root/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2025-01-12 11:40:17 +00:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

server : remove self-extend features (#9860)

Browse Source 
* server : remove self-extend

ggml-ci

* server : fix context limit check to use slot.n_past

ggml-ci
This commit is contained in:
Georgi Gerganov 2024-10-12 16:06:31 +03:00 committed by GitHub
parent 95c76e8e92
commit 1bde94dd02
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 57 additions and 142 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
common/arg.cpp
View File

@ -1163,14 +1163,14 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
[](common_params & params, int value) { [](common_params & params, int value) {
params.grp_attn_n = value; params.grp_attn_n = value;
} }
).set_env("LLAMA_ARG_GRP_ATTN_N")); ).set_env("LLAMA_ARG_GRP_ATTN_N").set_examples({LLAMA_EXAMPLE_MAIN, LLAMA_EXAMPLE_PASSKEY}));
add_opt(common_arg( add_opt(common_arg(
{"-gaw", "--grp-attn-w"}, "N", {"-gaw", "--grp-attn-w"}, "N",
string_format("group-attention width (default: %.1f)", (double)params.grp_attn_w), string_format("group-attention width (default: %d)", params.grp_attn_w),
[](common_params & params, int value) { [](common_params & params, int value) {
params.grp_attn_w = value; params.grp_attn_w = value;
} }
).set_env("LLAMA_ARG_GRP_ATTN_W")); ).set_env("LLAMA_ARG_GRP_ATTN_W").set_examples({LLAMA_EXAMPLE_MAIN}));
add_opt(common_arg( add_opt(common_arg(
{"-dkvc", "--dump-kv-cache"}, {"-dkvc", "--dump-kv-cache"},
"verbose print of the KV cache", "verbose print of the KV cache",

2
examples/server/README.md
View File

@ -60,8 +60,6 @@ The project is under active development, and we are [looking for feedback and co
| `--yarn-attn-factor N` | YaRN: scale sqrt(t) or attention magnitude (default: 1.0)<br/>(env: LLAMA_ARG_YARN_ATTN_FACTOR) | | `--yarn-attn-factor N` | YaRN: scale sqrt(t) or attention magnitude (default: 1.0)<br/>(env: LLAMA_ARG_YARN_ATTN_FACTOR) |
| `--yarn-beta-slow N` | YaRN: high correction dim or alpha (default: 1.0)<br/>(env: LLAMA_ARG_YARN_BETA_SLOW) | | `--yarn-beta-slow N` | YaRN: high correction dim or alpha (default: 1.0)<br/>(env: LLAMA_ARG_YARN_BETA_SLOW) |
| `--yarn-beta-fast N` | YaRN: low correction dim or beta (default: 32.0)<br/>(env: LLAMA_ARG_YARN_BETA_FAST) | | `--yarn-beta-fast N` | YaRN: low correction dim or beta (default: 32.0)<br/>(env: LLAMA_ARG_YARN_BETA_FAST) |
| `-gan, --grp-attn-n N` | group-attention factor (default: 1)<br/>(env: LLAMA_ARG_GRP_ATTN_N) |
| `-gaw, --grp-attn-w N` | group-attention width (default: 512.0)<br/>(env: LLAMA_ARG_GRP_ATTN_W) |
| `-dkvc, --dump-kv-cache` | verbose print of the KV cache | | `-dkvc, --dump-kv-cache` | verbose print of the KV cache |
| `-nkvo, --no-kv-offload` | disable KV offload<br/>(env: LLAMA_ARG_NO_KV_OFFLOAD) | | `-nkvo, --no-kv-offload` | disable KV offload<br/>(env: LLAMA_ARG_NO_KV_OFFLOAD) |
| `-ctk, --cache-type-k TYPE` | KV cache data type for K (default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_K) | | `-ctk, --cache-type-k TYPE` | KV cache data type for K (default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_K) |

