 1441792000, "tokens_per_second": 266630}
|
| 133 |
+
{"step": 11100, "max_steps": 38147, "train_loss": 2.912593, "learning_rate": 0.00016812530610837656, "grad_norm": 0.5476948618888855, "tokens_seen": 1454899200, "tokens_per_second": 266525}
|
| 134 |
+
{"step": 11200, "max_steps": 38147, "train_loss": 2.911587, "learning_rate": 0.00016754582731514031, "grad_norm": 0.5375077724456787, "tokens_seen": 1468006400, "tokens_per_second": 266531}
|
| 135 |
+
{"step": 11300, "max_steps": 38147, "train_loss": 2.890766, "learning_rate": 0.00016696228465485258, "grad_norm": 0.5167133808135986, "tokens_seen": 1481113600, "tokens_per_second": 266563}
|
| 136 |
+
{"step": 11400, "max_steps": 38147, "train_loss": 2.913491, "learning_rate": 0.00016637471933710052, "grad_norm": 0.532808244228363, "tokens_seen": 1494220800, "tokens_per_second": 266545}
|
| 137 |
+
{"step": 11500, "max_steps": 38147, "train_loss": 2.982643, "learning_rate": 0.0001657831728555501, "grad_norm": 0.5288663506507874, "tokens_seen": 1507328000, "tokens_per_second": 266504}
|
| 138 |
+
{"step": 11600, "max_steps": 38147, "train_loss": 2.890127, "learning_rate": 0.00016518768698501574, "grad_norm": 0.5118499994277954, "tokens_seen": 1520435200, "tokens_per_second": 266550}
|
| 139 |
+
{"step": 11700, "max_steps": 38147, "train_loss": 2.871277, "learning_rate": 0.0001645883037785102, "grad_norm": 0.5694160461425781, "tokens_seen": 1533542400, "tokens_per_second": 266542}
|
| 140 |
+
{"step": 11800, "max_steps": 38147, "train_loss": 2.863222, "learning_rate": 0.00016398506556427472, "grad_norm": 0.547090470790863, "tokens_seen": 1546649600, "tokens_per_second": 266559}
|
| 141 |
+
{"step": 11900, "max_steps": 38147, "train_loss": 2.871399, "learning_rate": 0.00016337801494279, "grad_norm": 0.5254355072975159, "tokens_seen": 1559756800, "tokens_per_second": 266540}
|
| 142 |
+
{"step": 12000, "max_steps": 38147, "train_loss": 2.99516, "learning_rate": 0.00016276719478376756, "grad_norm": 0.5378596782684326, "tokens_seen": 1572864000, "tokens_per_second": 266543}
|
| 143 |
+
{"step": 12000, "validation_loss": 2.985874, "validation_perplexity": 19.8038}
|
| 144 |
+
{"step": 12100, "max_steps": 38147, "train_loss": 2.946746, "learning_rate": 0.00016215264822312252, "grad_norm": 0.5134831070899963, "tokens_seen": 1585971200, "tokens_per_second": 262442}
|
| 145 |
+
{"step": 12200, "max_steps": 38147, "train_loss": 2.963517, "learning_rate": 0.00016153441865992708, "grad_norm": 0.5186991095542908, "tokens_seen": 1599078400, "tokens_per_second": 266535}
|
| 146 |
+
{"step": 12300, "max_steps": 38147, "train_loss": 2.86126, "learning_rate": 0.00016091254975334607, "grad_norm": 0.5443694591522217, "tokens_seen": 1612185600, "tokens_per_second": 266543}
|
| 147 |
+
{"step": 12400, "max_steps": 38147, "train_loss": 2.944561, "learning_rate": 0.00016028708541955322, "grad_norm": 0.5636797547340393, "tokens_seen": 1625292800, "tokens_per_second": 266550}
|
| 148 |
+
{"step": 12500, "max_steps": 38147, "train_loss": 2.981808, "learning_rate": 0.00015965806982863042, "grad_norm": 0.5743669867515564, "tokens_seen": 1638400000, "tokens_per_second": 266530}
|
| 149 |
+
{"step": 12600, "max_steps": 38147, "train_loss": 2.898306, "learning_rate": 0.00015902554740144798, "grad_norm": 0.5469308495521545, "tokens_seen": 1651507200, "tokens_per_second": 266564}
|
| 150 |
+
{"step": 12700, "max_steps": 38147, "train_loss": 2.9353, "learning_rate": 0.00015838956280652788, "grad_norm": 0.5421141982078552, "tokens_seen": 1664614400, "tokens_per_second": 266482}
|
| 151 |
+
{"step": 12800, "max_steps": 38147, "train_loss": 2.948559, "learning_rate": 0.00015775016095688921, "grad_norm": 0.527752161026001, "tokens_seen": 1677721600, "tokens_per_second": 266554}
|
| 152 |
+
{"step": 12900, "max_steps": 38147, "train_loss": 2.946706, "learning_rate": 0.00015710738700687641, "grad_norm": 0.5152209401130676, "tokens_seen": 1690828800, "tokens_per_second": 266539}
|
| 153 |
+
{"step": 13000, "max_steps": 38147, "train_loss": 2.895772, "learning_rate": 0.0001564612863489706, "grad_norm": 0.5434118509292603, "tokens_seen": 1703936000, "tokens_per_second": 266556}
|
| 154 |
+
{"step": 13100, "max_steps": 38147, "train_loss": 2.92532, "learning_rate": 0.00015581190461058382, "grad_norm": 0.5467553734779358, "tokens_seen": 1717043200, "tokens_per_second": 266552}
|
| 155 |
+
{"step": 13200, "max_steps": 38147, "train_loss": 2.877013, "learning_rate": 0.00015515928765083694, "grad_norm": 0.5095875263214111, "tokens_seen": 1730150400, "tokens_per_second": 266496}
|
| 156 |
+
{"step": 13300, "max_steps": 38147, "train_loss": 2.873789, "learning_rate": 0.00015450348155732105, "grad_norm": 0.5356979966163635, "tokens_seen": 1743257600, "tokens_per_second": 266480}
|
| 157 |
+
{"step": 13400, "max_steps": 38147, "train_loss": 2.847629, "learning_rate": 0.00015384453264284288, "grad_norm": 0.5649926662445068, "tokens_seen": 1756364800, "tokens_per_second": 266547}
|
| 158 |
+
{"step": 13500, "max_steps": 38147, "train_loss": 2.885597, "learning_rate": 0.00015318248744215408, "grad_norm": 0.5296280384063721, "tokens_seen": 1769472000, "tokens_per_second": 266545}
|
| 159 |
+
{"step": 13600, "max_steps": 38147, "train_loss": 2.89902, "learning_rate": 0.00015251739270866492, "grad_norm": 0.5799275636672974, "tokens_seen": 1782579200, "tokens_per_second": 266542}
|
| 160 |
+
{"step": 13700, "max_steps": 38147, "train_loss": 2.897563, "learning_rate": 0.00015184929541114275, "grad_norm": 0.5654956102371216, "tokens_seen": 1795686400, "tokens_per_second": 266541}
|
| 161 |
+
{"step": 13800, "max_steps": 38147, "train_loss": 2.892543, "learning_rate": 0.000151178242730395, "grad_norm": 0.5284501910209656, "tokens_seen": 1808793600, "tokens_per_second": 266547}
|
| 162 |
+
{"step": 13900, "max_steps": 38147, "train_loss": 2.927978, "learning_rate": 0.0001505042820559372, "grad_norm": 0.5702105164527893, "tokens_seen": 1821900800, "tokens_per_second": 266544}
|
| 163 |
+
{"step": 14000, "max_steps": 38147, "train_loss": 2.85512, "learning_rate": 0.00014982746098264648, "grad_norm": 0.5716403126716614, "tokens_seen": 1835008000, "tokens_per_second": 266603}
|
| 164 |
+
{"step": 14000, "validation_loss": 2.978342, "validation_perplexity": 19.6552}
|
| 165 |
+
{"step": 14100, "max_steps": 38147, "train_loss": 2.837504, "learning_rate": 0.00014914782730740032, "grad_norm": 0.5596545338630676, "tokens_seen": 1848115200, "tokens_per_second": 262606}
|
| 166 |
+
{"step": 14200, "max_steps": 38147, "train_loss": 2.900359, "learning_rate": 0.00014846542902570122, "grad_norm": 0.5157343149185181, "tokens_seen": 1861222400, "tokens_per_second": 266660}
|
| 167 |
+
{"step": 14300, "max_steps": 38147, "train_loss": 2.984998, "learning_rate": 0.00014778031432828738, "grad_norm": 0.546175479888916, "tokens_seen": 1874329600, "tokens_per_second": 266636}
|
| 168 |
+
{"step": 14400, "max_steps": 38147, "train_loss": 2.799061, "learning_rate": 0.00014709253159772917, "grad_norm": 0.5169898271560669, "tokens_seen": 1887436800, "tokens_per_second": 266692}
|
| 169 |
+
{"step": 14500, "max_steps": 38147, "train_loss": 2.93394, "learning_rate": 0.0001464021294050127, "grad_norm": 0.5111414790153503, "tokens_seen": 1900544000, "tokens_per_second": 266665}
|
| 170 |
+
{"step": 14600, "max_steps": 38147, "train_loss": 2.874222, "learning_rate": 0.00014570915650610956, "grad_norm": 0.608252227306366, "tokens_seen": 1913651200, "tokens_per_second": 266542}
|
| 171 |
+
{"step": 14700, "max_steps": 38147, "train_loss": 3.022098, "learning_rate": 0.0001450136618385338, "grad_norm": 0.520990788936615, "tokens_seen": 1926758400, "tokens_per_second": 266533}
|
| 172 |
+
{"step": 14800, "max_steps": 38147, "train_loss": 2.967238, "learning_rate": 0.000144315694517886, "grad_norm": 0.5397151112556458, "tokens_seen": 1939865600, "tokens_per_second": 266544}
|
| 173 |
+
{"step": 14900, "max_steps": 38147, "train_loss": 2.903136, "learning_rate": 0.00014361530383438445, "grad_norm": 0.5363152027130127, "tokens_seen": 1952972800, "tokens_per_second": 266545}
|
| 174 |
+
{"step": 15000, "max_steps": 38147, "train_loss": 2.837465, "learning_rate": 0.00014291253924938468, "grad_norm": 0.5414167046546936, "tokens_seen": 1966080000, "tokens_per_second": 266519}
|
| 175 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00015000
|
| 176 |
+
{"step": 15100, "max_steps": 38147, "train_loss": 2.951748, "learning_rate": 0.00014220745039188626, "grad_norm": 0.756091833114624, "tokens_seen": 1979187200, "tokens_per_second": 262797}
|
| 177 |
+
{"step": 15200, "max_steps": 38147, "train_loss": 2.910635, "learning_rate": 0.00014150008705502828, "grad_norm": 0.5328271985054016, "tokens_seen": 1992294400, "tokens_per_second": 266494}
|
| 178 |
+
{"step": 15300, "max_steps": 38147, "train_loss": 2.890371, "learning_rate": 0.0001407904991925726, "grad_norm": 0.5145664215087891, "tokens_seen": 2005401600, "tokens_per_second": 266530}
|
| 179 |
+
{"step": 15400, "max_steps": 38147, "train_loss": 2.915686, "learning_rate": 0.00014007873691537637, "grad_norm": 0.5268166065216064, "tokens_seen": 2018508800, "tokens_per_second": 266554}
|
| 180 |
+
{"step": 15500, "max_steps": 38147, "train_loss": 2.906188, "learning_rate": 0.0001393648504878533, "grad_norm": 0.5345094203948975, "tokens_seen": 2031616000, "tokens_per_second": 266546}
|
| 181 |
+
{"step": 15600, "max_steps": 38147, "train_loss": 2.892306, "learning_rate": 0.00013864889032442376, "grad_norm": 0.6043722629547119, "tokens_seen": 2044723200, "tokens_per_second": 266566}
|
| 182 |
+
{"step": 15700, "max_steps": 38147, "train_loss": 2.841777, "learning_rate": 0.00013793090698595469, "grad_norm": 0.5641171336174011, "tokens_seen": 2057830400, "tokens_per_second": 266563}
|
| 183 |
+
{"step": 15800, "max_steps": 38147, "train_loss": 2.899494, "learning_rate": 0.00013721095117618902, "grad_norm": 0.5548516511917114, "tokens_seen": 2070937600, "tokens_per_second": 266700}
|
| 184 |
+
{"step": 15900, "max_steps": 38147, "train_loss": 2.935511, "learning_rate": 0.00013648907373816485, "grad_norm": 0.5173591375350952, "tokens_seen": 2084044800, "tokens_per_second": 266726}
|
| 185 |
+
{"step": 16000, "max_steps": 38147, "train_loss": 2.883337, "learning_rate": 0.00013576532565062516, "grad_norm": 0.5861512422561646, "tokens_seen": 2097152000, "tokens_per_second": 266692}
|
| 186 |
+
{"step": 16000, "validation_loss": 2.966163, "validation_perplexity": 19.4173}
|
| 187 |
+
{"step": 16100, "max_steps": 38147, "train_loss": 2.869175, "learning_rate": 0.00013503975802441742, "grad_norm": 0.5603633522987366, "tokens_seen": 2110259200, "tokens_per_second": 262614}
|
| 188 |
+
{"step": 16200, "max_steps": 38147, "train_loss": 2.987583, "learning_rate": 0.0001343124220988844, "grad_norm": 0.5374985337257385, "tokens_seen": 2123366400, "tokens_per_second": 266722}
|
| 189 |
+
{"step": 16300, "max_steps": 38147, "train_loss": 2.914009, "learning_rate": 0.00013358336923824558, "grad_norm": 0.539892315864563, "tokens_seen": 2136473600, "tokens_per_second": 266726}
|
| 190 |
+
{"step": 16400, "max_steps": 38147, "train_loss": 2.910758, "learning_rate": 0.0001328526509279698, "grad_norm": 0.5089977979660034, "tokens_seen": 2149580800, "tokens_per_second": 266615}
|
| 191 |
+
{"step": 16500, "max_steps": 38147, "train_loss": 2.88835, "learning_rate": 0.00013212031877113932, "grad_norm": 0.5609439015388489, "tokens_seen": 2162688000, "tokens_per_second": 266680}
|
| 192 |
+
{"step": 16600, "max_steps": 38147, "train_loss": 2.860655, "learning_rate": 0.00013138642448480594, "grad_norm": 0.5659797191619873, "tokens_seen": 2175795200, "tokens_per_second": 266558}
|
| 193 |
+
{"step": 16700, "max_steps": 38147, "train_loss": 2.824591, "learning_rate": 0.00013065101989633818, "grad_norm": 0.5298318862915039, "tokens_seen": 2188902400, "tokens_per_second": 266556}
|
| 194 |
+
{"step": 16800, "max_steps": 38147, "train_loss": 2.871027, "learning_rate": 0.00012991415693976193, "grad_norm": 0.5541119575500488, "tokens_seen": 2202009600, "tokens_per_second": 266524}
|
| 195 |
+
{"step": 16900, "max_steps": 38147, "train_loss": 2.762313, "learning_rate": 0.0001291758876520923, "grad_norm": 0.5406379699707031, "tokens_seen": 2215116800, "tokens_per_second": 266500}
|
| 196 |
+
{"step": 17000, "max_steps": 38147, "train_loss": 2.931732, "learning_rate": 0.0001284362641696591, "grad_norm": 0.5427020788192749, "tokens_seen": 2228224000, "tokens_per_second": 266486}
|
| 197 |
+
{"step": 17100, "max_steps": 38147, "train_loss": 2.924223, "learning_rate": 0.00012769533872442502, "grad_norm": 0.5315942764282227, "tokens_seen": 2241331200, "tokens_per_second": 266557}
|
| 198 |
+
