Basaran Versions Save

Basaran is an open-source alternative to the OpenAI text completion API. It provides a compatible streaming API for your Hugging Face Transformers-based text generation models.

v0.21.1

8 months ago

[b6a6e7e] build(*): bump version to 0.21.1 (#250)
[b4b36dd] build(docker): build image for linux/arm64 (#249)
[d0d8792] build(ci): maximize build space (#248)
[a516722] build(ci): disable workflow dispatch (#247)
[6288582] build(docker): minor code style updates (#246)
[8ddc11a] build(deps): update transformers[sentencepiece] requirement (#245)

v0.21.0

8 months ago

[f2e8c90] build(*): bump version to 0.21.0 (#244)
[795d375] build(deps): update safetensors requirement from ~=0.3.2 to ~=0.3.3 (#243)
[ec987bd] build(deps): update transformers[sentencepiece] requirement (#242)
[534c27e] build(deps): revert accelerate requirement (#239)
[5cb714e] build(deps): update accelerate requirement from ~=0.20.3 to ~=0.21.0 (#228)
[08799a7] build(deps): add scipy to dependencies (#238)
[91fef14] feat(model): add MPS support for Apple Silicon
[a602fcc] build(deps): update safetensors requirement from ~=0.3.1 to ~=0.3.2 (#236)
[d5f1bdd] build(deps): update bitsandbytes requirement from ~=0.41.0 to ~=0.41.1 (#235)
[f1b0c1e] build(deps): update bitsandbytes requirement from ~=0.40.2 to ~=0.41.0 (#233)

v0.20.0

9 months ago

[a15d1ad] build(*): bump version to 0.20.0 (#231)
[fae81ea] build(deps): update transformers[sentencepiece] requirement (#230)
[4525856] build(deps): update bitsandbytes requirement from ~=0.40.0 to ~=0.40.2 (#229)
[33080a0] build(deps): update bitsandbytes requirement from ~=0.39.1 to ~=0.40.0 (#226)
[691ffda] build(deps): update huggingface-hub requirement (#225)
[b62584d] build(deps): update huggingface-hub requirement (#224)
[8cfbe3f] build(deps): update huggingface-hub requirement (#222)
[9a0c400] build(deps): update flask-cors requirement from ~=3.0.10 to ~=4.0.0 (#219)
[de49385] fix(model): set model to evaluation mode (#218)
[0288db4] build(deps): update bitsandbytes requirement from ~=0.39.0 to ~=0.39.1 (#217)
[52bc773] build(deps): update transformers[sentencepiece] requirement (#214)
[1677491] feat(model): pass kwargs to generate from model call (#212)

v0.19.0

11 months ago

[5609282] build(*): bump version to 0.19.0 (#211)
[9e6d0c6] build(deps): update transformers[sentencepiece] requirement (#210)
[5e34a84] feat(model): add support for 4-bit quantization (#209)
[0c4b8a5] build(deps): update accelerate requirement from ~=0.19.0 to ~=0.20.3 (#208)
[86783fa] build(deps): update transformers[sentencepiece] requirement (#207)
[62c3b49] build(deps): update huggingface-hub requirement (#201)

v0.18.1

11 months ago

[5ef5ef0] build(*): bump version to 0.18.1 (#200)
[3b53e72] fix(server): temporary workaround for multiple prompts (#199)
[9f2621f] build(deps): update bitsandbytes requirement from ~=0.38.1 to ~=0.39.0 (#196)

v0.18.0

11 months ago

[6230c5b] build(*): bump version to 0.18.0 (#195)
[0f4f540] build(deps): update huggingface-hub requirement (#188)
[f4fc091] build(deps): update transformers[sentencepiece] requirement (#194)
[7ab88d0] build(deps): update transformers[sentencepiece] requirement (#191)
[13fe0e3] build(deps): update transformers[sentencepiece] requirement (#190)
[e0e0df6] build(deps): update accelerate requirement from ~=0.18.0 to ~=0.19.0 (#189)

v0.17.2

1 year ago

[e28c5cd] build(*): bump version to 0.17.2 (#187)
[997f19a] build(docker): download in safetensors format for supported models (#186)
[4ea43b4] build(deps): update safetensors requirement from ~=0.3.0 to ~=0.3.1 (#184)
[c64e78e] build(deps): update huggingface-hub requirement (#182)

v0.17.1

1 year ago

[10e250e] build(*): bump version to 0.17.1 (#178)
[1c53cf1] build(utils): only download weights index of the selected tensor format (#177)

v0.17.0

1 year ago

[4b1ce7e] build(*): bump version to 0.17.0 (#176)
[07fea19] build(deps): add safetensors to dependencies (#175)
[c99ef9f] build(utils): allow to download in safetensors format (#174)
[3a821bf] build(snap): permit access to gpu hardware (#171)
[cb1e316] build(docker): add example for bundling stablelm-tuned-alpha-7b (#170)
[183da79] build(docker): update example to use huggyllama/llama-7b (#169)
[b962b62] build(docker): clean up example bundles (#168)
[10b153f] build(utils): only download files for pytorch (#167)
[a626b95] test(data): remove unused weights (#166)
[aad86ad] build(docker): add chat template for chatglm (#165)
[b12e907] build(docker): add chat template for stablelm (#164)
[14b734c] feat(server): add default chat template (#163)

v0.16.2

1 year ago

[22a2c1f] build(*): bump version to 0.16.2 (#162)
[2c97344] build(snap): fix unbound variable (#161)
[6dbf68b] fix(server): significantly increase the default limits (#159)