121
examples/server/server.cpp
View File

@ -193,12 +193,6 @@ struct server_slot {
llama_token sampled; llama_token sampled;
int32_t ga_i = 0; // group-attention state
int32_t ga_n = 1; // group-attention factor
int32_t ga_w = 512; // group-attention width
int32_t n_past_se = 0; // self-extend
// stats // stats
size_t n_sent_text = 0; // number of sent text character size_t n_sent_text = 0; // number of sent text character
size_t n_sent_token_probs = 0; size_t n_sent_token_probs = 0;
@ -225,8 +219,6 @@ struct server_slot {
n_sent_text = 0; n_sent_text = 0;
n_sent_token_probs = 0; n_sent_token_probs = 0;
cmpl_type = SERVER_TASK_CMPL_TYPE_NORMAL; cmpl_type = SERVER_TASK_CMPL_TYPE_NORMAL;
ga_i = 0;
n_past_se = 0;
generated_token_probs.clear(); generated_token_probs.clear();
} }
@ -705,22 +697,6 @@ struct server_context {
SLT_INF(slot, "new slot n_ctx_slot = %d\n", slot.n_ctx); SLT_INF(slot, "new slot n_ctx_slot = %d\n", slot.n_ctx);
const int ga_n = params.grp_attn_n;
const int ga_w = params.grp_attn_w;
if (ga_n != 1) {
GGML_ASSERT(ga_n > 0 && "ga_n must be positive"); // NOLINT
GGML_ASSERT(ga_w % ga_n == 0 && "ga_w must be a multiple of ga_n"); // NOLINT
//GGML_ASSERT(n_ctx_train % ga_w == 0 && "n_ctx_train must be a multiple of ga_w"); // NOLINT
//GGML_ASSERT(n_ctx >= n_ctx_train * ga_n && "n_ctx must be at least n_ctx_train * ga_n"); // NOLINT
SLT_INF(slot, "slot self-extend: ga_n = %d, ga_w = %d\n", ga_n, ga_w);
}
slot.ga_i = 0;
slot.ga_n = ga_n;
slot.ga_w = ga_w;
slot.sparams = params.sparams; slot.sparams = params.sparams;
slot.callback_on_release = [this](int) { slot.callback_on_release = [this](int) {
@ -916,11 +892,6 @@ struct server_context {
slot.sparams.grammar = json_value(data, "grammar", default_sparams.grammar); slot.sparams.grammar = json_value(data, "grammar", default_sparams.grammar);
} }
if (slot.params.cache_prompt && slot.ga_n != 1) {
slot.params.cache_prompt = false;
SLT_WRN(slot, "%s", "group-attention is not supported with prompt caching. disabling cache\n");
}
if (slot.n_predict > 0 && slot.params.n_predict > slot.n_predict) { if (slot.n_predict > 0 && slot.params.n_predict > slot.n_predict) {
// Might be better to reject the request with a 400 ? // Might be better to reject the request with a 400 ?
slot.params.n_predict = slot.n_predict; slot.params.n_predict = slot.n_predict;
@ -1131,12 +1102,13 @@ struct server_context {
} }
// if context shift is disabled, we stop when it reaches the context limit // if context shift is disabled, we stop when it reaches the context limit
if (slot.n_decoded >= slot.n_ctx) { if (slot.n_past >= slot.n_ctx) {
slot.truncated = true; slot.truncated = true;
slot.stopped_limit = true; slot.stopped_limit = true;
slot.has_next_token = false; slot.has_next_token = false;
SLT_DBG(slot, "stopped due to running out of context capacity, n_decoded = %d, n_ctx = %d\n", slot.n_decoded, slot.n_ctx); SLT_DBG(slot, "stopped due to running out of context capacity, n_past = %d, n_prompt_tokens = %d, n_decoded = %d, n_ctx = %d\n",
slot.n_decoded, slot.n_prompt_tokens, slot.n_past, slot.n_ctx);
} }
if (llama_token_is_eog(model, result.tok)) { if (llama_token_is_eog(model, result.tok)) {
