Merge pull request #236 from sobelio/workspace-deps

Workspace deps

Cargo.toml

@@ -1,9 +1,19 @@
 [workspace]
 members = ["crates/*"]
+resolver = "2"
 
 [workspace.metadata.release]
 shared-version = true
 
 [patch.crates-io]
 hnsw_rs = { git = "https://github.com/ruqqq/hnswlib-rs" }
 
+[workspace.dependencies]
+serde = { version = "1.0.163", features = ["derive"] }
+serde_json = "1.0.96"
+thiserror = "1.0.40"
+tokio = "1.28.0"
+async-trait = "0.1.68"
+anyhow = "1.0.71"
+lazy_static = "1.4.0"
+serde_yaml = "0.9.21"

crates/llm-chain-hnsw/Cargo.toml

@@ -13,13 +13,13 @@ repository = "https://github.com/sobelio/llm-chain/"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-async-trait = "0.1.68"
+async-trait.workspace = true
 hnsw_rs = "0.1.19"
-llm-chain = { path = "../llm-chain", version = "0.12.3", default-features = false }
-serde = "1.0.164"
-serde_json = "1.0.99"
-thiserror = "1.0.40"
-tokio = "1.28.2"
+llm-chain = { path = "../llm-chain", version = "0.12.0", default-features = false }
+serde.workspace = true
+serde_json.workspace = true
+thiserror.workspace = true
+tokio.workspace = true
 
 [dev-dependencies]
 llm-chain-openai = { path = "../llm-chain-openai" }

crates/llm-chain-llama/Cargo.toml

@@ -14,14 +14,14 @@ readme = "./README.md"
 repository = "https://github.com/sobelio/llm-chain/"
 
 [dependencies]
-anyhow = "1.0.72"
-async-trait = "0.1.68"
+anyhow.workspace = true
+async-trait.workspace = true
 llm-chain-llama-sys = { path = "../llm-chain-llama-sys", version = "0.12" }
-llm-chain = { path = "../llm-chain", version = "0.12.3" }
-serde = { version = "1.0.164", features = ["derive"] }
-thiserror = "1.0.40"
+llm-chain = { path = "../llm-chain", version = "0.12.0" }
+serde = { version = "1.0.163", features = ["derive"] }
+thiserror.workspace = true
 lazy_static = "1.4.0"
-tokio = "1.28.2"
+tokio.workspace = true
 
 [dev-dependencies]
 tokio = { version = "1.28.2", features = ["macros", "rt"] }

crates/llm-chain-llama/examples/simple.rs → crates/llm-chain-llama/examples/simple_llama.rs

@@ -1,7 +1,7 @@
 use llm_chain::options;
-use llm_chain::options::{ModelRef, Options};
+use llm_chain::options::ModelRef;
 use llm_chain::{executor, parameters, prompt};
-use std::{env::args, error::Error};
+use std::env::args;
 /// This example demonstrates how to use the llm-chain-llama crate to generate text using a
 /// LLaMA model.
 ///

crates/llm-chain-llama/src/context.rs

@@ -22,23 +22,14 @@ pub struct LLAMACPPErrorCode(i32);
 // Represents the configuration parameters for a LLamaContext.
 #[derive(Debug, Clone)]
 pub struct ContextParams {
+    pub n_parts: i32,
     pub n_ctx: i32,
-    pub n_batch: i32,
-    pub n_gpu_layers: i32,
-    pub main_gpu: i32,
-    pub tensor_split: *const f32,
-    pub seed: u32,
+    pub seed: i32,
     pub f16_kv: bool,
     pub vocab_only: bool,
     pub use_mlock: bool,
     pub use_mmap: bool,
     pub embedding: bool,
-    pub low_vram: bool,
-    pub rope_freq_base: f32,
-    pub rope_freq_scale: f32,
-    pub mul_mat_q: bool,
-    pub n_gqa: i32,
-    pub rms_norm_eps: f32,
 }
 
 unsafe impl Sync for ContextParams {}
@@ -66,11 +57,8 @@ impl Default for ContextParams {
 impl From<ContextParams> for llama_context_params {
     fn from(params: ContextParams) -> Self {
         llama_context_params {
+            n_parts: params.n_parts,
             n_ctx: params.n_ctx,
-            n_batch: params.n_batch,
-            n_gpu_layers: params.n_gpu_layers,
-            main_gpu: params.main_gpu,
-            tensor_split: params.tensor_split,
             seed: params.seed,
             f16_kv: params.f16_kv,
             logits_all: false,
@@ -80,12 +68,6 @@ impl From<ContextParams> for llama_context_params {
             embedding: params.embedding,
             progress_callback: None,
             progress_callback_user_data: null_mut(),
-            low_vram: params.low_vram,
-            rope_freq_base: params.rope_freq_base,
-            rope_freq_scale: params.rope_freq_scale,
-            mul_mat_q: params.mul_mat_q,
-            n_gqa: params.n_gqa,
-            rms_norm_eps: params.rms_norm_eps,
         }
     }
 }
@@ -94,22 +76,13 @@ impl From<llama_context_params> for ContextParams {
     fn from(params: llama_context_params) -> Self {
         ContextParams {
             n_ctx: params.n_ctx,
-            n_batch: params.n_batch,
-            n_gpu_layers: params.n_gpu_layers,
-            main_gpu: params.main_gpu,
-            tensor_split: params.tensor_split,
+            n_parts: params.n_parts,
             seed: params.seed,
             f16_kv: params.f16_kv,
             vocab_only: params.vocab_only,
             use_mlock: params.use_mlock,
             use_mmap: params.use_mmap,
             embedding: params.embedding,
-            low_vram: params.low_vram,
-            rope_freq_base: params.rope_freq_base,
-            rope_freq_scale: params.rope_freq_scale,
-            mul_mat_q: params.mul_mat_q,
-            n_gqa: params.n_gqa,
-            rms_norm_eps: params.rms_norm_eps,
         }
     }
 }