{"step": 17200, "max_steps": 38147, "train_loss": 2.894472, "learning_rate": 0.00012695316364029683, "grad_norm": 0.5872063040733337, "tokens_seen": 2254438400, "tokens_per_second": 266562}
|
| 199 |
+
{"step": 17300, "max_steps": 38147, "train_loss": 2.938319, "learning_rate": 0.00012620979132943033, "grad_norm": 0.5693638920783997, "tokens_seen": 2267545600, "tokens_per_second": 266570}
|
| 200 |
+
{"step": 17400, "max_steps": 38147, "train_loss": 2.849607, "learning_rate": 0.00012546527428852923, "grad_norm": 0.5248034596443176, "tokens_seen": 2280652800, "tokens_per_second": 266551}
|
| 201 |
+
{"step": 17500, "max_steps": 38147, "train_loss": 2.872927, "learning_rate": 0.00012471966509513754, "grad_norm": 0.5393016338348389, "tokens_seen": 2293760000, "tokens_per_second": 266545}
|
| 202 |
+
{"step": 17600, "max_steps": 38147, "train_loss": 2.907589, "learning_rate": 0.00012397301640392684, "grad_norm": 0.5371799468994141, "tokens_seen": 2306867200, "tokens_per_second": 266487}
|
| 203 |
+
{"step": 17700, "max_steps": 38147, "train_loss": 2.930816, "learning_rate": 0.00012322538094297755, "grad_norm": 0.5750351548194885, "tokens_seen": 2319974400, "tokens_per_second": 266558}
|
| 204 |
+
{"step": 17800, "max_steps": 38147, "train_loss": 2.80574, "learning_rate": 0.00012247681151005558, "grad_norm": 0.5639591813087463, "tokens_seen": 2333081600, "tokens_per_second": 266565}
|
| 205 |
+
{"step": 17900, "max_steps": 38147, "train_loss": 2.935044, "learning_rate": 0.00012172736096888354, "grad_norm": 0.5895889401435852, "tokens_seen": 2346188800, "tokens_per_second": 266558}
|
| 206 |
+
{"step": 18000, "max_steps": 38147, "train_loss": 2.947257, "learning_rate": 0.00012097708224540767, "grad_norm": 0.4961288869380951, "tokens_seen": 2359296000, "tokens_per_second": 266572}
|
| 207 |
+
{"step": 18000, "validation_loss": 2.956802, "validation_perplexity": 19.2364}
|
| 208 |
+
{"step": 18100, "max_steps": 38147, "train_loss": 2.860126, "learning_rate": 0.0001202260283240601, "grad_norm": 0.6287165284156799, "tokens_seen": 2372403200, "tokens_per_second": 262469}
|
| 209 |
+
{"step": 18200, "max_steps": 38147, "train_loss": 2.953648, "learning_rate": 0.00011947425224401725, "grad_norm": 0.5376870036125183, "tokens_seen": 2385510400, "tokens_per_second": 266548}
|
| 210 |
+
{"step": 18300, "max_steps": 38147, "train_loss": 2.961425, "learning_rate": 0.00011872180709545405, "grad_norm": 0.5595741868019104, "tokens_seen": 2398617600, "tokens_per_second": 266556}
|
| 211 |
+
{"step": 18400, "max_steps": 38147, "train_loss": 2.897191, "learning_rate": 0.00011796874601579491, "grad_norm": 0.5086379647254944, "tokens_seen": 2411724800, "tokens_per_second": 266540}
|
| 212 |
+
{"step": 18500, "max_steps": 38147, "train_loss": 2.917819, "learning_rate": 0.00011721512218596106, "grad_norm": 0.5639253258705139, "tokens_seen": 2424832000, "tokens_per_second": 266538}
|
| 213 |
+
{"step": 18600, "max_steps": 38147, "train_loss": 2.948039, "learning_rate": 0.00011646098882661496, "grad_norm": 0.6896440386772156, "tokens_seen": 2437939200, "tokens_per_second": 266549}
|
| 214 |
+
{"step": 18700, "max_steps": 38147, "train_loss": 2.964076, "learning_rate": 0.00011570639919440187, "grad_norm": 0.5386029481887817, "tokens_seen": 2451046400, "tokens_per_second": 266527}
|
| 215 |
+
{"step": 18800, "max_steps": 38147, "train_loss": 2.866205, "learning_rate": 0.00011495140657818889, "grad_norm": 0.5471669435501099, "tokens_seen": 2464153600, "tokens_per_second": 266506}
|
| 216 |
+
{"step": 18900, "max_steps": 38147, "train_loss": 2.840063, "learning_rate": 0.00011419606429530166, "grad_norm": 0.5953474640846252, "tokens_seen": 2477260800, "tokens_per_second": 266550}
|
| 217 |
+
{"step": 19000, "max_steps": 38147, "train_loss": 2.83049, "learning_rate": 0.0001134404256877593, "grad_norm": 0.5615992546081543, "tokens_seen": 2490368000, "tokens_per_second": 266566}
|
| 218 |
+
{"step": 19100, "max_steps": 38147, "train_loss": 2.859421, "learning_rate": 0.00011268454411850713, "grad_norm": 0.5288406610488892, "tokens_seen": 2503475200, "tokens_per_second": 266566}
|
| 219 |
+
{"step": 19200, "max_steps": 38147, "train_loss": 2.9146, "learning_rate": 0.00011192847296764846, "grad_norm": 0.5557245016098022, "tokens_seen": 2516582400, "tokens_per_second": 266567}
|
| 220 |
+
{"step": 19300, "max_steps": 38147, "train_loss": 2.890018, "learning_rate": 0.00011117226562867467, "grad_norm": 0.5548304915428162, "tokens_seen": 2529689600, "tokens_per_second": 266566}
|
| 221 |
+
{"step": 19400, "max_steps": 38147, "train_loss": 2.752083, "learning_rate": 0.0001104159755046948, "grad_norm": 0.5975026488304138, "tokens_seen": 2542796800, "tokens_per_second": 266567}
|
| 222 |
+
{"step": 19500, "max_steps": 38147, "train_loss": 2.88171, "learning_rate": 0.00010965965600466413, "grad_norm": 0.5508555173873901, "tokens_seen": 2555904000, "tokens_per_second": 266556}
|
| 223 |
+
{"step": 19600, "max_steps": 38147, "train_loss": 2.922331, "learning_rate": 0.00010890336053961243, "grad_norm": 0.5566782355308533, "tokens_seen": 2569011200, "tokens_per_second": 266548}
|
| 224 |
+
{"step": 19700, "max_steps": 38147, "train_loss": 2.842486, "learning_rate": 0.0001081471425188721, "grad_norm": 0.5678550004959106, "tokens_seen": 2582118400, "tokens_per_second": 266549}
|
| 225 |
+
{"step": 19800, "max_steps": 38147, "train_loss": 2.847586, "learning_rate": 0.00010739105534630656, "grad_norm": 0.5768691897392273, "tokens_seen": 2595225600, "tokens_per_second": 266561}
|
| 226 |
+
{"step": 19900, "max_steps": 38147, "train_loss": 2.923671, "learning_rate": 0.00010663515241653866, "grad_norm": 0.5536670684814453, "tokens_seen": 2608332800, "tokens_per_second": 266573}
|
| 227 |
+
{"step": 20000, "max_steps": 38147, "train_loss": 2.835271, "learning_rate": 0.0001058794871111801, "grad_norm": 0.5471276640892029, "tokens_seen": 2621440000, "tokens_per_second": 266575}
|
| 228 |
+
{"step": 20000, "validation_loss": 2.945952, "validation_perplexity": 19.0288}
|
| 229 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00020000
|
| 230 |
+
{"step": 20100, "max_steps": 38147, "train_loss": 2.859491, "learning_rate": 0.00010512411279506171, "grad_norm": 0.5950771570205688, "tokens_seen": 2634547200, "tokens_per_second": 258912}
|
| 231 |
+
{"step": 20200, "max_steps": 38147, "train_loss": 2.92326, "learning_rate": 0.00010436908281246469, "grad_norm": 0.5458201169967651, "tokens_seen": 2647654400, "tokens_per_second": 266576}
|
| 232 |
+
{"step": 20300, "max_steps": 38147, "train_loss": 2.900287, "learning_rate": 0.0001036144504833535, "grad_norm": 0.5458999276161194, "tokens_seen": 2660761600, "tokens_per_second": 266571}
|
| 233 |
+
{"step": 20400, "max_steps": 38147, "train_loss": 2.836447, "learning_rate": 0.00010286026909961053, "grad_norm": 0.5299811363220215, "tokens_seen": 2673868800, "tokens_per_second": 266546}
|
| 234 |
+
{"step": 20500, "max_steps": 38147, "train_loss": 2.841372, "learning_rate": 0.00010210659192127246, "grad_norm": 0.5424965620040894, "tokens_seen": 2686976000, "tokens_per_second": 266510}
|
| 235 |
+
{"step": 20600, "max_steps": 38147, "train_loss": 2.737658, "learning_rate": 0.00010135347217276924, "grad_norm": 0.6293522715568542, "tokens_seen": 2700083200, "tokens_per_second": 266518}
|
| 236 |
+
{"step": 20700, "max_steps": 38147, "train_loss": 2.834829, "learning_rate": 0.00010060096303916524, "grad_norm": 0.5857658386230469, "tokens_seen": 2713190400, "tokens_per_second": 266516}
|
| 237 |
+
{"step": 20800, "max_steps": 38147, "train_loss": 2.858683, "learning_rate": 9.984911766240343e-05, "grad_norm": 0.5542163848876953, "tokens_seen": 2726297600, "tokens_per_second": 266564}
|
| 238 |
+
{"step": 20900, "max_steps": 38147, "train_loss": 2.894853, "learning_rate": 9.909798913755258e-05, "grad_norm": 0.5500002503395081, "tokens_seen": 2739404800, "tokens_per_second": 266539}
|
| 239 |
+
{"step": 21000, "max_steps": 38147, "train_loss": 2.790889, "learning_rate": 9.834763050905747e-05, "grad_norm": 0.5608624815940857, "tokens_seen": 2752512000, "tokens_per_second": 266545}
|
| 240 |
+
{"step": 21100, "max_steps": 38147, "train_loss": 2.886356, "learning_rate": 9.759809476699313e-05, "grad_norm": 0.534460186958313, "tokens_seen": 2765619200, "tokens_per_second": 266563}
|
| 241 |
+
{"step": 21200, "max_steps": 38147, "train_loss": 2.811903, "learning_rate": 9.684943484332267e-05, "grad_norm": 0.5414276123046875, "tokens_seen": 2778726400, "tokens_per_second": 266558}
|
| 242 |
+
{"step": 21300, "max_steps": 38147, "train_loss": 2.900397, "learning_rate": 9.610170360815912e-05, "grad_norm": 0.5601606965065002, "tokens_seen": 2791833600, "tokens_per_second": 266500}
|
| 243 |
+
{"step": 21400, "max_steps": 38147, "train_loss": 2.82288, "learning_rate": 9.53549538660319e-05, "grad_norm": 0.5340302586555481, "tokens_seen": 2804940800, "tokens_per_second": 266550}
|
| 244 |
+
{"step": 21500, "max_steps": 38147, "train_loss": 2.819525, "learning_rate": 9.460923835215771e-05, "grad_norm": 0.5730591416358948, "tokens_seen": 2818048000, "tokens_per_second": 266546}
|
| 245 |
+
{"step": 21600, "max_steps": 38147, "train_loss": 2.880669, "learning_rate": 9.386460972871634e-05, "grad_norm": 0.5885633230209351, "tokens_seen": 2831155200, "tokens_per_second": 266550}
|
| 246 |
+
{"step": 21700, "max_steps": 38147, "train_loss": 2.972164, "learning_rate": 9.312112058113178e-05, "grad_norm": 0.5533197522163391, "tokens_seen": 2844262400, "tokens_per_second": 266540}
|
| 247 |
+
{"step": 21800, "max_steps": 38147, "train_loss": 2.847812, "learning_rate": 9.237882341435869e-05, "grad_norm": 0.5809752345085144, "tokens_seen": 2857369600, "tokens_per_second": 266549}
|
| 248 |
+
{"step": 21900, "max_steps": 38147, "train_loss": 2.930798, "learning_rate": 9.163777064917427e-05, "grad_norm": 0.5463840961456299, "tokens_seen": 2870476800, "tokens_per_second": 266568}
|
| 249 |
+
{"step": 22000, "max_steps": 38147, "train_loss": 2.91267, "learning_rate": 9.089801461847669e-05, "grad_norm": 0.5600722432136536, "tokens_seen": 2883584000, "tokens_per_second": 266562}
|
| 250 |
+
{"step": 22000, "validation_loss": 2.9367, "validation_perplexity": 18.8535}
|
| 251 |
+
{"step": 22100, "max_steps": 38147, "train_loss": 2.810362, "learning_rate": 9.015960756358901e-05, "grad_norm": 0.5334869027137756, "tokens_seen": 2896691200, "tokens_per_second": 262474}
|
| 252 |
+
{"step": 22200, "max_steps": 38147, "train_loss": 2.799604, "learning_rate": 8.942260163057022e-05, "grad_norm": 0.5521989464759827, "tokens_seen": 2909798400, "tokens_per_second": 266574}
|
| 253 |
+
{"step": 22300, "max_steps": 38147, "train_loss": 2.824443, "learning_rate": 8.868704886653239e-05, "grad_norm": 0.5473442077636719, "tokens_seen": 2922905600, "tokens_per_second": 266711}
|
| 254 |
+
{"step": 22400, "max_steps": 38147, "train_loss": 2.797859, "learning_rate": 8.795300121596539e-05, "grad_norm": 0.6124777793884277, "tokens_seen": 2936012800, "tokens_per_second": 266708}
|
| 255 |
+
{"step": 22500, "max_steps": 38147, "train_loss": 2.932135, "learning_rate": 8.722051051706846e-05, "grad_norm": 0.5962636470794678, "tokens_seen": 2949120000, "tokens_per_second": 266657}
|
| 256 |
+
{"step": 22600, "max_steps": 38147, "train_loss": 2.877236, "learning_rate": 8.648962849808942e-05, "grad_norm": 0.5380517244338989, "tokens_seen": 2962227200, "tokens_per_second": 266692}
|
| 257 |
+
{"step": 22700, "max_steps": 38147, "train_loss": 2.882246, "learning_rate": 8.576040677367164e-05, "grad_norm": 0.5944678783416748, "tokens_seen": 2975334400, "tokens_per_second": 266715}
|
| 258 |
+
{"step": 22800, "max_steps": 38147, "train_loss": 2.793537, "learning_rate": 8.50328968412091e-05, "grad_norm": 0.543428361415863, "tokens_seen": 2988441600, "tokens_per_second": 266681}
|
| 259 |
+
{"step": 22900, "max_steps": 38147, "train_loss": 2.864169, "learning_rate": 8.430715007720955e-05, "grad_norm": 0.5235292315483093, "tokens_seen": 3001548800, "tokens_per_second": 266681}
|
| 260 |
+
{"step": 23000, "max_steps": 38147, "train_loss": 2.838596, "learning_rate": 8.35832177336664e-05, "grad_norm": 0.5795411467552185, "tokens_seen": 3014656000, "tokens_per_second": 266688}
|
| 261 |
+
{"step": 23100, "max_steps": 38147, "train_loss": 2.949519, "learning_rate": 8.28611509344393e-05, "grad_norm": 0.5032678842544556, "tokens_seen": 3027763200, "tokens_per_second": 266683}
|
| 262 |
+
{"step": 23200, "max_steps": 38147, "train_loss": 2.859625, "learning_rate": 8.214100067164374e-05, "grad_norm": 0.5826283097267151, "tokens_seen": 3040870400, "tokens_per_second": 266667}
|
| 263 |
+
{"step": 23300, "max_steps": 38147, "train_loss": 2.908341, "learning_rate": 8.142281780205014e-05, "grad_norm": 0.5404061675071716, "tokens_seen": 3053977600, "tokens_per_second": 266728}
|
| 264 |
+