@ -1148,13 +1120,13 @@ struct server_context {
const auto n_ctx_train = llama_n_ctx_train(model); const auto n_ctx_train = llama_n_ctx_train(model);
if (slot.params.n_predict < 1 && slot.n_predict < 1 && slot.ga_n == 1 && slot.n_prompt_tokens + slot.n_decoded >= n_ctx_train) { if (slot.params.n_predict < 1 && slot.n_predict < 1 && slot.n_prompt_tokens + slot.n_decoded >= n_ctx_train) {
slot.truncated = true; slot.truncated = true;
slot.stopped_limit = true; slot.stopped_limit = true;
slot.has_next_token = false; // stop prediction slot.has_next_token = false; // stop prediction
SLT_WRN(slot, SLT_WRN(slot,
"n_predict (%d) is not set and self-context extend is disabled. " "n_predict (%d) is set for infinite generation. "
"Limiting generated tokens to n_ctx_train (%d) to avoid EOS-less generation infinite loop\n", "Limiting generated tokens to n_ctx_train (%d) to avoid EOS-less generation infinite loop\n",
slot.params.n_predict, n_ctx_train); slot.params.n_predict, n_ctx_train);
} }
@ -1826,8 +1798,7 @@ struct server_context {
// apply context-shift if needed // apply context-shift if needed
// TODO: simplify and improve // TODO: simplify and improve
for (server_slot & slot : slots) { for (server_slot & slot : slots) {
if (slot.ga_n == 1) { if (slot.is_processing() && slot.n_past + 1 >= slot.n_ctx) {
if (slot.is_processing() && slot.n_past >= slot.n_ctx - 1) {
if (!params.ctx_shift) { if (!params.ctx_shift) {
// this check is redundant (for good) // this check is redundant (for good)
// we should never get here, because generation should already stopped in process_token() // we should never get here, because generation should already stopped in process_token()
@ -1859,7 +1830,6 @@ struct server_context {
slot.truncated = true; slot.truncated = true;
} }
} }
}
// start populating the batch for this iteration // start populating the batch for this iteration
common_batch_clear(batch); common_batch_clear(batch);
@ -1872,9 +1842,7 @@ struct server_context {
slot.i_batch = batch.n_tokens; slot.i_batch = batch.n_tokens;
const int32_t slot_npast = slot.n_past_se > 0 ? slot.n_past_se : slot.n_past; common_batch_add(batch, slot.sampled, slot.n_past, { slot.id + 1 }, true);
common_batch_add(batch, slot.sampled, slot_npast, { slot.id + 1 }, true);
slot.n_past += 1; slot.n_past += 1;
@ -1993,6 +1961,8 @@ struct server_context {
} else { } else {
if (!params.ctx_shift) { if (!params.ctx_shift) {
// if context shift is disabled, we make sure prompt size is smaller than KV size // if context shift is disabled, we make sure prompt size is smaller than KV size
// TODO: there should be a separate parameter that control prompt truncation
// context shift should be applied only during the generation phase
if (slot.n_prompt_tokens >= slot.n_ctx) { if (slot.n_prompt_tokens >= slot.n_ctx) {
slot.release(); slot.release();
send_error(slot, "the request exceeds the available context size. try increasing the context size or enable context shift", ERROR_TYPE_INVALID_REQUEST); send_error(slot, "the request exceeds the available context size. try increasing the context size or enable context shift", ERROR_TYPE_INVALID_REQUEST);
@ -2005,7 +1975,7 @@ struct server_context {