crates/llm-chain-llama/src/options.rs

@@ -111,10 +111,7 @@ lazy_static! {
         MirostatTau: 5.0,
         MirostatEta: 0.1,
         PenalizeNl: true,
-        StopSequence: vec!["\n\n".to_string()],
-        NumGpuLayers: 0,
-        RopeFrequencyBase: 10000.0,
-        RopeFrequencyScale: 1.0
+        StopSequence: vec!["\n\n".to_string()]
     );
 }
 
@@ -123,15 +120,9 @@ pub(crate) fn get_executor_initial_opts(
 ) -> Result<(String, ContextParams), ExecutorCreationError> {
     let model = opt_extract!(opt, model, Model)?;
     let max_context_size = opt_extract!(opt, max_context_size, MaxContextSize)?;
-    let num_gpu_layers = opt_extract!(opt, num_gpu_layers, NumGpuLayers)?;
-    let rope_freq_base = opt_extract!(opt, rope_freq_base, RopeFrequencyBase)?;
-    let rope_freq_scale = opt_extract!(opt, rope_freq_scale, RopeFrequencyScale)?;
 
     let mut cp = ContextParams::new();
     cp.n_ctx = *max_context_size as i32;
-    cp.n_gpu_layers = *num_gpu_layers;
-    cp.rope_freq_base = *rope_freq_base;
-    cp.rope_freq_scale = *rope_freq_scale;
 
     Ok((model.to_path(), cp))
 }

crates/llm-chain-local/Cargo.toml

@@ -12,13 +12,13 @@ repository = "https://github.com/sobelio/llm-chain/"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-async-trait = "0.1.68"
-lazy_static = "1.4.0"
+async-trait.workspace = true
+lazy_static.workspace = true
 llm = "0.1.1"
 llm-chain = { path = "../llm-chain", version = "0.12.3", default-features = false }
 rand = "0.8.5"
-serde = { version = "1.0.164", features = ["derive"] }
-thiserror = "1.0.40"
+serde.workspace = true
+thiserror.workspace = true
 
 [dev-dependencies]
 tokio = { version = "1.28.2", features = ["macros", "rt"] }

crates/llm-chain-milvus/src/lib.rs

@@ -9,7 +9,7 @@ use milvus::{
     collection::SearchOption,
     data::FieldColumn,
     proto::{milvus::MutationResult, schema::i_ds::IdField},
-    value::{Value, ValueVec},
+    value::ValueVec,
 };
 use serde::{de::DeserializeOwned, Serialize};
 use std::{collections::HashMap, marker::PhantomData, sync::Arc};

crates/llm-chain-openai/Cargo.toml

@@ -14,20 +14,19 @@ repository = "https://github.com/sobelio/llm-chain/"
 
 [dependencies]
 futures = "0.3.28"
-async-openai = "0.10.3"
-async-trait = "0.1.68"
-llm-chain = { path = "../llm-chain", version = "0.12.3", default-features = false }
-serde = { version = "1.0.164" }
-strum = "0.25"
-strum_macros = "0.25"
-thiserror = "1.0.40"
-tiktoken-rs = { version = "0.4.2", features = ["async-openai"] }
-tokio = "1.28.2"
+async-openai = "0.16.2"
+async-trait.workspace = true
+llm-chain = { path = "../llm-chain", version = "0.12.0", default-features = false }
+serde.workspace = true
+strum = "0.24"
+strum_macros = "0.24"
+thiserror.workspace = true
+tiktoken-rs = { version = "0.5.7" }
+tokio.workspace = true
 
 [dev-dependencies]
 tokio = "1.28.2"
 qdrant-client = "1.3.0"
 llm-chain = { path = "../llm-chain" }
-anyhow = "1.0.72"
-serde_yaml = "0.9.27"
-
+anyhow = "1.0.70"
+serde_yaml = "0.9.21"

crates/llm-chain-openai/src/chatgpt/error.rs

@@ -0,0 +1,12 @@
+use async_openai::error::OpenAIError;
+use llm_chain::prompt::StringTemplateError;
+use thiserror::Error;
+
+#[derive(Debug, Error)]
+#[error(transparent)]
+pub enum OpenAIInnerError {
+    #[error(transparent)]
+    OpenAIError(#[from] OpenAIError),
+    #[error(transparent)]
+    StringTemplateError(#[from] StringTemplateError),
+}