{"step": 23400, "max_steps": 38147, "train_loss": 2.805495, "learning_rate": 8.070665304349216e-05, "grad_norm": 0.5462639331817627, "tokens_seen": 3067084800, "tokens_per_second": 266673}
|
| 265 |
+
{"step": 23500, "max_steps": 38147, "train_loss": 2.939718, "learning_rate": 7.99925569712852e-05, "grad_norm": 0.5242215394973755, "tokens_seen": 3080192000, "tokens_per_second": 266575}
|
| 266 |
+
{"step": 23600, "max_steps": 38147, "train_loss": 2.77057, "learning_rate": 7.928058001465473e-05, "grad_norm": 0.5677931904792786, "tokens_seen": 3093299200, "tokens_per_second": 266557}
|
| 267 |
+
{"step": 23700, "max_steps": 38147, "train_loss": 2.792751, "learning_rate": 7.857077245317494e-05, "grad_norm": 0.577501654624939, "tokens_seen": 3106406400, "tokens_per_second": 266521}
|
| 268 |
+
{"step": 23800, "max_steps": 38147, "train_loss": 2.794581, "learning_rate": 7.786318441321812e-05, "grad_norm": 0.5876113176345825, "tokens_seen": 3119513600, "tokens_per_second": 266544}
|
| 269 |
+
{"step": 23900, "max_steps": 38147, "train_loss": 2.86267, "learning_rate": 7.715786586441469e-05, "grad_norm": 0.5562711358070374, "tokens_seen": 3132620800, "tokens_per_second": 266543}
|
| 270 |
+
{"step": 24000, "max_steps": 38147, "train_loss": 2.833634, "learning_rate": 7.645486661612422e-05, "grad_norm": 0.541306734085083, "tokens_seen": 3145728000, "tokens_per_second": 266524}
|
| 271 |
+
{"step": 24000, "validation_loss": 2.927752, "validation_perplexity": 18.6856}
|
| 272 |
+
{"step": 24100, "max_steps": 38147, "train_loss": 2.857251, "learning_rate": 7.575423631391831e-05, "grad_norm": 0.5667536854743958, "tokens_seen": 3158835200, "tokens_per_second": 262408}
|
| 273 |
+
{"step": 24200, "max_steps": 38147, "train_loss": 2.838729, "learning_rate": 7.505602443607418e-05, "grad_norm": 0.5656136274337769, "tokens_seen": 3171942400, "tokens_per_second": 266462}
|
| 274 |
+
{"step": 24300, "max_steps": 38147, "train_loss": 2.884439, "learning_rate": 7.436028029008079e-05, "grad_norm": 0.592436671257019, "tokens_seen": 3185049600, "tokens_per_second": 266485}
|
| 275 |
+
{"step": 24400, "max_steps": 38147, "train_loss": 2.880978, "learning_rate": 7.366705300915675e-05, "grad_norm": 0.5604985952377319, "tokens_seen": 3198156800, "tokens_per_second": 266537}
|
| 276 |
+
{"step": 24500, "max_steps": 38147, "train_loss": 2.847855, "learning_rate": 7.297639154878039e-05, "grad_norm": 0.5508514642715454, "tokens_seen": 3211264000, "tokens_per_second": 266548}
|
| 277 |
+
{"step": 24600, "max_steps": 38147, "train_loss": 2.791588, "learning_rate": 7.22883446832328e-05, "grad_norm": 0.5729002952575684, "tokens_seen": 3224371200, "tokens_per_second": 266561}
|
| 278 |
+
{"step": 24700, "max_steps": 38147, "train_loss": 2.906638, "learning_rate": 7.160296100215316e-05, "grad_norm": 0.552596390247345, "tokens_seen": 3237478400, "tokens_per_second": 266598}
|
| 279 |
+
{"step": 24800, "max_steps": 38147, "train_loss": 2.839932, "learning_rate": 7.092028890710744e-05, "grad_norm": 0.607927680015564, "tokens_seen": 3250585600, "tokens_per_second": 266540}
|
| 280 |
+
{"step": 24900, "max_steps": 38147, "train_loss": 2.816602, "learning_rate": 7.024037660817038e-05, "grad_norm": 0.6242993474006653, "tokens_seen": 3263692800, "tokens_per_second": 266477}
|
| 281 |
+
{"step": 25000, "max_steps": 38147, "train_loss": 2.816746, "learning_rate": 6.956327212052076e-05, "grad_norm": 0.5614867210388184, "tokens_seen": 3276800000, "tokens_per_second": 266557}
|
| 282 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00025000
|
| 283 |
+
{"step": 25100, "max_steps": 38147, "train_loss": 2.766988, "learning_rate": 6.888902326105082e-05, "grad_norm": 0.5991885662078857, "tokens_seen": 3289907200, "tokens_per_second": 262356}
|
| 284 |
+
{"step": 25200, "max_steps": 38147, "train_loss": 2.854057, "learning_rate": 6.821767764498913e-05, "grad_norm": 0.558078944683075, "tokens_seen": 3303014400, "tokens_per_second": 266554}
|
| 285 |
+
{"step": 25300, "max_steps": 38147, "train_loss": 2.845538, "learning_rate": 6.754928268253827e-05, "grad_norm": 0.5611226558685303, "tokens_seen": 3316121600, "tokens_per_second": 266547}
|
| 286 |
+
{"step": 25400, "max_steps": 38147, "train_loss": 2.851045, "learning_rate": 6.688388557552657e-05, "grad_norm": 0.582211971282959, "tokens_seen": 3329228800, "tokens_per_second": 266543}
|
| 287 |
+
{"step": 25500, "max_steps": 38147, "train_loss": 2.825767, "learning_rate": 6.622153331407494e-05, "grad_norm": 0.547619104385376, "tokens_seen": 3342336000, "tokens_per_second": 266537}
|
| 288 |
+
{"step": 25600, "max_steps": 38147, "train_loss": 2.857817, "learning_rate": 6.556227267327817e-05, "grad_norm": 0.5610079765319824, "tokens_seen": 3355443200, "tokens_per_second": 266556}
|
| 289 |
+
{"step": 25700, "max_steps": 38147, "train_loss": 2.892435, "learning_rate": 6.490615020990199e-05, "grad_norm": 0.5642598867416382, "tokens_seen": 3368550400, "tokens_per_second": 266541}
|
| 290 |
+
{"step": 25800, "max_steps": 38147, "train_loss": 2.827709, "learning_rate": 6.4253212259095e-05, "grad_norm": 0.6536757349967957, "tokens_seen": 3381657600, "tokens_per_second": 266548}
|
| 291 |
+
{"step": 25900, "max_steps": 38147, "train_loss": 2.801008, "learning_rate": 6.360350493111658e-05, "grad_norm": 0.5914214849472046, "tokens_seen": 3394764800, "tokens_per_second": 266563}
|
| 292 |
+
{"step": 26000, "max_steps": 38147, "train_loss": 2.858851, "learning_rate": 6.295707410808061e-05, "grad_norm": 0.5447360277175903, "tokens_seen": 3407872000, "tokens_per_second": 266544}
|
| 293 |
+
{"step": 26000, "validation_loss": 2.918962, "validation_perplexity": 18.5221}
|
| 294 |
+
{"step": 26100, "max_steps": 38147, "train_loss": 2.800075, "learning_rate": 6.23139654407153e-05, "grad_norm": 0.5622711181640625, "tokens_seen": 3420979200, "tokens_per_second": 262408}
|
| 295 |
+
{"step": 26200, "max_steps": 38147, "train_loss": 2.85427, "learning_rate": 6.167422434513944e-05, "grad_norm": 0.5513009428977966, "tokens_seen": 3434086400, "tokens_per_second": 266544}
|
| 296 |
+
{"step": 26300, "max_steps": 38147, "train_loss": 2.849965, "learning_rate": 6.103789599965483e-05, "grad_norm": 0.6363672614097595, "tokens_seen": 3447193600, "tokens_per_second": 266545}
|
| 297 |
+
{"step": 26400, "max_steps": 38147, "train_loss": 2.879174, "learning_rate": 6.0405025341556167e-05, "grad_norm": 0.5475148558616638, "tokens_seen": 3460300800, "tokens_per_second": 266547}
|
| 298 |
+
{"step": 26500, "max_steps": 38147, "train_loss": 2.837725, "learning_rate": 5.977565706395736e-05, "grad_norm": 0.6237246990203857, "tokens_seen": 3473408000, "tokens_per_second": 266539}
|
| 299 |
+
{"step": 26600, "max_steps": 38147, "train_loss": 2.820456, "learning_rate": 5.9149835612635305e-05, "grad_norm": 0.5387876629829407, "tokens_seen": 3486515200, "tokens_per_second": 266526}
|
| 300 |
+
{"step": 26700, "max_steps": 38147, "train_loss": 2.899003, "learning_rate": 5.85276051828914e-05, "grad_norm": 0.5980178713798523, "tokens_seen": 3499622400, "tokens_per_second": 266528}
|
| 301 |
+
{"step": 26800, "max_steps": 38147, "train_loss": 2.846585, "learning_rate": 5.790900971643014e-05, "grad_norm": 0.5368542671203613, "tokens_seen": 3512729600, "tokens_per_second": 266530}
|
| 302 |
+
{"step": 26900, "max_steps": 38147, "train_loss": 2.852, "learning_rate": 5.72940928982562e-05, "grad_norm": 0.5674088001251221, "tokens_seen": 3525836800, "tokens_per_second": 266525}
|
| 303 |
+
{"step": 27000, "max_steps": 38147, "train_loss": 2.919401, "learning_rate": 5.668289815358931e-05, "grad_norm": 0.6030111312866211, "tokens_seen": 3538944000, "tokens_per_second": 266538}
|
| 304 |
+
{"step": 27100, "max_steps": 38147, "train_loss": 2.746781, "learning_rate": 5.6075468644797646e-05, "grad_norm": 0.5781011581420898, "tokens_seen": 3552051200, "tokens_per_second": 266521}
|
| 305 |
+
{"step": 27200, "max_steps": 38147, "train_loss": 2.889836, "learning_rate": 5.547184726834964e-05, "grad_norm": 0.591314971446991, "tokens_seen": 3565158400, "tokens_per_second": 266536}
|
| 306 |
+
{"step": 27300, "max_steps": 38147, "train_loss": 2.850757, "learning_rate": 5.48720766517848e-05, "grad_norm": 0.5595675706863403, "tokens_seen": 3578265600, "tokens_per_second": 266541}
|
| 307 |
+
{"step": 27400, "max_steps": 38147, "train_loss": 2.888055, "learning_rate": 5.4276199150703144e-05, "grad_norm": 0.5893188118934631, "tokens_seen": 3591372800, "tokens_per_second": 266497}
|
| 308 |
+
{"step": 27500, "max_steps": 38147, "train_loss": 2.850187, "learning_rate": 5.368425684577425e-05, "grad_norm": 0.5851203203201294, "tokens_seen": 3604480000, "tokens_per_second": 266544}
|
| 309 |
+
{"step": 27600, "max_steps": 38147, "train_loss": 2.822457, "learning_rate": 5.309629153976541e-05, "grad_norm": 0.6213855147361755, "tokens_seen": 3617587200, "tokens_per_second": 266538}
|
| 310 |
+
{"step": 27700, "max_steps": 38147, "train_loss": 2.705102, "learning_rate": 5.2512344754589603e-05, "grad_norm": 0.6402599811553955, "tokens_seen": 3630694400, "tokens_per_second": 266532}
|
| 311 |
+
{"step": 27800, "max_steps": 38147, "train_loss": 2.895195, "learning_rate": 5.1932457728373274e-05, "grad_norm": 0.527906596660614, "tokens_seen": 3643801600, "tokens_per_second": 266524}
|
| 312 |
+
{"step": 27900, "max_steps": 38147, "train_loss": 2.941834, "learning_rate": 5.1356671412543944e-05, "grad_norm": 0.5629961490631104, "tokens_seen": 3656908800, "tokens_per_second": 266492}
|
| 313 |
+
{"step": 28000, "max_steps": 38147, "train_loss": 2.893293, "learning_rate": 5.078502646893841e-05, "grad_norm": 0.5659019351005554, "tokens_seen": 3670016000, "tokens_per_second": 266479}
|
| 314 |
+
{"step": 28000, "validation_loss": 2.911568, "validation_perplexity": 18.3856}
|
| 315 |
+
{"step": 28100, "max_steps": 38147, "train_loss": 2.847016, "learning_rate": 5.021756326693115e-05, "grad_norm": 0.5569750666618347, "tokens_seen": 3683123200, "tokens_per_second": 262443}
|
| 316 |
+
{"step": 28200, "max_steps": 38147, "train_loss": 2.907253, "learning_rate": 4.9654321880583374e-05, "grad_norm": 0.5603170990943909, "tokens_seen": 3696230400, "tokens_per_second": 266532}
|
| 317 |
+
{"step": 28300, "max_steps": 38147, "train_loss": 2.79605, "learning_rate": 4.9095342085813225e-05, "grad_norm": 0.5688501596450806, "tokens_seen": 3709337600, "tokens_per_second": 266530}
|
| 318 |
+
{"step": 28400, "max_steps": 38147, "train_loss": 2.860015, "learning_rate": 4.854066335758653e-05, "grad_norm": 0.5661910176277161, "tokens_seen": 3722444800, "tokens_per_second": 266541}
|
| 319 |
+
{"step": 28500, "max_steps": 38147, "train_loss": 2.832889, "learning_rate": 4.799032486712935e-05, "grad_norm": 0.5644569993019104, "tokens_seen": 3735552000, "tokens_per_second": 266554}
|
| 320 |
+
{"step": 28600, "max_steps": 38147, "train_loss": 2.868754, "learning_rate": 4.7444365479161524e-05, "grad_norm": 0.582551121711731, "tokens_seen": 3748659200, "tokens_per_second": 266505}
|
| 321 |
+
{"step": 28700, "max_steps": 38147, "train_loss": 2.815263, "learning_rate": 4.6902823749152175e-05, "grad_norm": 0.551891028881073, "tokens_seen": 3761766400, "tokens_per_second": 266554}
|
| 322 |
+
{"step": 28800, "max_steps": 38147, "train_loss": 2.796259, "learning_rate": 4.636573792059689e-05, "grad_norm": 0.6193068027496338, "tokens_seen": 3774873600, "tokens_per_second": 266633}
|
| 323 |
+
{"step": 28900, "max_steps": 38147, "train_loss": 2.900352, "learning_rate": 4.5833145922317056e-05, "grad_norm": 0.5879529714584351, "tokens_seen": 3787980800, "tokens_per_second": 266547}
|
| 324 |
+
{"step": 29000, "max_steps": 38147, "train_loss": 2.837007, "learning_rate": 4.530508536578117e-05, "grad_norm": 0.5981994867324829, "tokens_seen": 3801088000, "tokens_per_second": 266560}
|
| 325 |
+
{"step": 29100, "max_steps": 38147, "train_loss": 2.829343, "learning_rate": 4.4781593542448884e-05, "grad_norm": 0.6040003299713135, "tokens_seen": 3814195200, "tokens_per_second": 266558}
|
| 326 |
+
{"step": 29200, "max_steps": 38147, "train_loss": 2.882287, "learning_rate": 4.426270742113743e-05, "grad_norm": 0.5958173274993896, "tokens_seen": 3827302400, "tokens_per_second": 266547}
|
| 327 |
+
{"step": 29300, "max_steps": 38147, "train_loss": 2.780126, "learning_rate": 4.374846364541088e-05, "grad_norm": 0.5621843934059143, "tokens_seen": 3840409600, "tokens_per_second": 266548}
|
| 328 |
+
{"step": 29400, "max_steps": 38147, "train_loss": 2.885348, "learning_rate": 4.32388985309925e-05, "grad_norm": 0.5459842681884766, "tokens_seen": 3853516800, "tokens_per_second": 266558}
|
| 329 |
+
{"step": 29500, "max_steps": 38147, "train_loss": 2.919526, "learning_rate": 4.2734048063199994e-05, "grad_norm": 0.5731099843978882, "tokens_seen": 3866624000, "tokens_per_second": 266565}
