Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[pull] master from ggerganov:master #31

Closed
wants to merge 7 commits into from
Closed

[pull] master from ggerganov:master #31

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
4ffc7a1
server : various fixes for the prompt field in /completion (#5300)
Niall- Feb 6, 2024
31e7903
server : add `dynatemp_range` and `dynatemp_exponent` (#5352)
m18coppola Feb 6, 2024
8a79c59
server : include total "num_slots" in props endpoint (#5349)
jparkerweb Feb 6, 2024
2c51661
CUDA: mul_mat_vec_q for batch sizes > 1 (#5351)
JohannesGaessler Feb 6, 2024
2e9c0bd
readme : add phi, orion 14b, internlm2, and yi-VL to readme (#5362)
BarfingLemurs Feb 6, 2024
f57fadc
Slight quantization improvement for Q4_K and Q5_K (#5361)
ikawrakow Feb 6, 2024
b08f22c
Update README.md (#5366)
ikawrakow Feb 6, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 18 additions & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -105,11 +105,14 @@ improved significantly thanks to many contributions. It is the main playground f
- [X] [MPT](https://github.com/ggerganov/llama.cpp/pull/3417)
- [X] [Bloom](https://github.com/ggerganov/llama.cpp/pull/3553)
- [x] [Yi models](https://huggingface.co/models?search=01-ai/Yi)
- [X] [StableLM-3b-4e1t](https://github.com/ggerganov/llama.cpp/pull/3586)
- [X] [StableLM models](https://huggingface.co/stabilityai)
- [x] [Deepseek models](https://huggingface.co/models?search=deepseek-ai/deepseek)
- [x] [Qwen models](https://huggingface.co/models?search=Qwen/Qwen)
- [x] [PLaMo-13B](https://github.com/ggerganov/llama.cpp/pull/3557)
- [x] [Phi models](https://huggingface.co/models?search=microsoft/phi)
- [x] [GPT-2](https://huggingface.co/gpt2)
- [x] [Orion 14B](https://github.com/ggerganov/llama.cpp/pull/5118)
- [x] [InternLM2](https://huggingface.co/models?search=internlm2)
- [x] [CodeShell](https://github.com/WisdomShell/codeshell)

**Multimodal models:**
Expand All @@ -119,6 +122,7 @@ improved significantly thanks to many contributions. It is the main playground f
- [x] [Obsidian](https://huggingface.co/NousResearch/Obsidian-3B-V0.5)
- [x] [ShareGPT4V](https://huggingface.co/models?search=Lin-Chen/ShareGPT4V)
- [x] [MobileVLM 1.7B/3B models](https://huggingface.co/models?search=mobileVLM)
- [x] [Yi-VL](https://huggingface.co/models?search=Yi-VL)


**Bindings:**
Expand Down Expand Up @@ -732,9 +736,21 @@ Several quantization methods are supported. They differ in the resulting model d
| 13B | bits/weight | 16.0 | 4.5 | 5.0 | 5.5 | 6.0 | 8.5 |

- [k-quants](https://github.com/ggerganov/llama.cpp/pull/1684)
- recent k-quants improvements
- recent k-quants improvements and new i-quants
- [#2707](https://github.com/ggerganov/llama.cpp/pull/2707)
- [#2807](https://github.com/ggerganov/llama.cpp/pull/2807)
- [#4773 - 2-bit i-quants (inference)](https://github.com/ggerganov/llama.cpp/pull/4773)
- [#4856 - 2-bit i-quants (inference)](https://github.com/ggerganov/llama.cpp/pull/4856)
- [#4861 - importance matrix](https://github.com/ggerganov/llama.cpp/pull/4861)
- [#4872 - MoE models](https://github.com/ggerganov/llama.cpp/pull/4872)
- [#4897 - 2-bit quantization](https://github.com/ggerganov/llama.cpp/pull/4897)
- [#4930 - imatrix for all k-quants](https://github.com/ggerganov/llama.cpp/pull/4930)
- [#4951 - imatrix on the GPU](https://github.com/ggerganov/llama.cpp/pull/4957)
- [#4969 - imatrix for legacy quants](https://github.com/ggerganov/llama.cpp/pull/4969)
- [#4996 - k-qunats tuning](https://github.com/ggerganov/llama.cpp/pull/4996)
- [#5060 - Q3_K_XS](https://github.com/ggerganov/llama.cpp/pull/5060)
- [#5196 - 3-bit i-quants](https://github.com/ggerganov/llama.cpp/pull/5196)
- [quantization tuning](https://github.com/ggerganov/llama.cpp/pull/5320), [another one](https://github.com/ggerganov/llama.cpp/pull/5334), and [another one](https://github.com/ggerganov/llama.cpp/pull/5361)

