jamba-900M-v0.13-KIx2

jamba-900M-v0.13-KIx2

<a href="https://colab.research.google.com/gist/pszemraj/62d037d0d93656ef2101d7e29e3b7220/jamba-test-sandbox.ipynb"> <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"/> </a>

The API widget is off as it isn't supported by hf yet - try the Colab

This is a pretraining experiment on the jamba arch as a "smol MoE".

Details:

• pretrained at context length 16384
• seen approx 20b tokens
• uses Claude3 tokenizer (as hf GPT2 tokenizer)
• hidden size 1024, 12 layers, 8 experts

achieves the following results on the evaluation set (most recent dataset):

• Loss: 3.0366
• Accuracy: 0.4514
• Num Input Tokens Seen: 1975517184

if I pretrain it further, other versions will be in new repos with incremented version (this is v0.13)

Quick eval

Quick eval for: pszemraj/jamba-H1024_L12-v0.13-KIx2

hf (pretrained=pszemraj/jamba-H1024_L12-v0.13-KIx2,trust_remote_code=True,dtype=float), gen_kwargs: (None), limit: 0.9999, num_fewshot: None, batch_size: 8

example outputs

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 80085
• gradient_accumulation_steps: 32
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 2.0

Training results

Framework versions

• Transformers 4.40.1
• Pytorch 2.2.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1