|
| 330 |
+
{"step": 29600, "max_steps": 38147, "train_loss": 2.77706, "learning_rate": 4.223394789440425e-05, "grad_norm": 0.624010443687439, "tokens_seen": 3879731200, "tokens_per_second": 266553}
|
| 331 |
+
{"step": 29700, "max_steps": 38147, "train_loss": 2.850321, "learning_rate": 4.173863334151176e-05, "grad_norm": 0.6066707372665405, "tokens_seen": 3892838400, "tokens_per_second": 266535}
|
| 332 |
+
{"step": 29800, "max_steps": 38147, "train_loss": 2.763236, "learning_rate": 4.1248139383470296e-05, "grad_norm": 0.5490103363990784, "tokens_seen": 3905945600, "tokens_per_second": 266557}
|
| 333 |
+
{"step": 29900, "max_steps": 38147, "train_loss": 2.751048, "learning_rate": 4.076250065879897e-05, "grad_norm": 0.5612136721611023, "tokens_seen": 3919052800, "tokens_per_second": 266558}
|
| 334 |
+
{"step": 30000, "max_steps": 38147, "train_loss": 2.759008, "learning_rate": 4.028175146314184e-05, "grad_norm": 0.5331870913505554, "tokens_seen": 3932160000, "tokens_per_second": 266549}
|
| 335 |
+
{"step": 30000, "validation_loss": 2.904722, "validation_perplexity": 18.2602}
|
| 336 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00030000
|
| 337 |
+
{"step": 30100, "max_steps": 38147, "train_loss": 2.803232, "learning_rate": 3.980592574684605e-05, "grad_norm": 0.5768374800682068, "tokens_seen": 3945267200, "tokens_per_second": 258866}
|
| 338 |
+
{"step": 30200, "max_steps": 38147, "train_loss": 2.840143, "learning_rate": 3.933505711256437e-05, "grad_norm": 0.5935572981834412, "tokens_seen": 3958374400, "tokens_per_second": 266555}
|
| 339 |
+
{"step": 30300, "max_steps": 38147, "train_loss": 2.880723, "learning_rate": 3.886917881288202e-05, "grad_norm": 0.5887243151664734, "tokens_seen": 3971481600, "tokens_per_second": 266547}
|
| 340 |
+
{"step": 30400, "max_steps": 38147, "train_loss": 2.895415, "learning_rate": 3.8408323747968466e-05, "grad_norm": 0.5813783407211304, "tokens_seen": 3984588800, "tokens_per_second": 266539}
|
| 341 |
+
{"step": 30500, "max_steps": 38147, "train_loss": 2.812178, "learning_rate": 3.79525244632541e-05, "grad_norm": 0.5698392987251282, "tokens_seen": 3997696000, "tokens_per_second": 266540}
|
| 342 |
+
{"step": 30600, "max_steps": 38147, "train_loss": 2.800062, "learning_rate": 3.7501813147131727e-05, "grad_norm": 0.5620970129966736, "tokens_seen": 4010803200, "tokens_per_second": 266523}
|
| 343 |
+
{"step": 30700, "max_steps": 38147, "train_loss": 2.834445, "learning_rate": 3.705622162868352e-05, "grad_norm": 0.6195650100708008, "tokens_seen": 4023910400, "tokens_per_second": 266534}
|
| 344 |
+
{"step": 30800, "max_steps": 38147, "train_loss": 2.806707, "learning_rate": 3.661578137543329e-05, "grad_norm": 0.5828126668930054, "tokens_seen": 4037017600, "tokens_per_second": 266542}
|
| 345 |
+
{"step": 30900, "max_steps": 38147, "train_loss": 2.786363, "learning_rate": 3.618052349112416e-05, "grad_norm": 0.580255925655365, "tokens_seen": 4050124800, "tokens_per_second": 266541}
|
| 346 |
+
{"step": 31000, "max_steps": 38147, "train_loss": 2.825516, "learning_rate": 3.575047871352223e-05, "grad_norm": 0.5789481401443481, "tokens_seen": 4063232000, "tokens_per_second": 266520}
|
| 347 |
+
{"step": 31100, "max_steps": 38147, "train_loss": 2.801404, "learning_rate": 3.5325677412245626e-05, "grad_norm": 0.5894032120704651, "tokens_seen": 4076339200, "tokens_per_second": 266567}
|
| 348 |
+
{"step": 31200, "max_steps": 38147, "train_loss": 2.862363, "learning_rate": 3.490614958661992e-05, "grad_norm": 0.5975892543792725, "tokens_seen": 4089446400, "tokens_per_second": 266553}
|
| 349 |
+
{"step": 31300, "max_steps": 38147, "train_loss": 2.895666, "learning_rate": 3.44919248635597e-05, "grad_norm": 0.5712851881980896, "tokens_seen": 4102553600, "tokens_per_second": 266547}
|
| 350 |
+
{"step": 31400, "max_steps": 38147, "train_loss": 2.820938, "learning_rate": 3.4083032495476144e-05, "grad_norm": 0.6001507639884949, "tokens_seen": 4115660800, "tokens_per_second": 266536}
|
| 351 |
+
{"step": 31500, "max_steps": 38147, "train_loss": 2.859482, "learning_rate": 3.367950135821136e-05, "grad_norm": 0.5555902123451233, "tokens_seen": 4128768000, "tokens_per_second": 266481}
|
| 352 |
+
{"step": 31600, "max_steps": 38147, "train_loss": 2.824783, "learning_rate": 3.3281359948999143e-05, "grad_norm": 0.5704244375228882, "tokens_seen": 4141875200, "tokens_per_second": 266490}
|
| 353 |
+
{"step": 31700, "max_steps": 38147, "train_loss": 2.802811, "learning_rate": 3.288863638445238e-05, "grad_norm": 0.6088986396789551, "tokens_seen": 4154982400, "tokens_per_second": 266550}
|
| 354 |
+
{"step": 31800, "max_steps": 38147, "train_loss": 2.787406, "learning_rate": 3.250135839857772e-05, "grad_norm": 0.5637134313583374, "tokens_seen": 4168089600, "tokens_per_second": 266539}
|
| 355 |
+
{"step": 31900, "max_steps": 38147, "train_loss": 2.863412, "learning_rate": 3.2119553340816827e-05, "grad_norm": 0.6168907880783081, "tokens_seen": 4181196800, "tokens_per_second": 266545}
|
| 356 |
+
{"step": 32000, "max_steps": 38147, "train_loss": 2.813874, "learning_rate": 3.174324817411495e-05, "grad_norm": 0.5987451076507568, "tokens_seen": 4194304000, "tokens_per_second": 266549}
|
| 357 |
+
{"step": 32000, "validation_loss": 2.900096, "validation_perplexity": 18.1759}
|
| 358 |
+
{"step": 32100, "max_steps": 38147, "train_loss": 2.755264, "learning_rate": 3.137246947301701e-05, "grad_norm": 0.5844534039497375, "tokens_seen": 4207411200, "tokens_per_second": 262483}
|
| 359 |
+
{"step": 32200, "max_steps": 38147, "train_loss": 2.991299, "learning_rate": 3.1007243421790615e-05, "grad_norm": 0.6205332279205322, "tokens_seen": 4220518400, "tokens_per_second": 266503}
|
| 360 |
+
{"step": 32300, "max_steps": 38147, "train_loss": 2.765144, "learning_rate": 3.064759581257723e-05, "grad_norm": 0.5922822952270508, "tokens_seen": 4233625600, "tokens_per_second": 266537}
|
| 361 |
+
{"step": 32400, "max_steps": 38147, "train_loss": 2.833967, "learning_rate": 3.0293552043570544e-05, "grad_norm": 0.5997751355171204, "tokens_seen": 4246732800, "tokens_per_second": 266530}
|
| 362 |
+
{"step": 32500, "max_steps": 38147, "train_loss": 2.781722, "learning_rate": 2.9945137117222933e-05, "grad_norm": 0.5541488528251648, "tokens_seen": 4259840000, "tokens_per_second": 266520}
|
| 363 |
+
{"step": 32600, "max_steps": 38147, "train_loss": 2.785665, "learning_rate": 2.960237563847986e-05, "grad_norm": 0.5725107192993164, "tokens_seen": 4272947200, "tokens_per_second": 266539}
|
| 364 |
+
{"step": 32700, "max_steps": 38147, "train_loss": 2.838851, "learning_rate": 2.926529181304208e-05, "grad_norm": 0.5699809789657593, "tokens_seen": 4286054400, "tokens_per_second": 266551}
|
| 365 |
+
{"step": 32800, "max_steps": 38147, "train_loss": 2.860283, "learning_rate": 2.8933909445656494e-05, "grad_norm": 0.5856370329856873, "tokens_seen": 4299161600, "tokens_per_second": 266549}
|
| 366 |
+
{"step": 32900, "max_steps": 38147, "train_loss": 2.827816, "learning_rate": 2.8608251938434895e-05, "grad_norm": 0.5361691117286682, "tokens_seen": 4312268800, "tokens_per_second": 266549}
|
| 367 |
+
{"step": 33000, "max_steps": 38147, "train_loss": 2.884901, "learning_rate": 2.8288342289201348e-05, "grad_norm": 0.5725577473640442, "tokens_seen": 4325376000, "tokens_per_second": 266561}
|
| 368 |
+
{"step": 33100, "max_steps": 38147, "train_loss": 2.89825, "learning_rate": 2.797420308986819e-05, "grad_norm": 0.5881522297859192, "tokens_seen": 4338483200, "tokens_per_second": 266553}
|
| 369 |
+
{"step": 33200, "max_steps": 38147, "train_loss": 2.886364, "learning_rate": 2.76658565248404e-05, "grad_norm": 0.5617952942848206, "tokens_seen": 4351590400, "tokens_per_second": 266550}
|
| 370 |
+
{"step": 33300, "max_steps": 38147, "train_loss": 2.882295, "learning_rate": 2.736332436944915e-05, "grad_norm": 0.5786371231079102, "tokens_seen": 4364697600, "tokens_per_second": 266558}
|
| 371 |
+
{"step": 33400, "max_steps": 38147, "train_loss": 2.819813, "learning_rate": 2.706662798841395e-05, "grad_norm": 0.5660547018051147, "tokens_seen": 4377804800, "tokens_per_second": 266530}
|
| 372 |
+
{"step": 33500, "max_steps": 38147, "train_loss": 2.843132, "learning_rate": 2.6775788334333824e-05, "grad_norm": 0.5831009745597839, "tokens_seen": 4390912000, "tokens_per_second": 266481}
|
| 373 |
+
{"step": 33600, "max_steps": 38147, "train_loss": 2.822223, "learning_rate": 2.6490825946207792e-05, "grad_norm": 0.7210411429405212, "tokens_seen": 4404019200, "tokens_per_second": 266544}
|
| 374 |
+
{"step": 33700, "max_steps": 38147, "train_loss": 2.735867, "learning_rate": 2.621176094798421e-05, "grad_norm": 0.5961363911628723, "tokens_seen": 4417126400, "tokens_per_second": 266545}
|
| 375 |
+
{"step": 33800, "max_steps": 38147, "train_loss": 2.817519, "learning_rate": 2.5938613047139777e-05, "grad_norm": 0.6047573089599609, "tokens_seen": 4430233600, "tokens_per_second": 266523}
|
| 376 |
+
{"step": 33900, "max_steps": 38147, "train_loss": 2.863799, "learning_rate": 2.5671401533287816e-05, "grad_norm": 0.5683556199073792, "tokens_seen": 4443340800, "tokens_per_second": 266624}
|
| 377 |
+
{"step": 34000, "max_steps": 38147, "train_loss": 2.811208, "learning_rate": 2.5410145276815912e-05, "grad_norm": 0.603577733039856, "tokens_seen": 4456448000, "tokens_per_second": 266652}
|
| 378 |
+
{"step": 34000, "validation_loss": 2.895255, "validation_perplexity": 18.0881}
|
| 379 |
+
{"step": 34100, "max_steps": 38147, "train_loss": 2.855515, "learning_rate": 2.515486272755338e-05, "grad_norm": 0.6094146966934204, "tokens_seen": 4469555200, "tokens_per_second": 262551}
|
| 380 |
+
{"step": 34200, "max_steps": 38147, "train_loss": 2.733138, "learning_rate": 2.4905571913468396e-05, "grad_norm": 0.6777827143669128, "tokens_seen": 4482662400, "tokens_per_second": 266673}
|
| 381 |
+
{"step": 34300, "max_steps": 38147, "train_loss": 2.882184, "learning_rate": 2.4662290439394705e-05, "grad_norm": 0.5954318642616272, "tokens_seen": 4495769600, "tokens_per_second": 266662}
|
| 382 |
+
{"step": 34400, "max_steps": 38147, "train_loss": 2.799615, "learning_rate": 2.442503548578856e-05, "grad_norm": 0.5584408044815063, "tokens_seen": 4508876800, "tokens_per_second": 266648}
|
| 383 |
+
{"step": 34500, "max_steps": 38147, "train_loss": 2.807887, "learning_rate": 2.4193823807515308e-05, "grad_norm": 0.5776689648628235, "tokens_seen": 4521984000, "tokens_per_second": 266601}
|
| 384 |
+
{"step": 34600, "max_steps": 38147, "train_loss": 2.83416, "learning_rate": 2.3968671732666227e-05, "grad_norm": 0.5929839015007019, "tokens_seen": 4535091200, "tokens_per_second": 266524}
|
| 385 |
+
{"step": 34700, "max_steps": 38147, "train_loss": 2.907646, "learning_rate": 2.3749595161405472e-05, "grad_norm": 0.6023045182228088, "tokens_seen": 4548198400, "tokens_per_second": 266493}
|
| 386 |
+
{"step": 34800, "max_steps": 38147, "train_loss": 2.870698, "learning_rate": 2.3536609564847074e-05, "grad_norm": 0.6041939854621887, "tokens_seen": 4561305600, "tokens_per_second": 266526}
|
| 387 |
+
{"step": 34900, "max_steps": 38147, "train_loss": 2.772315, "learning_rate": 2.3329729983962546e-05, "grad_norm": 0.5803220868110657, "tokens_seen": 4574412800, "tokens_per_second": 266535}
|
| 388 |
+
{"step": 35000, "max_steps": 38147, "train_loss": 2.879679, "learning_rate": 2.3128971028518633e-05, "grad_norm": 0.5865634083747864, "tokens_seen": 4587520000, "tokens_per_second": 266519}
|
| 389 |
+
Saved checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00035000
|
| 390 |
+
{"step": 35100, "max_steps": 38147, "train_loss": 2.790793, "learning_rate": 2.2934346876045487e-05, "grad_norm": 0.5853574275970459, "tokens_seen": 4600627200, "tokens_per_second": 262582}
|
| 391 |
+
{"step": 35200, "max_steps": 38147, "train_loss": 2.833371, "learning_rate": 2.274587127083561e-05, "grad_norm": 0.58817058801651, "tokens_seen": 4613734400, "tokens_per_second": 266453}
|
| 392 |
+
{"step": 35300, "max_steps": 38147, "train_loss": 2.877543, "learning_rate": 2.2563557522973084e-05, "grad_norm": 0.6056636571884155, "tokens_seen": 4626841600, "tokens_per_second": 266469}
|
| 393 |
+
{"step": 35400, "max_steps": 38147, "train_loss": 2.813346, "learning_rate": 2.2387418507393733e-05, "grad_norm": 0.6120516061782837, "tokens_seen": 4639948800, "tokens_per_second": 266530}
|
| 394 |
+
{"step": 35500, "max_steps": 38147, "train_loss": 2.86982, "learning_rate": 2.221746666297585e-05, "grad_norm": 0.5743942856788635, "tokens_seen": 4653056000, "tokens_per_second": 266525}
|
| 395 |
+
{"step": 35600, "max_steps": 38147, "train_loss": 2.757251, "learning_rate": 2.2053713991661767e-05, "grad_norm": 0.6028977036476135, "tokens_seen": 4666163200, "tokens_per_second": 266537}
|
| 396 |
+