### Perplexity (measuring model quality)

Expand Down
4 changes: 4 additions & 0 deletions examples/server/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -137,6 +137,10 @@ node index.js

`temperature`: Adjust the randomness of the generated text (default: 0.8).

`dynatemp_range`: Dynamic temperature range (default: 0.0, 0.0 = disabled).

`dynatemp_exponent`: Dynamic temperature exponent (default: 1.0).

`top_k`: Limit the next token selection to the K most probable tokens (default: 40).

`top_p`: Limit the next token selection to a subset of tokens with a cumulative probability above a threshold P (default: 0.95).
Expand Down
81 changes: 53 additions & 28 deletions examples/server/server.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -432,6 +432,7 @@ struct llama_server_context
}

default_generation_settings_for_props = get_formated_generation(slots.front());
default_generation_settings_for_props["num_slots"] = params.n_parallel;
default_generation_settings_for_props["seed"] = -1;

batch = llama_batch_init(n_ctx, 0, params.n_parallel);
Expand Down Expand Up @@ -524,27 +525,29 @@ struct llama_server_context
slot->oaicompat_model = "";
}

slot->params.stream = json_value(data, "stream", false);
slot->params.cache_prompt = json_value(data, "cache_prompt", false);
slot->params.n_predict = json_value(data, "n_predict", default_params.n_predict);
slot->sparams.top_k = json_value(data, "top_k", default_sparams.top_k);
slot->sparams.top_p = json_value(data, "top_p", default_sparams.top_p);
slot->sparams.min_p = json_value(data, "min_p", default_sparams.min_p);
slot->sparams.tfs_z = json_value(data, "tfs_z", default_sparams.tfs_z);
slot->sparams.typical_p = json_value(data, "typical_p", default_sparams.typical_p);
slot->sparams.temp = json_value(data, "temperature", default_sparams.temp);
slot->sparams.penalty_last_n = json_value(data, "repeat_last_n", default_sparams.penalty_last_n);
slot->sparams.penalty_repeat = json_value(data, "repeat_penalty", default_sparams.penalty_repeat);
slot->sparams.penalty_freq = json_value(data, "frequency_penalty", default_sparams.penalty_freq);
slot->sparams.penalty_present = json_value(data, "presence_penalty", default_sparams.penalty_present);
slot->sparams.mirostat = json_value(data, "mirostat", default_sparams.mirostat);
slot->sparams.mirostat_tau = json_value(data, "mirostat_tau", default_sparams.mirostat_tau);
slot->sparams.mirostat_eta = json_value(data, "mirostat_eta", default_sparams.mirostat_eta);
slot->sparams.penalize_nl = json_value(data, "penalize_nl", default_sparams.penalize_nl);
slot->params.n_keep = json_value(data, "n_keep", slot->params.n_keep);
slot->params.seed = json_value(data, "seed", default_params.seed);
slot->sparams.grammar = json_value(data, "grammar", default_sparams.grammar);
slot->sparams.n_probs = json_value(data, "n_probs", default_sparams.n_probs);
slot->params.stream = json_value(data, "stream", false);
slot->params.cache_prompt = json_value(data, "cache_prompt", false);
slot->params.n_predict = json_value(data, "n_predict", default_params.n_predict);
slot->sparams.top_k = json_value(data, "top_k", default_sparams.top_k);
slot->sparams.top_p = json_value(data, "top_p", default_sparams.top_p);
slot->sparams.min_p = json_value(data, "min_p", default_sparams.min_p);
slot->sparams.tfs_z = json_value(data, "tfs_z", default_sparams.tfs_z);
slot->sparams.typical_p = json_value(data, "typical_p", default_sparams.typical_p);
slot->sparams.temp = json_value(data, "temperature", default_sparams.temp);
slot->sparams.dynatemp_range = json_value(data, "dynatemp_range", default_sparams.dynatemp_range);
slot->sparams.dynatemp_exponent = json_value(data, "dynatemp_exponent", default_sparams.dynatemp_exponent);
slot->sparams.penalty_last_n = json_value(data, "repeat_last_n", default_sparams.penalty_last_n);
slot->sparams.penalty_repeat = json_value(data, "repeat_penalty", default_sparams.penalty_repeat);
slot->sparams.penalty_freq = json_value(data, "frequency_penalty", default_sparams.penalty_freq);
slot->sparams.penalty_present = json_value(data, "presence_penalty", default_sparams.penalty_present);
slot->sparams.mirostat = json_value(data, "mirostat", default_sparams.mirostat);
slot->sparams.mirostat_tau = json_value(data, "mirostat_tau", default_sparams.mirostat_tau);
slot->sparams.mirostat_eta = json_value(data, "mirostat_eta", default_sparams.mirostat_eta);
slot->sparams.penalize_nl = json_value(data, "penalize_nl", default_sparams.penalize_nl);
slot->params.n_keep = json_value(data, "n_keep", slot->params.n_keep);
slot->params.seed = json_value(data, "seed", default_params.seed);
slot->sparams.grammar = json_value(data, "grammar", default_sparams.grammar);
slot->sparams.n_probs = json_value(data, "n_probs", default_sparams.n_probs);