slot.params.n_keep = std::min(slot.n_ctx - 4, slot.params.n_keep); slot.params.n_keep = std::min(slot.n_ctx - 4, slot.params.n_keep);
// if input prompt is too big, truncate it (if group attention self-extend is disabled) // if input prompt is too big, truncate it (if group attention self-extend is disabled)
if (slot.ga_n == 1 && slot.n_prompt_tokens >= slot.n_ctx) { if (slot.n_prompt_tokens >= slot.n_ctx) {
const int n_left = slot.n_ctx - slot.params.n_keep; const int n_left = slot.n_ctx - slot.params.n_keep;
const int n_block_size = n_left / 2; const int n_block_size = n_left / 2;
@ -2032,12 +2002,7 @@ struct server_context {
common_sampler_reset(slot.smpl); common_sampler_reset(slot.smpl);
if (!slot.params.cache_prompt) { if (slot.params.cache_prompt) {
slot.n_past_se = 0;
slot.ga_i = 0;
} else {
GGML_ASSERT(slot.ga_n == 1);
// reuse any previously computed tokens that are common with the new prompt // reuse any previously computed tokens that are common with the new prompt
slot.n_past = common_part(slot.cache_tokens, prompt_tokens); slot.n_past = common_part(slot.cache_tokens, prompt_tokens);
@ -2053,9 +2018,6 @@ struct server_context {
SLT_WRN(slot, "need to evaluate at least 1 token to generate logits, n_past = %d, n_prompt_tokens = %d\n", slot.n_past, slot.n_prompt_tokens); SLT_WRN(slot, "need to evaluate at least 1 token to generate logits, n_past = %d, n_prompt_tokens = %d\n", slot.n_past, slot.n_prompt_tokens);
slot.n_past--; slot.n_past--;
if (slot.ga_i > 0) {
slot.n_past_se--;
}
} }
slot.n_prompt_tokens_processed = 0; slot.n_prompt_tokens_processed = 0;
@ -2081,52 +2043,31 @@ struct server_context {
} }
// keep only the common part // keep only the common part
int p0 = slot.n_past; if (!llama_kv_cache_seq_rm(ctx, slot.id + 1, slot.n_past, -1)) {
if (!llama_kv_cache_seq_rm(ctx, slot.id + 1, p0, -1)) {
// could not partially delete (likely using a non-Transformer model) // could not partially delete (likely using a non-Transformer model)
llama_kv_cache_seq_rm(ctx, slot.id + 1, -1, -1); llama_kv_cache_seq_rm(ctx, slot.id + 1, -1, -1);
p0 = 0;
// there is no common part left // there is no common part left
slot.n_past = 0; slot.n_past = 0;
slot.n_past_se = 0;
slot.ga_i = 0;
common_sampler_reset(slot.smpl); common_sampler_reset(slot.smpl);
} }
SLT_INF(slot, "kv cache rm [%d, end)\n", slot.n_past);
// remove the non-common part from the cache // remove the non-common part from the cache
slot.cache_tokens.resize(slot.n_past); slot.cache_tokens.resize(slot.n_past);
SLT_INF(slot, "kv cache rm [%d, end)\n", p0);
int32_t slot_npast = slot.n_past_se > 0 ? slot.n_past_se : slot.n_past;
int32_t ga_i = slot.ga_i;
int32_t ga_n = slot.ga_n;
int32_t ga_w = slot.ga_w;
// add prompt tokens for processing in the current batch // add prompt tokens for processing in the current batch
// TODO: the self-extend stuff here is a mess - simplify and/or abstract it somehow while (slot.n_past < slot.n_prompt_tokens && batch.n_tokens < n_batch) {
for (; slot.n_past < slot.n_prompt_tokens && batch.n_tokens < n_batch; ++slot.n_past) { common_batch_add(batch, prompt_tokens[slot.n_past], slot.n_past, { slot.id + 1 }, false);
if (slot.ga_n != 1) {