{"step": 35700, "max_steps": 38147, "train_loss": 2.807111, "learning_rate": 2.189617205761028e-05, "grad_norm": 0.6074195504188538, "tokens_seen": 4679270400, "tokens_per_second": 266539}
|
| 397 |
+
{"step": 35800, "max_steps": 38147, "train_loss": 2.824218, "learning_rate": 2.1744851986380013e-05, "grad_norm": 0.5873327255249023, "tokens_seen": 4692377600, "tokens_per_second": 266557}
|
| 398 |
+
{"step": 35900, "max_steps": 38147, "train_loss": 2.88764, "learning_rate": 2.1599764464143705e-05, "grad_norm": 0.5763123035430908, "tokens_seen": 4705484800, "tokens_per_second": 266500}
|
| 399 |
+
{"step": 36000, "max_steps": 38147, "train_loss": 2.789732, "learning_rate": 2.1460919736933597e-05, "grad_norm": 0.5788039565086365, "tokens_seen": 4718592000, "tokens_per_second": 266569}
|
| 400 |
+
{"step": 36000, "validation_loss": 2.891825, "validation_perplexity": 18.0262}
|
| 401 |
+
{"step": 36100, "max_steps": 38147, "train_loss": 2.732829, "learning_rate": 2.132832760991782e-05, "grad_norm": 0.5719741582870483, "tokens_seen": 4731699200, "tokens_per_second": 262502}
|
| 402 |
+
{"step": 36200, "max_steps": 38147, "train_loss": 2.788649, "learning_rate": 2.1201997446707975e-05, "grad_norm": 0.5978461503982544, "tokens_seen": 4744806400, "tokens_per_second": 266706}
|
| 403 |
+
{"step": 36300, "max_steps": 38147, "train_loss": 2.888856, "learning_rate": 2.1081938168697866e-05, "grad_norm": 0.6924115419387817, "tokens_seen": 4757913600, "tokens_per_second": 266715}
|
| 404 |
+
{"step": 36400, "max_steps": 38147, "train_loss": 2.856264, "learning_rate": 2.0968158254433507e-05, "grad_norm": 0.5733427405357361, "tokens_seen": 4771020800, "tokens_per_second": 266712}
|
| 405 |
+
{"step": 36500, "max_steps": 38147, "train_loss": 2.848382, "learning_rate": 2.0860665739014325e-05, "grad_norm": 0.5853831171989441, "tokens_seen": 4784128000, "tokens_per_second": 266707}
|
| 406 |
+
{"step": 36600, "max_steps": 38147, "train_loss": 2.822363, "learning_rate": 2.0759468213525767e-05, "grad_norm": 0.5953521728515625, "tokens_seen": 4797235200, "tokens_per_second": 266585}
|
| 407 |
+
{"step": 36700, "max_steps": 38147, "train_loss": 2.803841, "learning_rate": 2.0664572824503198e-05, "grad_norm": 0.6017800569534302, "tokens_seen": 4810342400, "tokens_per_second": 266526}
|
| 408 |
+
{"step": 36800, "max_steps": 38147, "train_loss": 2.834847, "learning_rate": 2.0575986273427188e-05, "grad_norm": 0.5908879637718201, "tokens_seen": 4823449600, "tokens_per_second": 266539}
|
| 409 |
+
{"step": 36900, "max_steps": 38147, "train_loss": 2.847193, "learning_rate": 2.0493714816250315e-05, "grad_norm": 0.6009106636047363, "tokens_seen": 4836556800, "tokens_per_second": 266543}
|
| 410 |
+
{"step": 37000, "max_steps": 38147, "train_loss": 2.900528, "learning_rate": 2.041776426295533e-05, "grad_norm": 0.5982612371444702, "tokens_seen": 4849664000, "tokens_per_second": 266558}
|
| 411 |
+
{"step": 37100, "max_steps": 38147, "train_loss": 2.825163, "learning_rate": 2.0348139977144858e-05, "grad_norm": 0.6053802967071533, "tokens_seen": 4862771200, "tokens_per_second": 266495}
|
| 412 |
+
{"step": 37200, "max_steps": 38147, "train_loss": 2.835938, "learning_rate": 2.028484687566266e-05, "grad_norm": 0.5705294013023376, "tokens_seen": 4875878400, "tokens_per_second": 266549}
|
| 413 |
+
{"step": 37300, "max_steps": 38147, "train_loss": 2.841648, "learning_rate": 2.0227889428246363e-05, "grad_norm": 0.6156661510467529, "tokens_seen": 4888985600, "tokens_per_second": 266548}
|
| 414 |
+
{"step": 37400, "max_steps": 38147, "train_loss": 2.818843, "learning_rate": 2.0177271657211818e-05, "grad_norm": 0.6606355309486389, "tokens_seen": 4902092800, "tokens_per_second": 266541}
|
| 415 |
+
{"step": 37500, "max_steps": 38147, "train_loss": 2.900203, "learning_rate": 2.013299713716906e-05, "grad_norm": 0.5641033053398132, "tokens_seen": 4915200000, "tokens_per_second": 266538}
|
| 416 |
+
{"step": 37600, "max_steps": 38147, "train_loss": 2.782548, "learning_rate": 2.0095068994769872e-05, "grad_norm": 0.5759052634239197, "tokens_seen": 4928307200, "tokens_per_second": 266532}
|
| 417 |
+
{"step": 37700, "max_steps": 38147, "train_loss": 2.782311, "learning_rate": 2.0063489908486975e-05, "grad_norm": 0.5701453685760498, "tokens_seen": 4941414400, "tokens_per_second": 266557}
|
| 418 |
+
{"step": 37800, "max_steps": 38147, "train_loss": 2.820711, "learning_rate": 2.0038262108424856e-05, "grad_norm": 0.588422954082489, "tokens_seen": 4954521600, "tokens_per_second": 266547}
|
| 419 |
+
{"step": 37900, "max_steps": 38147, "train_loss": 2.882058, "learning_rate": 2.0019387376162305e-05, "grad_norm": 0.5679551362991333, "tokens_seen": 4967628800, "tokens_per_second": 266540}
|
| 420 |
+
{"step": 38000, "max_steps": 38147, "train_loss": 2.804259, "learning_rate": 2.0006867044626598e-05, "grad_norm": 0.5903794169425964, "tokens_seen": 4980736000, "tokens_per_second": 266543}
|
| 421 |
+
{"step": 38000, "validation_loss": 2.889988, "validation_perplexity": 17.9931}
|
| 422 |
+
{"step": 38100, "max_steps": 38147, "train_loss": 2.808716, "learning_rate": 2.000070199799935e-05, "grad_norm": 0.5782372951507568, "tokens_seen": 4993843200, "tokens_per_second": 262438}
|
| 423 |
+
Final checkpoint: runs/kawk50m-fixed-recovery-6b-5090/checkpoint-00038147
|
reports/kawk50m-ko-instruct-v1/baseline-evaluation.json
ADDED
|
@@ -0,0 +1,56 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"checkpoint": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415",
|
| 3 |
+
"tokenizer": "/workspace/KAWK50M/artifacts/tokenizer-20k-v3",
|
| 4 |
+
"validation_loss": 2.54827207645266,
|
| 5 |
+
"validation_perplexity": 12.784993194667898,
|
| 6 |
+
"evaluated_supervised_tokens": 341104,
|
| 7 |
+
"assistant_only_loss": true,
|
| 8 |
+
"samples": [
|
| 9 |
+
{
|
| 10 |
+
"prompt": "대한민국의 수도는 어디인가요? 한 문장으로 답해주세요.",
|
| 11 |
+
"response": "한국에서 가장 오래된 수도 중 하나인 서울은 1950년대에 지어진 최초의 수도로, 현재는 300년이 넘은 역사를 가지고 있습니다. 이 수도는 1872년에 처음 지어졌으며, 1964년에는 1960년대에 다시 지어졌습니다. 서울은 1970년대에 고덕동과 고덕동을 연결하는 도로가 개통되면서 번성했습니다. 또한, 1980년대에는 고덕동과 고덕동을 연결하는 도로가 개통되어 서울의 중심지 역할을 했습니다. 서울은 고덕동과 고덕동을 연결하는 도로를 건설하여 서울의 중심지 역할을 수행했습니다.",
|
| 12 |
+
"metrics": {
|
| 13 |
+
"characters": 270,
|
| 14 |
+
"hangul_visible_ratio": 0.8379629629629629,
|
| 15 |
+
"unique_word_ratio": 0.7818181818181819
|
| 16 |
+
}
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"prompt": "비가 오는 날 외출할 때 준비할 물건 세 가지를 알려주세요.",
|
| 20 |
+
"response": "비오는 날 외출 시 준비해야 할 물품은 무엇인가요?\n### 1. 옷장:\n옷을 보관하는 장소는 어떤 것이 있나요?\n옷을 보관하기 위해 필요한 물품은 무엇인가요?\n### 2. 가방:\n가방을 보관할 수 있는 아이템은 무엇인가요?\n### 3. 지갑:\n가방을 보관할 수 있는 아이템은 무엇인가요?\n### 4. 지갑:\n가방을 보관할 수 있는 아이템은 무엇인가요?\n### 5. 지갑:\n가방을 보관할 수 있는 아이템은 무엇인가요?\n### 6. 지갑:\n가방을 보관할 수 있는 아이템은 무엇인가요?\n### 7. 지갑:\n가",
|
| 21 |
+
"metrics": {
|
| 22 |
+
"characters": 281,
|
| 23 |
+
"hangul_visible_ratio": 0.7619047619047619,
|
| 24 |
+
"unique_word_ratio": 0.4722222222222222
|
| 25 |
+
}
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"prompt": "다음 문장을 자연스럽게 고쳐주세요: 나는 어제 학교에 갑니다.",
|
| 29 |
+
"response": "이 문장은 1980년대의 팝 음악과 2000년대의 록 음악 사이에 존재하는 전형적인 멜로디를 사용합니다.\n그것은 팝 음악의 멜로디와 록 음악의 멜로디가 서로 다른 방식으로 결합되어 있습니다.\n### 도우미:\n이 문장은 팝 음악의 멜로디와 록 음악의 멜로디가 서로 다른 방식으로 결합되어 있습니다.\n그것은 팝 음악의 멜로디와 록 음악의 멜로디가 서로 다른 방식으로 결합되어 있습니다.\n### 도우미:\n이 문장은 팝 음악의 멜로디와 록 음악의 멜로디",
|
| 30 |
+
"metrics": {
|
| 31 |
+
"characters": 249,
|
| 32 |
+
"hangul_visible_ratio": 0.8936170212765957,
|
| 33 |
+
"unique_word_ratio": 0.4032258064516129
|
| 34 |
+
}
|
| 35 |
+
},
|
| 36 |
+
{
|
| 37 |
+
"prompt": "한국어로 자기소개를 두 문장으로 해주세요.",
|
| 38 |
+
"response": "안녕하세요, 저는 한국에 있는 외국인으로서 한국어로 자기소개를 하고 싶습니다.\n이름:\n한국어로 자기소개를 하려면 어떻게 해야 할까요?\n도우미:\n저는 한국어로 자기소개를 하기 위해 다음과 같은 방법을 사용할 수 있습니다.\n1. 인터넷을 통해 자신의 언어를 검색합니다.\n2. 한국어로 자기소개를 하려면 어떻게 해야 할까요?\n3. 인터넷에서 자신의 언어를 검색하고, 자신의 언어를 다른 언어로 바꾸거나 변경할 수 있습니다.\n4. 인터넷에서 자신의 언어를 변경하거나 변경할 수 있습니다.\n5. 자신의 언어를 다른 언어로 바꾸거나 변경할 수 있습니다.\n6. 자신의 언어를 다른 언어로 바꿀 수도 있습니다.\n7. 자신의 언어를",
|
| 39 |
+
"metrics": {
|
| 40 |
+
"characters": 344,
|
| 41 |
+
"hangul_visible_ratio": 0.8973384030418251,
|
| 42 |
+
"unique_word_ratio": 0.5365853658536586
|
| 43 |
+
}
|
| 44 |
+
},
|
| 45 |
+
{
|
| 46 |
+
"prompt": "사용자가 위험한 요청을 하면 어떻게 대응해야 하나요?",
|
| 47 |
+
"response": "이메일이나 문자메시지로 사용자의 요청을 처리하는 것은 매우 어렵습니다. 하지만 사용자는 이러한 요청을 처리할 수 있는 도구를 제공합니다.\n### 2. 사용자의 요청 처리 방법\n사용자는 이메일, 문자 메시지, 또는 기타 방법으로 사용자의 요청을 처리할 수 있습니다.\n### 3. 사용자의 요청 처리 방법\n사용자는 사용자의 요청을 처리하기 위해 다음과 같은 방법을 사용할 수 있습니다.\n1. 사용자의 요청을 처리합니다.\n2. 사용자의 요청을 처리하기 위해 사용자의 요청을 처리합니다.\n3. 사용자의 요청을 처리하기 위해 사용자의 요청을 처리합니다.\n4. 사용자의 요청을 처리하기 위해 사용자의 요청을 처리합니다.\n5. 사용자의 요청을",
|
| 48 |
+
"metrics": {
|
| 49 |
+
"characters": 352,
|
| 50 |
+
"hangul_visible_ratio": 0.8888888888888888,
|
| 51 |
+
"unique_word_ratio": 0.46987951807228917
|
| 52 |
+
}
|
| 53 |
+
}
|
| 54 |
+
],
|
| 55 |
+
"note": "Korean instruction evaluation; responses are sampled, not facts checked."
|
| 56 |
+
}
|
reports/kawk50m-ko-instruct-v1/train.log
ADDED
|
@@ -0,0 +1,112 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"parameters": 51542528,
|
| 3 |
+
"device": "cuda",
|
| 4 |
+
"sequence_length": 5120,
|
| 5 |
+
"train_sequences": 15626,
|
| 6 |
+
"validation_sequences": 200,
|
| 7 |
+
"epochs": 2.0,
|
| 8 |
+
"sequences_per_optimizer_step": 24,
|
| 9 |
+
"tokens_per_optimizer_step": 122880,
|
| 10 |
+
"max_steps": 1303,
|
| 11 |
+
"planned_capacity_tokens": 160112640,
|
| 12 |
+
"start_step": 0,
|
| 13 |
+
"initialized_from": "/workspace/KAWK50M/runs/kawk50m-fixed-ko-cpt-3b-5k-5090/checkpoint-00024415",
|
| 14 |
+
"assistant_only_loss": true,
|
| 15 |
+
"precision": "bf16",
|
| 16 |
+
"compile": true
|
| 17 |
+
}
|
| 18 |
+
{"step": 1, "max_steps": 1303, "train_loss": 2.450684, "learning_rate": 1.282051282051282e-06, "grad_norm": 2.340644359588623, "supervised_tokens_in_step": 83853, "tokens_per_second": 6532}
|
| 19 |
+
{"step": 20, "max_steps": 1303, "train_loss": 2.347481, "learning_rate": 2.564102564102564e-05, "grad_norm": 0.8594114780426025, "supervised_tokens_in_step": 86722, "tokens_per_second": 115987}
|
| 20 |
+
Stop checkpoint: runs/kawk50m-ko-instruct-v1-5090/checkpoint-00000020
|
| 21 |
+
{
|
| 22 |
+
"parameters": 51542528,
|
| 23 |
+
"device": "cuda",
|
| 24 |
+
"sequence_length": 5120,
|
| 25 |
+
"train_sequences": 15626,
|
| 26 |
+
"validation_sequences": 200,
|
| 27 |
+
"epochs": 2.0,
|
| 28 |
+
"sequences_per_optimizer_step": 24,
|
| 29 |
+
"tokens_per_optimizer_step": 122880,
|
| 30 |
+
"max_steps": 1303,
|
| 31 |
+
"planned_capacity_tokens": 160112640,
|
| 32 |
+
"start_step": 20,
|
| 33 |
+
"initialized_from": null,
|
| 34 |
+
"assistant_only_loss": true,
|
| 35 |
+
"precision": "bf16",
|
| 36 |
+
"compile": true
|
| 37 |
+
}
|
| 38 |
+
{"step": 40, "max_steps": 1303, "train_loss": 2.333369, "learning_rate": 4.999993050430598e-05, "grad_norm": 0.8435298204421997, "supervised_tokens_in_step": 84949, "tokens_per_second": 100529}
|
| 39 |
+
{"step": 60, "max_steps": 1303, "train_loss": 2.324109, "learning_rate": 4.996935934013126e-05, "grad_norm": 0.8027786612510681, "supervised_tokens_in_step": 85477, "tokens_per_second": 115613}
|
| 40 |
+
{"step": 80, "max_steps": 1303, "train_loss": 2.271922, "learning_rate": 4.988327873548549e-05, "grad_norm": 0.761134147644043, "supervised_tokens_in_step": 85546, "tokens_per_second": 115473}
|
| 41 |
+
{"step": 100, "max_steps": 1303, "train_loss": 2.345757, "learning_rate": 4.974190134824319e-05, "grad_norm": 0.7603924870491028, "supervised_tokens_in_step": 87840, "tokens_per_second": 115765}
|
| 42 |
+
{"step": 120, "max_steps": 1303, "train_loss": 2.286366, "learning_rate": 4.9545576444244234e-05, "grad_norm": 0.7394975423812866, "supervised_tokens_in_step": 85646, "tokens_per_second": 115659}
|
| 43 |
+
{"step": 140, "max_steps": 1303, "train_loss": 2.319667, "learning_rate": 4.929478903444984e-05, "grad_norm": 0.797751784324646, "supervised_tokens_in_step": 89247, "tokens_per_second": 115754}
|
| 44 |
+
{"step": 160, "max_steps": 1303, "train_loss": 2.319262, "learning_rate": 4.8990158676747e-05, "grad_norm": 0.7572959065437317, "supervised_tokens_in_step": 85585, "tokens_per_second": 115685}
|
| 45 |
+
{"step": 180, "max_steps": 1303, "train_loss": 2.296479, "learning_rate": 4.863243794536136e-05, "grad_norm": 0.7168862223625183, "supervised_tokens_in_step": 88009, "tokens_per_second": 115681}
|
| 46 |
+
{"step": 200, "max_steps": 1303, "train_loss": 2.279972, "learning_rate": 4.8222510571659895e-05, "grad_norm": 0.7220127582550049, "supervised_tokens_in_step": 88893, "tokens_per_second": 115636}
|
| 47 |
+
/usr/local/lib/python3.12/dist-packages/torch/_inductor/compile_fx.py:282: UserWarning: TensorFloat32 tensor cores for float32 matrix multiplication available but not enabled. Consider setting `torch.set_float32_matmul_precision('high')` for better performance.