// infill
if (data.count("input_prefix") != 0)
Expand Down Expand Up @@ -1002,6 +1005,8 @@ struct llama_server_context
{"model", params.model_alias},
{"seed", slot.params.seed},
{"temperature", slot.sparams.temp},
{"dynatemp_range", slot.sparams.dynatemp_range},
{"dynatemp_exponent", slot.sparams.dynatemp_exponent},
{"top_k", slot.sparams.top_k},
{"top_p", slot.sparams.top_p},
{"min_p", slot.sparams.min_p},
Expand Down Expand Up @@ -1163,13 +1168,30 @@ struct llama_server_context
task.multitask_id = multitask_id;

// when a completion task's prompt array is not a singleton, we split it into multiple requests
if (task.data.count("prompt") && task.data.at("prompt").size() > 1)
{
split_multiprompt_task(task_id, task);
}

// otherwise, it's a single-prompt task, we actually queue it
queue_tasks.post(task);
// if there's numbers in the prompt array it will be treated as an array of tokens
if (task.data.count("prompt") != 0 && task.data.at("prompt").size() > 1) {
bool numbers = false;
for (const auto& e : task.data.at("prompt")) {
if (e.is_number()) {
numbers = true;
break;
}
}

// NOTE: split_multiprompt_task() does not handle a mix of strings and numbers,
// it will completely stall the server. I don't know where the bug for this is.
//
// if there are numbers, it needs to be treated like a single prompt,
// queue_tasks handles a mix of strings and numbers just fine.
if (numbers) {
queue_tasks.post(task);
} else {
split_multiprompt_task(task_id, task);
}
} else {
queue_tasks.post(task);
}
}

// for multiple images processing
Expand Down Expand Up @@ -1251,7 +1273,10 @@ struct llama_server_context
void split_multiprompt_task(int multitask_id, task_server& multiprompt_task)
{
int prompt_count = multiprompt_task.data.at("prompt").size();
assert(prompt_count > 1);
if (prompt_count <= 1) {
send_error(multiprompt_task, "error while handling multiple prompts");
return;
}

// generate all the ID for subtask
std::vector<int> subtask_ids(prompt_count);
Expand Down
Loading
Loading