0.0.63

What's Changed

• Fixed an iPython loading error introduced in 0.0.62
• Fixed a negative index setting error
• Use regex to parse OpenAI models by @a2975667 in #240

New Contributors

• @a2975667 made their first contribution in #240

Full Changelog: 0.0.62...0.0.63

0.0.62

What's Changed

• fix: Changed the example output to match the behaviour of the code. by @FunnyPhantom in #176
• Allow variables to start with "or" or "else" by @Mihaiii in #179
• Fix examples for issues #172 by @Keiku in #182
• New variable_stack for the parser by @slundberg in #193
• fixed openai streaming with REST by @marcotcr in #208
• Now using pyparsing and supporting infix notation! by @slundberg in #230

New Contributors

• @FunnyPhantom made their first contribution in #176
• @Keiku made their first contribution in #182

Full Changelog: 0.0.61...0.0.62

0.0.61

• Adds working support for Azure OpenAI
  Full Changelog: 0.0.60...0.0.61

0.0.60

• This is another quick patch for a cache issue when loading transformer models manually.
  Full Changelog: 0.0.59...0.0.60

0.0.59

• Just a patch to address a cache bug.

Full Changelog: 0.0.58...0.0.59

0.0.58

What's Changed

• Support for multi-token token healing (healing that requires backing up more than one token).
• Support falcon models.
• Fix GPT proverb example with multiple gens by @bcstyle in #137
• Fix geneach demo code by @Shaunwei in #149
• Enhance the cache ability by @SimFG in #123
• Fix partials and empty blocks by @justinbowes in #155
• Adding Microsoft SECURITY.MD by @microsoft-github-policy-service in #168

New Contributors

• @bcstyle made their first contribution in #137
• @Shaunwei made their first contribution in #149
• @SimFG made their first contribution in #123
• @justinbowes made their first contribution in #155
• @microsoft-github-policy-service made their first contribution in #168

Full Changelog: 0.0.57...0.0.58

0.0.57

What's Changed

• Now silent=True is the default when streaming the program output.
• Fixed some bugs with caching, and also select with special EOS tokens in LLaMA.
• Fix typos by @amqdn in #135

New Contributors

• @amqdn made their first contribution in #135

Full Changelog: 0.0.56...0.0.57

0.0.56

What's Changed

• New support for stop_regex with OpenAI models.
• New support for streaming the results of program execution programmatically both sync and async.
• Update test_unless.py by @Jvr2022 in #96
• Add gpt-4-32k and gpt-4-32k-0314 to OpenAI Models by @flavius-popan in #98
• Support models that doesn't output past_key_values by @younesbelkada in #91
• add quantization parameters for transformers models by @jquesnelle in #8
• Fix unnecessary log.txt file creation by @gerelltroche in #106
• Fix TypeError when serializing out.variables() to JSON by @h-k-nyosu in #120
• fix: Add missing msal dependency. by @shawnohare in #68
• Export LLMSession and SyncSession by @abetlen in #66
• add timeout parameter to POST request by @satarupaguha11 in #125

New Contributors

• @Jvr2022 made their first contribution in #96
• @flavius-popan made their first contribution in #98
• @younesbelkada made their first contribution in #91
• @jquesnelle made their first contribution in #8
• @gerelltroche made their first contribution in #106
• @h-k-nyosu made their first contribution in #120
• @shawnohare made their first contribution in #68
• @abetlen made their first contribution in #66
• @satarupaguha11 made their first contribution in #125

Full Changelog: 0.0.54...0.0.56

0.0.54

• Switch the select command over to a token-id-based process to account for possible non-greedy tokenizer exceptions.

Full Changelog: 0.0.53...0.0.54

0.0.53

What's Changed

• Allow for select statements to have options that are prefixes of other options (and fix one select bug with long options).
• fix repo typo by @marcosmagallanes in #86
• fix typo by @kajarenc in #85

New Contributors

• @marcosmagallanes made their first contribution in #86
• @kajarenc made their first contribution in #85

Full Changelog: 0.0.52...0.0.53