|
| 48 |
+
warnings.warn(
|
| 49 |
+
{"step": 200, "validation_loss": 2.403591, "validation_perplexity": 11.0628, "validation_supervised_tokens": 341104}
|
| 50 |
+
{"step": 220, "max_steps": 1303, "train_loss": 2.27585, "learning_rate": 4.776138926093622e-05, "grad_norm": 0.7102628350257874, "supervised_tokens_in_step": 85294, "tokens_per_second": 81196}
|
| 51 |
+
{"step": 240, "max_steps": 1303, "train_loss": 2.252894, "learning_rate": 4.725021319057228e-05, "grad_norm": 0.7745625376701355, "supervised_tokens_in_step": 84501, "tokens_per_second": 115715}
|
| 52 |
+
{"step": 260, "max_steps": 1303, "train_loss": 2.312163, "learning_rate": 4.669024519575705e-05, "grad_norm": 0.7289627194404602, "supervised_tokens_in_step": 85363, "tokens_per_second": 115689}
|
| 53 |
+
{"step": 280, "max_steps": 1303, "train_loss": 2.227963, "learning_rate": 4.6082868649714635e-05, "grad_norm": 0.7343097925186157, "supervised_tokens_in_step": 85313, "tokens_per_second": 115678}
|
| 54 |
+
{"step": 300, "max_steps": 1303, "train_loss": 2.253481, "learning_rate": 4.5429584046149166e-05, "grad_norm": 0.7375707626342773, "supervised_tokens_in_step": 88018, "tokens_per_second": 115699}
|
| 55 |
+
{"step": 320, "max_steps": 1303, "train_loss": 2.317738, "learning_rate": 4.4732005292349395e-05, "grad_norm": 0.7279749512672424, "supervised_tokens_in_step": 85609, "tokens_per_second": 115741}
|
| 56 |
+
{"step": 340, "max_steps": 1303, "train_loss": 2.268028, "learning_rate": 4.399185572211055e-05, "grad_norm": 0.7550586462020874, "supervised_tokens_in_step": 85623, "tokens_per_second": 115692}
|
| 57 |
+
{"step": 360, "max_steps": 1303, "train_loss": 2.293647, "learning_rate": 4.321096383832345e-05, "grad_norm": 0.7362149953842163, "supervised_tokens_in_step": 85354, "tokens_per_second": 115731}
|
| 58 |
+
{"step": 380, "max_steps": 1303, "train_loss": 2.217218, "learning_rate": 4.239125879574867e-05, "grad_norm": 0.7212095260620117, "supervised_tokens_in_step": 83026, "tokens_per_second": 115685}
|
| 59 |
+
{"step": 400, "max_steps": 1303, "train_loss": 2.242836, "learning_rate": 4.153476563513506e-05, "grad_norm": 0.7235660552978516, "supervised_tokens_in_step": 89569, "tokens_per_second": 115714}
|
| 60 |
+
{"step": 400, "validation_loss": 2.367918, "validation_perplexity": 10.6751, "validation_supervised_tokens": 341104}
|
| 61 |
+
{"step": 420, "max_steps": 1303, "train_loss": 2.154666, "learning_rate": 4.064360028045682e-05, "grad_norm": 0.7504831552505493, "supervised_tokens_in_step": 77337, "tokens_per_second": 110396}
|
| 62 |
+
{"step": 440, "max_steps": 1303, "train_loss": 2.249795, "learning_rate": 3.971996431162805e-05, "grad_norm": 0.793662965297699, "supervised_tokens_in_step": 85744, "tokens_per_second": 115753}
|
| 63 |
+
{"step": 460, "max_steps": 1303, "train_loss": 2.220676, "learning_rate": 3.876613952560843e-05, "grad_norm": 0.7667786478996277, "supervised_tokens_in_step": 81542, "tokens_per_second": 115756}
|
| 64 |
+
{"step": 480, "max_steps": 1303, "train_loss": 2.206832, "learning_rate": 3.778448229933679e-05, "grad_norm": 0.7093261480331421, "supervised_tokens_in_step": 84235, "tokens_per_second": 115761}
|
| 65 |
+
{"step": 500, "max_steps": 1303, "train_loss": 2.09693, "learning_rate": 3.677741776841846e-05, "grad_norm": 0.7661908268928528, "supervised_tokens_in_step": 78815, "tokens_per_second": 115740}
|
| 66 |
+
Saved checkpoint: runs/kawk50m-ko-instruct-v1-5090/checkpoint-00000500
|
| 67 |
+
{"step": 520, "max_steps": 1303, "train_loss": 2.304523, "learning_rate": 3.574743383594776e-05, "grad_norm": 0.7090751528739929, "supervised_tokens_in_step": 86564, "tokens_per_second": 113600}
|
| 68 |
+
{"step": 540, "max_steps": 1303, "train_loss": 2.214948, "learning_rate": 3.469707502626648e-05, "grad_norm": 0.7331322431564331, "supervised_tokens_in_step": 76710, "tokens_per_second": 115632}
|
| 69 |
+
{"step": 560, "max_steps": 1303, "train_loss": 2.299292, "learning_rate": 3.362893619884243e-05, "grad_norm": 0.7434960007667542, "supervised_tokens_in_step": 85048, "tokens_per_second": 115524}
|
| 70 |
+
{"step": 580, "max_steps": 1303, "train_loss": 2.099081, "learning_rate": 3.254565613779749e-05, "grad_norm": 0.7385064959526062, "supervised_tokens_in_step": 75457, "tokens_per_second": 115650}
|
| 71 |
+
{"step": 600, "max_steps": 1303, "train_loss": 2.2911, "learning_rate": 3.1449911032922045e-05, "grad_norm": 0.7000813484191895, "supervised_tokens_in_step": 88458, "tokens_per_second": 115412}
|
| 72 |
+
{"step": 600, "validation_loss": 2.345304, "validation_perplexity": 10.4364, "validation_supervised_tokens": 341104}
|
| 73 |
+
{"step": 620, "max_steps": 1303, "train_loss": 2.082757, "learning_rate": 3.034440786828045e-05, "grad_norm": 0.7202531695365906, "supervised_tokens_in_step": 76615, "tokens_per_second": 110341}
|
| 74 |
+
{"step": 640, "max_steps": 1303, "train_loss": 2.2964, "learning_rate": 2.9231877734740916e-05, "grad_norm": 0.6959240436553955, "supervised_tokens_in_step": 85483, "tokens_per_second": 115773}
|
| 75 |
+
{"step": 660, "max_steps": 1303, "train_loss": 2.18364, "learning_rate": 2.8115069082950614e-05, "grad_norm": 0.7376174926757812, "supervised_tokens_in_step": 85536, "tokens_per_second": 115723}
|
| 76 |
+
{"step": 680, "max_steps": 1303, "train_loss": 2.034453, "learning_rate": 2.6996740933424363e-05, "grad_norm": 0.6972545385360718, "supervised_tokens_in_step": 86494, "tokens_per_second": 115748}
|
| 77 |
+
{"step": 700, "max_steps": 1303, "train_loss": 2.160494, "learning_rate": 2.5879656060520946e-05, "grad_norm": 0.7508715987205505, "supervised_tokens_in_step": 82246, "tokens_per_second": 115712}
|
| 78 |
+
{"step": 720, "max_steps": 1303, "train_loss": 1.999812, "learning_rate": 2.4766574167145764e-05, "grad_norm": 0.7145387530326843, "supervised_tokens_in_step": 81685, "tokens_per_second": 115711}
|
| 79 |
+
{"step": 740, "max_steps": 1303, "train_loss": 2.092892, "learning_rate": 2.3660245067041364e-05, "grad_norm": 0.7211691737174988, "supervised_tokens_in_step": 83164, "tokens_per_second": 115726}
|
| 80 |
+
{"step": 760, "max_steps": 1303, "train_loss": 2.084289, "learning_rate": 2.256340189150864e-05, "grad_norm": 0.7231587767601013, "supervised_tokens_in_step": 82188, "tokens_per_second": 115740}
|
| 81 |
+
{"step": 780, "max_steps": 1303, "train_loss": 2.019705, "learning_rate": 2.147875433734121e-05, "grad_norm": 0.7101276516914368, "supervised_tokens_in_step": 80317, "tokens_per_second": 115725}
|
| 82 |
+
{"step": 800, "max_steps": 1303, "train_loss": 2.080972, "learning_rate": 2.040898197265353e-05, "grad_norm": 0.7223166823387146, "supervised_tokens_in_step": 83287, "tokens_per_second": 115599}
|
| 83 |
+
{"step": 800, "validation_loss": 2.335017, "validation_perplexity": 10.3296, "validation_supervised_tokens": 341104}
|
| 84 |
+
{"step": 820, "max_steps": 1303, "train_loss": 2.087878, "learning_rate": 1.9356727617140458e-05, "grad_norm": 0.7293902039527893, "supervised_tokens_in_step": 82008, "tokens_per_second": 110247}
|
| 85 |
+
{"step": 840, "max_steps": 1303, "train_loss": 2.12016, "learning_rate": 1.832459081312199e-05, "grad_norm": 0.6915454864501953, "supervised_tokens_in_step": 84157, "tokens_per_second": 115367}
|
| 86 |
+
{"step": 860, "max_steps": 1303, "train_loss": 2.165454, "learning_rate": 1.731512140350256e-05, "grad_norm": 0.7230251431465149, "supervised_tokens_in_step": 82607, "tokens_per_second": 115400}
|
| 87 |
+
{"step": 880, "max_steps": 1303, "train_loss": 2.074008, "learning_rate": 1.6330813232510332e-05, "grad_norm": 0.6996212005615234, "supervised_tokens_in_step": 84391, "tokens_per_second": 115428}
|
| 88 |
+
{"step": 900, "max_steps": 1303, "train_loss": 2.012958, "learning_rate": 1.537409798477839e-05, "grad_norm": 0.7183899879455566, "supervised_tokens_in_step": 84450, "tokens_per_second": 115550}
|
| 89 |
+
{"step": 920, "max_steps": 1303, "train_loss": 2.167079, "learning_rate": 1.4447339177988183e-05, "grad_norm": 0.6651352047920227, "supervised_tokens_in_step": 86696, "tokens_per_second": 115753}
|
| 90 |
+
{"step": 940, "max_steps": 1303, "train_loss": 2.202042, "learning_rate": 1.3552826323916119e-05, "grad_norm": 0.674192488193512, "supervised_tokens_in_step": 88343, "tokens_per_second": 115770}
|
| 91 |
+
{"step": 960, "max_steps": 1303, "train_loss": 2.171797, "learning_rate": 1.2692769272307864e-05, "grad_norm": 0.6765074133872986, "supervised_tokens_in_step": 88750, "tokens_per_second": 115778}
|
| 92 |
+
{"step": 980, "max_steps": 1303, "train_loss": 2.085758, "learning_rate": 1.1869292751553982e-05, "grad_norm": 0.7256887555122375, "supervised_tokens_in_step": 83377, "tokens_per_second": 115757}
|
| 93 |
+
{"step": 1000, "max_steps": 1303, "train_loss": 2.127656, "learning_rate": 1.1084431119653585e-05, "grad_norm": 0.7082517147064209, "supervised_tokens_in_step": 85872, "tokens_per_second": 115464}
|
| 94 |
+
{"step": 1000, "validation_loss": 2.326874, "validation_perplexity": 10.2459, "validation_supervised_tokens": 341104}
|
| 95 |
+
Saved checkpoint: runs/kawk50m-ko-instruct-v1-5090/checkpoint-00001000
|
| 96 |
+
{"step": 1020, "max_steps": 1303, "train_loss": 2.033393, "learning_rate": 1.0340123338433502e-05, "grad_norm": 0.7110400795936584, "supervised_tokens_in_step": 81863, "tokens_per_second": 108929}
|
| 97 |
+
{"step": 1040, "max_steps": 1303, "train_loss": 2.064529, "learning_rate": 9.638208183439063e-06, "grad_norm": 0.7083670496940613, "supervised_tokens_in_step": 84796, "tokens_per_second": 115765}
|
| 98 |
+
{"step": 1060, "max_steps": 1303, "train_loss": 2.118068, "learning_rate": 8.980419701330082e-06, "grad_norm": 0.7071433067321777, "supervised_tokens_in_step": 81092, "tokens_per_second": 115679}
|
| 99 |
+
{"step": 1080, "max_steps": 1303, "train_loss": 2.064492, "learning_rate": 8.36838292600437e-06, "grad_norm": 0.6941749453544617, "supervised_tokens_in_step": 82037, "tokens_per_second": 115425}
|
| 100 |
+
{"step": 1100, "max_steps": 1303, "train_loss": 2.037831, "learning_rate": 7.803609864031949e-06, "grad_norm": 0.6952521204948425, "supervised_tokens_in_step": 82017, "tokens_per_second": 115419}
|
| 101 |
+
{"step": 1120, "max_steps": 1303, "train_loss": 2.091592, "learning_rate": 7.287495759317571e-06, "grad_norm": 0.719948947429657, "supervised_tokens_in_step": 82368, "tokens_per_second": 115710}
|
| 102 |
+
{"step": 1140, "max_steps": 1303, "train_loss": 2.142646, "learning_rate": 6.821315646219733e-06, "grad_norm": 0.6843776106834412, "supervised_tokens_in_step": 82018, "tokens_per_second": 115759}
|
| 103 |
+
{"step": 1160, "max_steps": 1303, "train_loss": 2.099951, "learning_rate": 6.406221199641351e-06, "grad_norm": 0.6954435706138611, "supervised_tokens_in_step": 80084, "tokens_per_second": 115739}
|
| 104 |
+
{"step": 1180, "max_steps": 1303, "train_loss": 2.079199, "learning_rate": 6.043237889873913e-06, "grad_norm": 0.7361739873886108, "supervised_tokens_in_step": 80202, "tokens_per_second": 115484}
|
| 105 |
+
{"step": 1200, "max_steps": 1303, "train_loss": 2.065013, "learning_rate": 5.733262449223885e-06, "grad_norm": 0.707120954990387, "supervised_tokens_in_step": 83275, "tokens_per_second": 115429}
|
| 106 |
+
{"step": 1200, "validation_loss": 2.322639, "validation_perplexity": 10.2026, "validation_supervised_tokens": 341104}
|
| 107 |
+
{"step": 1220, "max_steps": 1303, "train_loss": 2.151211, "learning_rate": 5.477060656679872e-06, "grad_norm": 0.6623756289482117, "supervised_tokens_in_step": 86610, "tokens_per_second": 110407}
|
| 108 |
+
{"step": 1240, "max_steps": 1303, "train_loss": 2.194439, "learning_rate": 5.2752654460935424e-06, "grad_norm": 0.6283762454986572, "supervised_tokens_in_step": 90614, "tokens_per_second": 115738}
|
| 109 |
+
{"step": 1260, "max_steps": 1303, "train_loss": 2.174987, "learning_rate": 5.128375342547875e-06, "grad_norm": 0.6523645520210266, "supervised_tokens_in_step": 88230, "tokens_per_second": 115401}
|
| 110 |
+
{"step": 1280, "max_steps": 1303, "train_loss": 2.070068, "learning_rate": 5.036753230775589e-06, "grad_norm": 0.6893972754478455, "supervised_tokens_in_step": 86951, "tokens_per_second": 115404}
|
| 111 |
+
{"step": 1300, "max_steps": 1303, "train_loss": 2.102395, "learning_rate": 5.0006254586703764e-06, "grad_norm": 0.7105104327201843, "supervised_tokens_in_step": 82481, "tokens_per_second": 115412}
|
| 112 |
+
Final checkpoint: runs/kawk50m-ko-instruct-v1-5090/checkpoint-00001303
|
requirements-data.txt
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
-r requirements.txt
|
| 2 |
+
datasets>=4,<5
|
| 3 |
+
huggingface-hub>=0.36,<2
|
| 4 |
+
pyarrow>=20,<25
|
| 5 |
+
|
requirements-dev.txt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
-r requirements.txt
|
| 2 |
+
pytest>=8,<10
|
| 3 |
+
|
requirements.txt
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
torch>=2.4,<3
|
| 2 |
+
transformers>=4.48,<5
|
| 3 |
+
sentencepiece>=0.2,<1
|
| 4 |
+
protobuf>=4.25,<7
|
| 5 |
+
PyYAML>=6,<7
|
| 6 |
+
numpy>=1.26,<3
|
| 7 |
+
safetensors>=0.4,<1
|
| 8 |
+
tqdm>=4.66,<5
|
runpod/README.md
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# RunPod RTX 5090 실행
|
| 2 |
+
|
| 3 |
+
`KAWK-1.5-50M Korean Base`의 기준 환경은 RTX 5090 32GB 한 장입니다.
