llm icon indicating copy to clipboard operation
llm copied to clipboard
verified publisher icon rustformers

Write a 0.2 changelog

Open philpax opened this issue 2 years ago • 8 comments

There's been quite a few changes from 0.1. We should document them for people updating their applications.

philpax avatar May 18 '23 17:05 philpax

Thoughts on getting this into the next release? https://github.com/rustformers/llm/pull/224

danforbes avatar May 18 '23 17:05 danforbes

Fine by me!

philpax avatar May 18 '23 18:05 philpax

I guess we'll want to look at this in order to figure out what goes into the changelog https://github.com/rustformers/llm/compare/v0.1.1...main

danforbes avatar Jun 01 '23 13:06 danforbes

Looking at the history for the inference.rs example is another good way to track changes to the public API https://github.com/rustformers/llm/commits/main/crates/llm/examples/inference.rs

danforbes avatar Jun 01 '23 13:06 danforbes

we can do so automatically here https://docs.github.com/en/repositories/releasing-projects-on-github/automatically-generated-release-notes

hhamud avatar Aug 07 '23 13:08 hhamud

Example here

What's Changed

  • refactor: move load to KnownModel trait by @philpax in https://github.com/rustformers/llm/pull/199
  • Refactors from llm-chain integration by @danforbes in https://github.com/rustformers/llm/pull/200
  • Fine-grain inference feedback by @danforbes in https://github.com/rustformers/llm/pull/206
  • Update upstream GGML dependency by @danforbes in https://github.com/rustformers/llm/pull/215
  • Implement RedPajama as a NeoX Specialization by @danforbes in https://github.com/rustformers/llm/pull/204
  • Implement LoRA patching via the Loader by @LLukas22 in https://github.com/rustformers/llm/pull/211
  • Dynamically calculate the LoRA patching context size by @LLukas22 in https://github.com/rustformers/llm/pull/225
  • NeoX -> GPT-NeoX by @danforbes in https://github.com/rustformers/llm/pull/233
  • Update GGML dependency by @danforbes in https://github.com/rustformers/llm/pull/226
  • fix: cancel the loading spinner in error cases by @tehmatt in https://github.com/rustformers/llm/pull/236
  • Added support for multiple LoRA adapters by @LLukas22 in https://github.com/rustformers/llm/pull/232
  • fix: qnt0 build by @philpax in https://github.com/rustformers/llm/pull/242
  • Typo fix in extern function name by @tanmaysachan in https://github.com/rustformers/llm/pull/247
  • Implement MPT Model by @danforbes in https://github.com/rustformers/llm/pull/218
  • feat: support quantizing with a GGML container by @philpax in https://github.com/rustformers/llm/pull/245
  • Add perplexity calculation and cli-arg by @tanmaysachan in https://github.com/rustformers/llm/pull/248
  • Update Gpt-NeoX implementation by @LLukas22 in https://github.com/rustformers/llm/pull/237
  • Fix perplexity calculation by @tanmaysachan in https://github.com/rustformers/llm/pull/251
  • GGML README by @danforbes in https://github.com/rustformers/llm/pull/224
  • Lingering NeoX reference by @danforbes in https://github.com/rustformers/llm/pull/254
  • Model architectures should appear in alphabetic order by @danforbes in https://github.com/rustformers/llm/pull/255
  • More model standardization by @danforbes in https://github.com/rustformers/llm/pull/256
  • README updates for v0.2 by @danforbes in https://github.com/rustformers/llm/pull/253
  • More CONTRIBUTING updates by @danforbes in https://github.com/rustformers/llm/pull/258
  • Update headings by @danforbes in https://github.com/rustformers/llm/pull/259
  • Context size consistency & fixes by @danforbes in https://github.com/rustformers/llm/pull/260
  • Remove LLaMA "old_loader" by @danforbes in https://github.com/rustformers/llm/pull/262
  • refactor: remove model-associated inference params by @philpax in https://github.com/rustformers/llm/pull/272
  • Pull upstream changes into GptNeox by @LLukas22 in https://github.com/rustformers/llm/pull/270
  • Fix embeddings extraction by @skirodev in https://github.com/rustformers/llm/pull/273
  • Update GGML Dependency & Bindings by @danforbes in https://github.com/rustformers/llm/pull/275
  • Add scratchbuffers to MPT by @LLukas22 in https://github.com/rustformers/llm/pull/278
  • Fixed broken GPT-J quantization by @LLukas22 in https://github.com/rustformers/llm/pull/283
  • Add HuggingFace's Tokenizer by @RedBoxing in https://github.com/rustformers/llm/pull/271
  • Sampler rework by @philpax in https://github.com/rustformers/llm/pull/280
  • Add alternative key shortcuts for shift enter by @pdufour in https://github.com/rustformers/llm/pull/287
  • Fix error in Usage section of README by @danforbes in https://github.com/rustformers/llm/pull/290
  • Made Sampler and InferenceParameters threadsafe by @LLukas22 in https://github.com/rustformers/llm/pull/292
  • Fix embeddings extraction for all models by @skirodev in https://github.com/rustformers/llm/pull/291
  • Fix --help panicing in debug mode by @tehmatt in https://github.com/rustformers/llm/pull/300
  • Default to f16 model memory k/v in llm CLI and InferenceSessionConfig by @KerfuffleV2 in https://github.com/rustformers/llm/pull/296
  • Switch from GGML to LLama.cpp as GGML source. by @LLukas22 in https://github.com/rustformers/llm/pull/302
  • NixOS flake stuff needs to be in root by @danforbes in https://github.com/rustformers/llm/pull/306
  • Add bindings section to README by @danforbes in https://github.com/rustformers/llm/pull/309
  • adding secondbrain to applications section on Readme by @juliooa in https://github.com/rustformers/llm/pull/310
  • Added support for cuBLAS and CLBlast in ggml. by @darxkies in https://github.com/rustformers/llm/pull/282
  • Fix GPT-2 Positional Embeddings by @LLukas22 in https://github.com/rustformers/llm/pull/322
  • Implement Metal support by @pixelspark in https://github.com/rustformers/llm/pull/311
  • fix(metal): patch ggml-metal.m with shader code by @philpax in https://github.com/rustformers/llm/pull/328
  • K-quants by @nightscape in https://github.com/rustformers/llm/pull/326
  • feat: automatically run checks before commit by @philpax in https://github.com/rustformers/llm/pull/329
  • Added InferenceSessionRef to exports, for easier serializing and deserializing. by @JuliaMerz in https://github.com/rustformers/llm/pull/335
  • Add Falcon Support by @LLukas22 in https://github.com/rustformers/llm/pull/313
  • Added tokenizers-remote by @jafioti in https://github.com/rustformers/llm/pull/339
  • Basic Integration Testing by @LLukas22 in https://github.com/rustformers/llm/pull/337
  • feat(test): check hyperparameters can roundtrip by @philpax in https://github.com/rustformers/llm/pull/344
  • Improve model architecture error message by @viirya in https://github.com/rustformers/llm/pull/348
  • fix #216 - split prompts for chat mode in CLI by @philpax in https://github.com/rustformers/llm/pull/346
  • Add placeholder check for message prompt by @viirya in https://github.com/rustformers/llm/pull/351
  • Fix linux CLBlast build and add CI for accelerated builds by @LLukas22 in https://github.com/rustformers/llm/pull/358
  • Add CI for cuda builds by @LLukas22 in https://github.com/rustformers/llm/pull/360
  • Add ability to delete tokens (undo feed) by @steventrouble in https://github.com/rustformers/llm/pull/334
  • Update gpt2 to use wte if no lm_head by @steventrouble in https://github.com/rustformers/llm/pull/362
  • Add CI for OpenCL on windows by @LLukas22 in https://github.com/rustformers/llm/pull/366
  • allow chat to halt new token generation on stop_sequence by @averypelle in https://github.com/rustformers/llm/pull/364
  • feat(tracing): add tracing to llm and llm-base crates by @radu-matei in https://github.com/rustformers/llm/pull/367
  • CUDA/OpenCL Acceleration by @LLukas22 in https://github.com/rustformers/llm/pull/325
  • Fix MSRV to 1.65.0 by @philpax in https://github.com/rustformers/llm/pull/375
  • Peg CUDA architecture by @LLukas22 in https://github.com/rustformers/llm/pull/376
  • Update README.md by @basraayman in https://github.com/rustformers/llm/pull/379
  • Update README.md by @baskrahmer in https://github.com/rustformers/llm/pull/380
  • Better GPU Documentation by @LLukas22 in https://github.com/rustformers/llm/pull/385
  • add Falcon 40B model support by @skirodev in https://github.com/rustformers/llm/pull/368
  • Custom RoPE Scaling by @LLukas22 in https://github.com/rustformers/llm/pull/389
  • Fix #393 halting feed_prompt by @clarkmcc in https://github.com/rustformers/llm/pull/394
  • Use llm_samplers crate for sampler backend by @KerfuffleV2 in https://github.com/rustformers/llm/pull/359