while (slot_npast >= ga_i + ga_w) {
const int bd = (ga_w/ga_n)*(ga_n - 1);
slot_npast -= bd;
ga_i += ga_w/ga_n;
}
}
common_batch_add(batch, prompt_tokens[slot.n_past], slot_npast, { slot.id + 1 }, false);
if (slot.params.cache_prompt) { if (slot.params.cache_prompt) {
slot.cache_tokens.push_back(prompt_tokens[slot.n_past]); slot.cache_tokens.push_back(prompt_tokens[slot.n_past]);
} }
slot.n_prompt_tokens_processed++; slot.n_prompt_tokens_processed++;
slot_npast++; slot.n_past++;
} }
SLT_INF(slot, "prompt processing progress, n_past = %d, n_tokens = %d, progress = %f\n", slot.n_past, batch.n_tokens, (float) slot.n_prompt_tokens_processed / slot.n_prompt_tokens); SLT_INF(slot, "prompt processing progress, n_past = %d, n_tokens = %d, progress = %f\n", slot.n_past, batch.n_tokens, (float) slot.n_prompt_tokens_processed / slot.n_prompt_tokens);
@ -2167,34 +2108,6 @@ struct server_context {
for (int32_t i = 0; i < batch.n_tokens; i += n_batch) { for (int32_t i = 0; i < batch.n_tokens; i += n_batch) {
const int32_t n_tokens = std::min(n_batch, batch.n_tokens - i); const int32_t n_tokens = std::min(n_batch, batch.n_tokens - i);
for (auto & slot : slots) {
if (slot.ga_n != 1) {
// context extension via Self-Extend
// TODO: simplify and/or abstract this
while (slot.n_past_se >= slot.ga_i + slot.ga_w) {
const int ib = (slot.ga_n * slot.ga_i) / slot.ga_w;
const int bd = (slot.ga_w / slot.ga_n) * (slot.ga_n - 1);
const int dd = (slot.ga_w / slot.ga_n) - ib * bd - slot.ga_w;
SLT_DBG(slot, "shift: [%6d, %6d] + %6d -> [%6d, %6d]\n", slot.ga_i, slot.n_past_se, ib * bd, slot.ga_i + ib * bd, slot.n_past_se + ib * bd);
SLT_DBG(slot, "div: [%6d, %6d] / %6d -> [%6d, %6d]\n", slot.ga_i + ib * bd, slot.ga_i + ib * bd + slot.ga_w, slot.ga_n, (slot.ga_i + ib * bd) / slot.ga_n, (slot.ga_i + ib * bd + slot.ga_w) / slot.ga_n);
SLT_DBG(slot, "shift: [%6d, %6d] + %6d -> [%6d, %6d]\n", slot.ga_i + ib * bd + slot.ga_w, slot.n_past_se + ib * bd, dd, slot.ga_i + ib * bd + slot.ga_w + dd, slot.n_past_se + ib * bd + dd);
llama_kv_cache_seq_add(ctx, slot.id + 1, slot.ga_i, slot.n_past_se, ib * bd);
llama_kv_cache_seq_div(ctx, slot.id + 1, slot.ga_i + ib * bd, slot.ga_i + ib * bd + slot.ga_w, slot.ga_n);
llama_kv_cache_seq_add(ctx, slot.id + 1, slot.ga_i + ib * bd + slot.ga_w, slot.n_past_se + ib * bd, dd);
slot.n_past_se -= bd;
slot.ga_i += slot.ga_w / slot.ga_n;
SLT_DBG(slot, "\nn_past_old = %d, n_past = %d, ga_i = %d\n\n", slot.n_past_se + bd, slot.n_past_se, slot.ga_i);
}
slot.n_past_se += n_tokens;
}
}
llama_batch batch_view = { llama_batch batch_view = {
n_tokens, n_tokens,
batch.token + i, batch.token + i,

4
examples/server/tests/features/ctx_shift.feature
View File

@ -13,6 +13,10 @@ Feature: llama.cpp server
And 32 as batch size And 32 as batch size
And 2 slots And 2 slots
# the prompt is 301 tokens
# the slot context is 256/2 = 128 tokens
# the prompt is truncated to keep the last 109 tokens
# 64 tokens are generated thanks to shifting the context when it gets full
Scenario: Inference with context shift Scenario: Inference with context shift
And 64 server max tokens to predict And 64 server max tokens to predict
Then the server is starting Then the server is starting