|
| 4 |
+
엄격 한국어 20B 베이스 사전학습 후 같은 가중치를 5,120 컨텍스트로
|
| 5 |
+
확장하여 3B 토큰을 계속학습합니다.
|
| 6 |
+
|
| 7 |
+
## Pod 권장값
|
| 8 |
+
|
| 9 |
+
- GPU: NVIDIA GeForce RTX 5090 1장
|
| 10 |
+
- 이미지: CUDA 12.8 계열 RunPod PyTorch
|
| 11 |
+
- Container disk: 20GB 이상
|
| 12 |
+
- Persistent volume: **200GB 권장**
|
| 13 |
+
- 프로젝트 경로: `/workspace/KAWK50M`
|
| 14 |
+
|
| 15 |
+
20B `uint16` 토큰은 약 40GB입니다. 원문·정제본, 비교 체크포인트,
|
| 16 |
+
중간 체크포인트까지 유지하므로 100GB volume은 빠듯합니다.
|
| 17 |
+
|
| 18 |
+
## 환경 준비
|
| 19 |
+
|
| 20 |
+
```bash
|
| 21 |
+
cd /workspace/KAWK50M
|
| 22 |
+
bash runpod/bootstrap.sh
|
| 23 |
+
```
|
| 24 |
+
|
| 25 |
+
`bootstrap.sh`는 시스템 Python을 변경하지 않고 `.venv`를 만들며 이미지의
|
| 26 |
+
CUDA PyTorch를 재사용합니다.
|
| 27 |
+
|
| 28 |
+
## 전체 빌드
|
| 29 |
+
|
| 30 |
+
초기 341만 엄격 정제 문서를 토큰화합니다.
|
| 31 |
+
|
| 32 |
+
```bash
|
| 33 |
+
bash runpod/tokenize_v2_initial.sh
|
| 34 |
+
```
|
| 35 |
+
|
| 36 |
+
고정된 FineWeb2 revision에서 20B까지 직접 스트리밍·필터·토큰화합니다.
|
| 37 |
+
원문을 별도로 저장하지 않으므로 volume 사용량을 줄일 수 있습니다.
|
| 38 |
+
|
| 39 |
+
```bash
|
| 40 |
+
bash runpod/stream_v2_20b.sh
|
| 41 |
+
```
|
| 42 |
+
|
| 43 |
+
중단 후 재개:
|
| 44 |
+
|
| 45 |
+
```bash
|
| 46 |
+
RESUME=1 bash runpod/stream_v2_20b.sh
|
| 47 |
+
```
|
| 48 |
+
|
| 49 |
+
20B manifest가 생긴 뒤 전체 후속 단계를 자동 실행:
|
| 50 |
+
|
| 51 |
+
```bash
|
| 52 |
+
bash runpod/continue_kawk_1_5_after_stream.sh
|
| 53 |
+
```
|
| 54 |
+
|
| 55 |
+
이 스크립트는 다음을 순서대로 수행합니다.
|
| 56 |
+
|
| 57 |
+
1. 1,024 컨텍스트 20B 베이스 사전학습
|
| 58 |
+
2. 고정 revision·허용 라이선스 한국어 RAG/대화 필터링
|
| 59 |
+
3. 코드, LaTeX, 수학/코드 source, 영어 중심 행 제외
|
| 60 |
+
4. 2.9B 장문 + 0.1B 구조 데이터 CPT 혼합
|
| 61 |
+
5. 5,120 컨텍스트 3B 계속학습
|
| 62 |
+
6. validation loss, UTF-8 byte loss, 생성 평가
|
| 63 |
+
7. 모델 카드와 SHA-256 체크섬 생성
|
| 64 |
+
|
| 65 |
+
## 속도 기준
|
| 66 |
+
|
| 67 |
+
실측 RTX 5090 합성 벤치마크:
|
| 68 |
+
|
| 69 |
+
- 1,024 컨텍스트: 약 266K tokens/s
|
| 70 |
+
- 5,120 컨텍스트, batch 4 × accumulation 6: 약 116K tokens/s
|
| 71 |
+
- 5,120 단계 peak VRAM: 약 8.5GiB
|
| 72 |
+
|
| 73 |
+
계산만 보면 20B는 약 21시간, 3B CPT는 약 7.2시간입니다. 데이터 공급,
|
| 74 |
+
평가와 체크포인트 저장 시간은 별도입니다.
|
| 75 |
+
|
| 76 |
+
## 로그와 결과
|
| 77 |
+
|
| 78 |
+
```text
|
| 79 |
+
/workspace/kawk-v2-stream-20b-wrapper.log
|
| 80 |
+
/workspace/kawk-v2-stream-20b-worker-00.log ... worker-03.log
|
| 81 |
+
/workspace/kawk-1.5-build-continuation.log
|
| 82 |
+
reports/v2-20b/
|
| 83 |
+
reports/kawk-1.5-50m-v2/
|
| 84 |
+
runs/kawk50m-v2-20b-5090/
|
| 85 |
+
runs/kawk-1.5-50m-v2-cpt-3b-5090/
|
| 86 |
+
```
|
| 87 |
+
|
| 88 |
+
상태 확인:
|
| 89 |
+
|
| 90 |
+
```bash
|
| 91 |
+
cat data/tokenized-v2-20b/manifest.json
|
| 92 |
+
tail -n 30 reports/v2-20b/train.log
|
| 93 |
+
tail -n 30 reports/kawk-1.5-50m-v2/train.log
|
| 94 |
+
nvidia-smi
|
| 95 |
+
```
|
| 96 |
+
|
| 97 |
+
Pod를 삭제하기 전에 최종 체크포인트, `MODEL_CARD.md`, 평가 JSON,
|
| 98 |
+
`checksums.sha256`을 로컬이나 별도 영구 저장소에 복사하십시오.
|
runpod/bootstrap.sh
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
set -euo pipefail
|
| 3 |
+
|
| 4 |
+
PROJECT_DIR="${PROJECT_DIR:-/workspace/KAWK50M}"
|
| 5 |
+
VENV_DIR="${VENV_DIR:-${PROJECT_DIR}/.venv}"
|
| 6 |
+
|
| 7 |
+
if [[ ! -d "${PROJECT_DIR}" ]]; then
|
| 8 |
+
echo "Project directory not found: ${PROJECT_DIR}" >&2
|
| 9 |
+
exit 1
|
| 10 |
+
fi
|
| 11 |
+
|
| 12 |
+
cd "${PROJECT_DIR}"
|
| 13 |
+
if [[ ! -x "${VENV_DIR}/bin/python" ]]; then
|
| 14 |
+
python -m venv --system-site-packages "${VENV_DIR}"
|
| 15 |
+
fi
|
| 16 |
+
|
| 17 |
+
# Keep the CUDA-enabled PyTorch bundled with the RunPod image available while
|
| 18 |
+
# installing this project into an isolated environment (PEP 668 compatible).
|
| 19 |
+
source "${VENV_DIR}/bin/activate"
|
| 20 |
+
python -m pip install --upgrade pip
|
| 21 |
+
python -m pip install -e .
|
| 22 |
+
mkdir -p data/raw data/clean data/tokenized artifacts runs
|
| 23 |
+
|
| 24 |
+
nvidia-smi
|
| 25 |
+
python -m scripts.model_info --config configs/model_50m.yaml
|
| 26 |
+
|
| 27 |
+
echo
|
| 28 |
+
echo "Bootstrap complete."
|
| 29 |
+
echo "Virtual environment: ${VENV_DIR}"
|
| 30 |
+
echo "After tokenizer and token binaries are present, run:"
|
| 31 |
+
echo " bash runpod/train_5090.sh"
|
runpod/download_5b_parallel.sh
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
set -euo pipefail
|
| 3 |
+
|
| 4 |
+
PROJECT_DIR="${PROJECT_DIR:-/workspace/KAWK50M}"
|
| 5 |
+
OUTPUT_ROOT="${OUTPUT_ROOT:-data/raw/fineweb2-kor-5b}"
|
| 6 |
+
REVISION="${REVISION:-af9c13333eb981300149d5ca60a8e9d659b276b9}"
|
| 7 |
+
WORKERS="${WORKERS:-4}"
|
| 8 |
+
TOTAL_DOCUMENTS="${TOTAL_DOCUMENTS:-6500000}"
|
| 9 |
+
|
| 10 |
+
cd "${PROJECT_DIR}"
|
| 11 |
+
source "${PROJECT_DIR}/.venv/bin/activate"
|
| 12 |
+
|
| 13 |
+
if (( TOTAL_DOCUMENTS % WORKERS != 0 )); then
|
| 14 |
+
echo "TOTAL_DOCUMENTS must be divisible by WORKERS" >&2
|
| 15 |
+
exit 1
|
| 16 |
+
fi
|
| 17 |
+
|
| 18 |
+
documents_per_worker=$((TOTAL_DOCUMENTS / WORKERS))
|
| 19 |
+
mkdir -p "${OUTPUT_ROOT}"
|
| 20 |
+
pids=()
|
| 21 |
+
|
| 22 |
+
for ((worker = 0; worker < WORKERS; worker++)); do
|
| 23 |
+
worker_dir="${OUTPUT_ROOT}/part-$(printf '%02d' "${worker}")"
|
| 24 |
+
worker_log="/workspace/kawk50m-download-5b-worker-$(printf '%02d' "${worker}").log"
|
| 25 |
+
if [[ -e "${worker_dir}" ]]; then
|
| 26 |
+
echo "Refusing to overwrite existing worker directory: ${worker_dir}" >&2
|
| 27 |
+
exit 1
|
| 28 |
+
fi
|
| 29 |
+
python -u -m scripts.download_fineweb2 \
|
| 30 |
+
--revision "${REVISION}" \
|
| 31 |
+
--output-dir "${worker_dir}" \
|
| 32 |
+
--max-documents "${documents_per_worker}" \
|
| 33 |
+
--documents-per-shard 50000 \
|
| 34 |
+
--shuffle-buffer 10000 \
|
| 35 |
+
--seed "$((42 + worker))" \
|
| 36 |
+
--compression-level 1 \
|
| 37 |
+
--dataset-num-shards "${WORKERS}" \
|
| 38 |
+
--dataset-shard-index "${worker}" \
|
| 39 |
+
> "${worker_log}" 2>&1 &
|
| 40 |
+
pids+=("$!")
|
| 41 |
+
echo "worker=${worker} pid=${pids[-1]} log=${worker_log}"
|
| 42 |
+
done
|
| 43 |
+
|
| 44 |
+
status=0
|
| 45 |
+
for pid in "${pids[@]}"; do
|
| 46 |
+
if ! wait "${pid}"; then
|
| 47 |
+
status=1
|
| 48 |
+
fi
|
| 49 |
+
done
|
| 50 |
+
|
| 51 |
+
if (( status != 0 )); then
|
| 52 |
+
echo "At least one download worker failed." >&2
|
| 53 |
+
exit "${status}"
|
| 54 |
+
fi
|
| 55 |
+
|
| 56 |
+
python - <<'PY'
|
| 57 |
+
from __future__ import annotations
|
| 58 |
+
|
| 59 |
+
import json
|
| 60 |
+
from pathlib import Path
|
| 61 |
+
|
| 62 |
+
root = Path("data/raw/fineweb2-kor-5b")
|
| 63 |
+
states = []
|
| 64 |
+
for state_path in sorted(root.glob("part-*/download_state.json")):
|
| 65 |
+
states.append(json.loads(state_path.read_text(encoding="utf-8")))
|
| 66 |
+
summary = {
|
| 67 |
+
"workers": len(states),
|
| 68 |
+
"documents": sum(int(state["documents_written"]) for state in states),
|
| 69 |
+
"utf8_text_gib": sum(int(state["utf8_bytes"]) for state in states) / 1024**3,
|
| 70 |
+
"shards": sum(int(state["completed_shards"]) for state in states),
|
| 71 |
+
"complete": bool(states) and all(bool(state["complete"]) for state in states),
|
| 72 |
+
"revision": sorted({state["revision"] for state in states}),
|
| 73 |
+
}
|
| 74 |
+
(root / "download_summary.json").write_text(
|
| 75 |
+
json.dumps(summary, ensure_ascii=False, indent=2) + "\n",
|
| 76 |
+
encoding="utf-8",
|
| 77 |
+
)
|
| 78 |
+
print(json.dumps(summary, ensure_ascii=False, indent=2))
|
| 79 |
+
PY
|
runpod/prepare_5b_after_download.sh
ADDED
|
@@ -0,0 +1,55 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
set -euo pipefail
|
| 3 |
+
|
| 4 |
+
PROJECT_DIR="${PROJECT_DIR:-/workspace/KAWK50M}"
|
| 5 |
+
DOWNLOAD_PID="${DOWNLOAD_PID:?Set DOWNLOAD_PID to the active coordinator PID}"
|
| 6 |
+
RAW_ROOT="${RAW_ROOT:-data/raw/fineweb2-kor-5b}"
|
| 7 |
+
CLEAN_DIR="${CLEAN_DIR:-data/clean-5b}"
|
| 8 |
+
TOKENIZED_DIR="${TOKENIZED_DIR:-data/tokenized-5b}"
|
| 9 |
+
TOKENIZER_DIR="${TOKENIZER_DIR:-artifacts/tokenizer-20k-pilot}"
|
| 10 |
+
|
| 11 |
+
cd "${PROJECT_DIR}"
|
| 12 |
+
source "${PROJECT_DIR}/.venv/bin/activate"
|
| 13 |
+
|
| 14 |
+
for output_dir in "${CLEAN_DIR}" "${TOKENIZED_DIR}"; do
|
| 15 |
+
if [[ -e "${output_dir}" ]]; then
|
| 16 |
+
echo "Refusing to overwrite existing output: ${output_dir}" >&2
|
| 17 |
+
exit 1
|
| 18 |
+
fi
|
| 19 |
+
done
|
| 20 |
+
|
| 21 |
+
echo "Waiting for download coordinator PID ${DOWNLOAD_PID}..."
|
| 22 |
+
while kill -0 "${DOWNLOAD_PID}" 2>/dev/null; do
|
| 23 |
+
sleep 30
|
| 24 |
+
done
|
| 25 |
+
|
| 26 |
+
python - <<'PY'
|
| 27 |
+
from __future__ import annotations
|
| 28 |
+
|
| 29 |
+
import json
|
| 30 |
+
from pathlib import Path
|
| 31 |
+
|
| 32 |
+
summary_path = Path("data/raw/fineweb2-kor-5b/download_summary.json")
|
| 33 |
+
if not summary_path.is_file():
|
| 34 |
+
raise SystemExit(f"Download summary missing: {summary_path}")
|
| 35 |
+
summary = json.loads(summary_path.read_text(encoding="utf-8"))
|
| 36 |
+
if not summary.get("complete") or int(summary.get("documents", 0)) != 6_500_000:
|
| 37 |
+
raise SystemExit(f"Download incomplete: {summary}")
|
| 38 |
+
print(json.dumps(summary, ensure_ascii=False, indent=2))
|
| 39 |
+
PY
|
| 40 |
+
|
| 41 |
+
echo "Starting clean/filter/dedup..."
