StarCoder2 is the next generation of transparently trained open code LLMs that comes in three sizes: 3B, 7B and 15B parameters.

3b 7b 15b

436.7K 2 months ago

Readme

StarCode 2

Supporting a context window of up to 16,384 tokens, StarCoder2 is the next generation of transparently trained open code LLMs.

  • starcoder2:instruct: a 15B model that follows natural and human-written instructions
  • starcoder2:15b was trained on 600+ programming languages and 4+ trillion tokens.
  • starcoder2:7b was trained on 17 programming languages and 3.5+ trillion tokens.
  • starcoder2:3b was trained on 17 programming languages and 3+ trillion tokens.

StarCoder2-15B is the best in its size class and matches 33B+ models on many evaluations. StarCoder2-3B matches the performance of StarCoder1-15B.

References