New Contributors

  • @skirodev made their first contribution in https://github.com/rustformers/llm/pull/273
  • @RedBoxing made their first contribution in https://github.com/rustformers/llm/pull/271
  • @pdufour made their first contribution in https://github.com/rustformers/llm/pull/287
  • @juliooa made their first contribution in https://github.com/rustformers/llm/pull/310
  • @darxkies made their first contribution in https://github.com/rustformers/llm/pull/282
  • @pixelspark made their first contribution in https://github.com/rustformers/llm/pull/311
  • @nightscape made their first contribution in https://github.com/rustformers/llm/pull/326
  • @JuliaMerz made their first contribution in https://github.com/rustformers/llm/pull/335
  • @jafioti made their first contribution in https://github.com/rustformers/llm/pull/339
  • @viirya made their first contribution in https://github.com/rustformers/llm/pull/348
  • @steventrouble made their first contribution in https://github.com/rustformers/llm/pull/334
  • @averypelle made their first contribution in https://github.com/rustformers/llm/pull/364
  • @radu-matei made their first contribution in https://github.com/rustformers/llm/pull/367
  • @basraayman made their first contribution in https://github.com/rustformers/llm/pull/379
  • @baskrahmer made their first contribution in https://github.com/rustformers/llm/pull/380
  • @clarkmcc made their first contribution in https://github.com/rustformers/llm/pull/394

Full Changelog: https://github.com/rustformers/llm/compare/v0.1.1...v0.2

hhamud avatar Aug 07 '23 14:08 hhamud

Though I think this ci add on looks interesting https://intuit.github.io/auto/docs

hhamud avatar Aug 07 '23 14:08 hhamud

Aye, that's a good start; we'll likely need to edit it to remove the redundancies and focus on the things that are relevant to users, but I really like the new contributors list!

philpax avatar Aug 08 '23 11:08 philpax