|
| 42 |
+
python -u -m scripts.prepare_corpus \
|
| 43 |
+
--input "${RAW_ROOT}/part-*/*.jsonl.gz" \
|
| 44 |
+
--output-dir "${CLEAN_DIR}" \
|
| 45 |
+
--documents-per-shard 50000 \
|
| 46 |
+
--validation-ratio 0.002 \
|
| 47 |
+
--test-ratio 0.001 \
|
| 48 |
+
--seed 42 \
|
| 49 |
+
--workers 16 \
|
| 50 |
+
--worker-batch-size 64
|
| 51 |
+
|
| 52 |
+
echo "Starting tokenization..."
|
| 53 |
+
bash runpod/tokenize_5b_parallel.sh
|
| 54 |
+
|
| 55 |
+
echo "5B data preparation complete."
|
runpod/stream_v2_20b.sh
ADDED
|
@@ -0,0 +1,123 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
set -euo pipefail
|
| 3 |
+
|
| 4 |
+
PROJECT_DIR="${PROJECT_DIR:-/workspace/KAWK50M}"
|
| 5 |
+
OUTPUT_ROOT="${OUTPUT_ROOT:-data/tokenized-v2-20b/stream}"
|
| 6 |
+
INITIAL_MANIFEST="${INITIAL_MANIFEST:-data/tokenized-v2-20b/initial/manifest.json}"
|
| 7 |
+
TOKENIZER_DIR="${TOKENIZER_DIR:-artifacts/tokenizer-20k-v3}"
|
| 8 |
+
REVISION="${REVISION:-af9c13333eb981300149d5ca60a8e9d659b276b9}"
|
| 9 |
+
TARGET_TOKENS="${TARGET_TOKENS:-20000000000}"
|
| 10 |
+
DATASET_SHARDS="${DATASET_SHARDS:-4}"
|
| 11 |
+
SKIP_PER_SHARD="${SKIP_PER_SHARD:-1625000}"
|
| 12 |
+
FILTER_WORKERS_PER_STREAM="${FILTER_WORKERS_PER_STREAM:-4}"
|
| 13 |
+
RESUME="${RESUME:-0}"
|
| 14 |
+
|
| 15 |
+
cd "${PROJECT_DIR}"
|
| 16 |
+
source "${PROJECT_DIR}/.venv/bin/activate"
|
| 17 |
+
ulimit -c 0
|
| 18 |
+
|
| 19 |
+
if [[ ! -f "${INITIAL_MANIFEST}" ]]; then
|
| 20 |
+
echo "Initial token manifest is missing: ${INITIAL_MANIFEST}" >&2
|
| 21 |
+
exit 1
|
| 22 |
+
fi
|
| 23 |
+
if [[ -e "${OUTPUT_ROOT}" && "${RESUME}" != "1" ]]; then
|
| 24 |
+
echo "Refusing to overwrite existing stream output: ${OUTPUT_ROOT}" >&2
|
| 25 |
+
exit 1
|
| 26 |
+
fi
|
| 27 |
+
mkdir -p "${OUTPUT_ROOT}"
|
| 28 |
+
|
| 29 |
+
read -r initial_tokens tokens_per_worker < <(
|
| 30 |
+
python - "${INITIAL_MANIFEST}" "${TARGET_TOKENS}" "${DATASET_SHARDS}" <<'PY'
|
| 31 |
+
import json
|
| 32 |
+
import math
|
| 33 |
+
import sys
|
| 34 |
+
|
| 35 |
+
initial = int(json.load(open(sys.argv[1], encoding="utf-8"))["tokens"])
|
| 36 |
+
target = int(sys.argv[2])
|
| 37 |
+
workers = int(sys.argv[3])
|
| 38 |
+
remaining = target - initial
|
| 39 |
+
if remaining <= 0:
|
| 40 |
+
raise SystemExit("Initial corpus already reaches the target")
|
| 41 |
+
print(initial, math.ceil(remaining / workers))
|
| 42 |
+
PY
|
| 43 |
+
)
|
| 44 |
+
echo "initial_tokens=${initial_tokens} tokens_per_stream=${tokens_per_worker}"
|
| 45 |
+
|
| 46 |
+
pids=()
|
| 47 |
+
for ((worker = 0; worker < DATASET_SHARDS; worker++)); do
|
| 48 |
+
worker_dir="${OUTPUT_ROOT}/part-$(printf '%02d' "${worker}")"
|
| 49 |
+
worker_log="/workspace/kawk-v2-stream-20b-worker-$(printf '%02d' "${worker}").log"
|
| 50 |
+
resume_args=()
|
| 51 |
+
if [[ -e "${worker_dir}" ]]; then
|
| 52 |
+
if [[ "${RESUME}" != "1" || ! -f "${worker_dir}/state.json" ]]; then
|
| 53 |
+
echo "Worker output exists without a resumable state: ${worker_dir}" >&2
|
| 54 |
+
exit 1
|
| 55 |
+
fi
|
| 56 |
+
resume_args+=(--resume)
|
| 57 |
+
fi
|
| 58 |
+
python -u -m scripts.stream_tokenize_fineweb2 \
|
| 59 |
+
--revision "${REVISION}" \
|
| 60 |
+
--tokenizer-dir "${TOKENIZER_DIR}" \
|
| 61 |
+
--output-dir "${worker_dir}" \
|
| 62 |
+
--dataset-num-shards "${DATASET_SHARDS}" \
|
| 63 |
+
--dataset-shard-index "${worker}" \
|
| 64 |
+
--skip-documents "${SKIP_PER_SHARD}" \
|
| 65 |
+
--max-output-tokens "${tokens_per_worker}" \
|
| 66 |
+
--tokens-per-chunk 250000000 \
|
| 67 |
+
--shuffle-buffer 10000 \
|
| 68 |
+
--seed "$((42 + worker))" \
|
| 69 |
+
--workers "${FILTER_WORKERS_PER_STREAM}" \
|
| 70 |
+
--worker-batch-size 64 \
|
| 71 |
+
--tokenize-batch-size 512 \
|
| 72 |
+
"${resume_args[@]}" \
|
| 73 |
+
> "${worker_log}" 2>&1 &
|
| 74 |
+
pids+=("$!")
|
| 75 |
+
echo "worker=${worker} pid=${pids[-1]} log=${worker_log}"
|
| 76 |
+
done
|
| 77 |
+
|
| 78 |
+
status=0
|
| 79 |
+
for pid in "${pids[@]}"; do
|
| 80 |
+
if ! wait "${pid}"; then
|
| 81 |
+
status=1
|
| 82 |
+
fi
|
| 83 |
+
done
|
| 84 |
+
if (( status != 0 )); then
|
| 85 |
+
echo "At least one stream-tokenization worker failed." >&2
|
| 86 |
+
exit "${status}"
|
| 87 |
+
fi
|
| 88 |
+
|
| 89 |
+
python - "${INITIAL_MANIFEST}" "${OUTPUT_ROOT}" "${TARGET_TOKENS}" <<'PY'
|
| 90 |
+
from __future__ import annotations
|
| 91 |
+
|
| 92 |
+
import json
|
| 93 |
+
import sys
|
| 94 |
+
from pathlib import Path
|
| 95 |
+
|
| 96 |
+
initial_manifest = Path(sys.argv[1])
|
| 97 |
+
output_root = Path(sys.argv[2])
|
| 98 |
+
target = int(sys.argv[3])
|
| 99 |
+
initial = json.loads(initial_manifest.read_text(encoding="utf-8"))
|
| 100 |
+
states = [
|
| 101 |
+
json.loads(path.read_text(encoding="utf-8"))
|
| 102 |
+
for path in sorted(output_root.glob("part-*/state.json"))
|
| 103 |
+
]
|
| 104 |
+
summary = {
|
| 105 |
+
"initial_tokens": int(initial["tokens"]),
|
| 106 |
+
"stream_workers": len(states),
|
| 107 |
+
"stream_tokens": sum(int(state["tokens_written"]) for state in states),
|
| 108 |
+
"total_tokens": int(initial["tokens"])
|
| 109 |
+
+ sum(int(state["tokens_written"]) for state in states),
|
| 110 |
+
"complete": bool(states) and all(bool(state["complete"]) for state in states),
|
| 111 |
+
"revision": sorted({state["revision"] for state in states}),
|
| 112 |
+
}
|
| 113 |
+
if not summary["complete"] or summary["total_tokens"] < target:
|
| 114 |
+
raise SystemExit(f"Incomplete 20B stream: {summary}")
|
| 115 |
+
(output_root.parent / "manifest.json").write_text(
|
| 116 |
+
json.dumps(summary, ensure_ascii=False, indent=2) + "\n",
|
| 117 |
+
encoding="utf-8",
|
| 118 |
+
)
|
| 119 |
+
print(json.dumps(summary, ensure_ascii=False, indent=2))
|
| 120 |
+
PY
|
| 121 |
+
|
| 122 |
+
python -m scripts.preflight --config configs/pretrain_v2_20b_5090.yaml
|
| 123 |
+
echo "KAWK50M v2 20B token corpus complete."
|
runpod/tokenize_5b_parallel.sh
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
set -euo pipefail
|
| 3 |
+
|
| 4 |
+
PROJECT_DIR="${PROJECT_DIR:-/workspace/KAWK50M}"
|
| 5 |
+
CLEAN_DIR="${CLEAN_DIR:-data/clean-5b}"
|
| 6 |
+
TOKENIZED_DIR="${TOKENIZED_DIR:-data/tokenized-5b}"
|
| 7 |
+
TOKENIZER_DIR="${TOKENIZER_DIR:-artifacts/tokenizer-20k-pilot}"
|
| 8 |
+
WORKERS="${WORKERS:-16}"
|
| 9 |
+
|
| 10 |
+
cd "${PROJECT_DIR}"
|
| 11 |
+
source "${PROJECT_DIR}/.venv/bin/activate"
|
| 12 |
+
|
| 13 |
+
if [[ -e "${TOKENIZED_DIR}" ]]; then
|
| 14 |
+
echo "Refusing to overwrite existing output: ${TOKENIZED_DIR}" >&2
|
| 15 |
+
exit 1
|
| 16 |
+
fi
|
| 17 |
+
mkdir -p "${TOKENIZED_DIR}/parts"
|
| 18 |
+
|
| 19 |
+
tokenize_parts() {
|
| 20 |
+
local split="$1"
|
| 21 |
+
local split_dir="${TOKENIZED_DIR}/parts/${split}"
|
| 22 |
+
mkdir -p "${split_dir}"
|
| 23 |
+
find "${CLEAN_DIR}" -maxdepth 1 -type f -name "${split}-*.jsonl" -print0 \
|
| 24 |
+
| sort -z \
|
| 25 |
+
| xargs -0 -r -P "${WORKERS}" -I '{}' bash -c '
|
| 26 |
+
input="$1"
|
| 27 |
+
output_dir="$2"
|
| 28 |
+
tokenizer_dir="$3"
|
| 29 |
+
stem="$(basename "${input%.jsonl}")"
|
| 30 |
+
python -u -m scripts.tokenize_shard \
|
| 31 |
+
--input "${input}" \
|
| 32 |
+
--tokenizer-dir "${tokenizer_dir}" \
|
| 33 |
+
--output "${output_dir}/${stem}.bin" \
|
| 34 |
+
--batch-size 64
|
| 35 |
+
' _ '{}' "${split_dir}" "${TOKENIZER_DIR}"
|
| 36 |
+
}
|
| 37 |
+
|
| 38 |
+
tokenize_parts train
|
| 39 |
+
tokenize_parts validation
|
| 40 |
+
tokenize_parts test
|
| 41 |
+
|
| 42 |
+
python -m scripts.combine_token_parts \
|
| 43 |
+
--input-dir "${TOKENIZED_DIR}/parts/train" \
|
| 44 |
+
--output "${TOKENIZED_DIR}/train.bin" \
|
| 45 |
+
--max-tokens 5000000000
|
| 46 |
+
python -m scripts.combine_token_parts \
|
| 47 |
+
--input-dir "${TOKENIZED_DIR}/parts/validation" \
|
| 48 |
+
--output "${TOKENIZED_DIR}/validation.bin" \
|
| 49 |
+
--max-tokens 2000000
|
| 50 |
+
python -m scripts.combine_token_parts \
|
| 51 |
+
--input-dir "${TOKENIZED_DIR}/parts/test" \
|
| 52 |
+
--output "${TOKENIZED_DIR}/test.bin" \
|
| 53 |
+
--max-tokens 1000000
|
| 54 |
+
|
| 55 |
+
python - <<'PY'
|
| 56 |
+
from __future__ import annotations
|
| 57 |
+
|
| 58 |
+
import json
|
| 59 |
+
from pathlib import Path
|
| 60 |
+
|
| 61 |
+
root = Path("data/tokenized-5b")
|
| 62 |
+
splits = {}
|
| 63 |
+
for split in ("train", "validation", "test"):
|
| 64 |
+
path = root / f"{split}.metadata.json"
|
| 65 |
+
splits[split] = json.loads(path.read_text(encoding="utf-8"))
|
| 66 |
+
metadata = {
|
| 67 |
+
"tokenizer": str(Path("artifacts/tokenizer-20k-pilot").resolve()),
|
| 68 |
+
"vocab_size": 20000,
|
| 69 |
+
"eos_token_id": 2,
|
| 70 |
+
"splits": splits,
|
| 71 |
+
}
|
| 72 |
+
(root / "metadata.json").write_text(
|
| 73 |
+
json.dumps(metadata, ensure_ascii=False, indent=2) + "\n",
|
| 74 |
+
encoding="utf-8",
|
| 75 |
+
)
|
| 76 |
+
print(json.dumps(metadata, ensure_ascii=False, indent=2))
|
| 77 |
+
PY
|
| 78 |
+
|
| 79 |
+
echo "Parallel 5B tokenization complete."
|
runpod/train_5090.sh
ADDED
|
@@ -0,0 +1,37 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
#!/usr/bin/env bash
|
| 2 |
+
set -euo pipefail
|
| 3 |
+
|
| 4 |
+
PROJECT_DIR="${PROJECT_DIR:-/workspace/KAWK50M}"
|
| 5 |
+
VENV_DIR="${VENV_DIR:-${PROJECT_DIR}/.venv}"
|
| 6 |
+
CONFIG="${CONFIG:-configs/pretrain_5b_5090.yaml}"
|
| 7 |
+
RESUME="${RESUME:-}"
|
| 8 |
+
STOP_AFTER_STEPS="${STOP_AFTER_STEPS:-}"
|
| 9 |
+
|
| 10 |
+
cd "${PROJECT_DIR}"
|
| 11 |
+
if [[ ! -x "${VENV_DIR}/bin/python" ]]; then
|
| 12 |
+
echo "Virtual environment not found: ${VENV_DIR}" >&2
|
| 13 |
+
echo "Run: bash runpod/bootstrap.sh" >&2
|
| 14 |
+
exit 1
|
| 15 |
+
fi
|
| 16 |
+
source "${VENV_DIR}/bin/activate"
|
| 17 |
+
|
| 18 |
+
python -m scripts.preflight --config "${CONFIG}"
|
| 19 |
+
|
| 20 |
+
if [[ "${SKIP_BENCHMARK:-0}" != "1" ]]; then
|
| 21 |
+
python -m scripts.benchmark \
|
| 22 |
+
--batch-size 32 \
|
| 23 |
+
--gradient-accumulation 4 \
|
| 24 |
+
--compile \
|
| 25 |
+
--warmup-steps 1 \
|
| 26 |
+
--steps 3
|
| 27 |
+
fi
|
| 28 |
+
|
| 29 |
+
command=(python -m scripts.pretrain --config "${CONFIG}")
|
| 30 |
+
if [[ -n "${RESUME}" ]]; then
|
| 31 |
+
command+=(--resume "${RESUME}")
|
| 32 |
+
fi
|
| 33 |
+
if [[ -n "${STOP_AFTER_STEPS}" ]]; then
|
| 34 |
+
command+=(--stop-after-steps "${STOP_AFTER_STEPS}")
|
| 35 |
+
fi
|
| 36 |
+
|
| 37 |
+
"${command[@]}"
|
scripts/__init__.py
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""Command-line entry modules. Run them with ``python -m scripts.<name>``."""
|
| 2 |
+
|