huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-04-23 16:02:10 +00:00
Code Issues Packages Projects Releases Wiki Activity

chore(trtllm): post-rebase commit

Browse Source
This commit is contained in:
Morgan Funtowicz 2024-10-21 12:31:24 +02:00
parent 85c03e33a9
commit fb00f985ae
2 changed files with 19 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

68
Cargo.lock generated
View File

@ -2706,9 +2706,9 @@ dependencies = [
[[package]]
name = "opentelemetry"
version = "0.23.0"
version = "0.24.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1b69a91d4893e713e06f724597ad630f1fa76057a5e1026c0ca67054a9032a76"
checksum = "4c365a63eec4f55b7efeceb724f1336f26a9cf3427b70e59e2cd2a5b947fba96"
dependencies = [
"futures-core",
"futures-sink",
@ -2819,19 +2819,17 @@ dependencies = [
[[package]]
name = "opentelemetry_sdk"
version = "0.23.0"
version = "0.24.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "ae312d58eaa90a82d2e627fd86e075cf5230b3f11794e2ed74199ebbe572d4fd"
checksum = "692eac490ec80f24a17828d49b40b60f5aeaccdfe6a503f939713afd22bc28df"
dependencies = [
"async-trait",
"futures-channel",
"futures-executor",
"futures-util",
"glob",
"lazy_static",
"once_cell",
"opentelemetry 0.23.0",
"ordered-float 4.3.0",
"opentelemetry 0.24.0",
"percent-encoding",
"rand",
"thiserror",
@ -4185,16 +4183,17 @@ dependencies = [
"cmake",
"cxx",
"cxx-build",
"hashbrown 0.14.5",
"hf-hub",
"log",
"parking_lot",
"pkg-config",
"text-generation-router",
"thiserror",
"tokenizers 0.19.1",
"tokenizers",
"tokio",
"tokio-stream",
"tracing",
"tracing-opentelemetry 0.24.0",
"tracing-opentelemetry 0.25.0",
"tracing-subscriber",
]
@ -4212,7 +4211,7 @@ dependencies = [
"tabled",
"text-generation-client",
"thiserror",
"tokenizers 0.20.0",
"tokenizers",
"tokio",
"tracing",
"tracing-subscriber",
@ -4292,7 +4291,7 @@ dependencies = [
"serde_json",
"sysinfo",
"thiserror",
"tokenizers 0.20.0",
"tokenizers",
"tokio",
"tokio-stream",
"tower-http",
@ -4341,7 +4340,7 @@ dependencies = [
"slotmap",
"text-generation-router",
"thiserror",
"tokenizers 0.20.0",
"tokenizers",
"tokio",
"tokio-stream",
"tonic 0.10.2",
@ -4392,7 +4391,7 @@ dependencies = [
"slotmap",
"text-generation-router",
"thiserror",
"tokenizers 0.20.0",
"tokenizers",
"tokio",
"tokio-stream",
"tonic 0.10.2",
@ -4514,39 +4513,6 @@ version = "0.1.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "1f3ccbac311fea05f86f61904b462b55fb3df8837a366dfc601a0161d0532f20"
[[package]]
name = "tokenizers"
version = "0.19.1"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "e500fad1dd3af3d626327e6a3fe5050e664a6eaa4708b8ca92f1794aaf73e6fd"
dependencies = [
"aho-corasick",
"derive_builder",
"esaxx-rs",
"getrandom",
"hf-hub",
"indicatif",
"itertools 0.12.1",
"lazy_static",
"log",
"macro_rules_attribute",
"monostate",
"onig",
"paste",
"rand",
"rayon",
"rayon-cond",
"regex",
"regex-syntax 0.8.5",
"serde",
"serde_json",
"spm_precompiled",
"thiserror",
"unicode-normalization-alignments",
"unicode-segmentation",
"unicode_categories",
]
[[package]]
name = "tokenizers"
version = "0.20.0"
@ -4933,14 +4899,14 @@ dependencies = [
[[package]]
name = "tracing-opentelemetry"
version = "0.24.0"
version = "0.25.0"
source = "registry+https://github.com/rust-lang/crates.io-index"
checksum = "f68803492bf28ab40aeccaecc7021096bd256baf7ca77c3d425d89b35a7be4e4"
checksum = "a9784ed4da7d921bc8df6963f8c80a0e4ce34ba6ba76668acadd3edbd985ff3b"
dependencies = [
"js-sys",
"once_cell",
"opentelemetry 0.23.0",
"opentelemetry_sdk 0.23.0",
"opentelemetry 0.24.0",
"opentelemetry_sdk 0.24.1",
"smallvec",
"tracing",
"tracing-core",

4
backends/trtllm/src/main.rs
View File

@ -9,6 +9,7 @@ use tracing::info;
use text_generation_backends_trtllm::errors::TensorRtLlmBackendError;
use text_generation_backends_trtllm::TensorRtLlmBackendV2;
use text_generation_router::server::{create_post_processor, get_base_tokenizer};
use text_generation_router::usage_stats::UsageStatsLevel;
use text_generation_router::{server, HubTokenizerConfig};
/// App Configuration
@ -312,8 +313,7 @@ async fn main() -> Result<(), TensorRtLlmBackendError> {
messages_api_enabled,
true,
max_client_batch_size,
false,
false,
UsageStatsLevel::Off,
)
.await?;
Ok(())